Ex Parte Schaller et al

Patent Trial and Appeal Board

Feb 25, 2015

11530338 (P.T.A.B. Feb. 25, 2015)

UNITED STATES PATENT AND TRADEMARK OFFICE ____________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________ Ex parte RAIMUND SCHALLER, ARMIN HOLZNER, RICHARD EHRENFELDNER, MICHAEL HOECHTL, WOLFGANG KERN, FRANZ STELZER, and ARMIN TEMEL ____________ Appeal 2013-002146 Application 11/530,338 Technology Center 1700 ____________ Before CATHERINE Q. TIMM, GEORGE C. BEST, and MICHELLE N. ANKENBRAND, Administrative Patent Judges. BEST, Administrative Patent Judge. DECISION ON APPEAL The Examiner has finally rejected claims 1-10, 48-51, 64-69, 72-80, and 82 of Application 11/530,338 under 35 U.S.C. § 103(a) as obvious. Final Rejection (FR) (Nov. 25, 2011). Appellants1 seek reversal of these rejections pursuant to 35 U.S.C. § 134(a). We have jurisdiction under 35 U.S.C. § 6(b). 1 Semperit Aktiengesellschaft Holding is identified as the real party in interest. App. Br. 6. Appeal 2013-002146 Application 11/530,338 2 For the reasons set forth below, we REVERSE. BACKGROUND Reactions between thiols and unsaturated organic compounds were observed more than a century ago.2 Photocatalysis of such reactions was observed by 1928,3 and a free radical chain-reaction mechanism was proposed in 1940.4 At least as early as 1948, persulfate-catalyzed reactions between thiols and a variety of unsaturated polymers (including polybutadiene, polyisoprene, and emulsion-polymerized styrene-butadiene rubber) were reported.5 More recent work has included, for example, the photochemical crosslinking of polystyrene-polybutadiene-polystyrene block copolymers with multifunctional thiols.6 2 N.B. Cramer & C.N. Bowman, Kinetics of Thiol-Ene and Thiol-Acrylate Photopolymerizations with Real-Time Fourier Transform Infrared, 39 J. Polymer Sci.: Part A: Polymer Chem. 3311 (2001) (citing Posner, 38 Ber. 646 (1905)). 3 F. Ashworth & G.N. Burkhardt, Effects Induced by the Phenyl Group. Part I. Addition of Polar Reagents to Styrene and the Behavior of the Halogenated Ethylbenzenes, J. Chem. Soc. 1791, 1797 (1928). 4 F.R. Mayo & C. Walling, The Peroxide Effect in the Addition of Reagents to Unsaturated Compounds and in Rearrangement Reactions 27 Chem. Rev. 351, 387-394 (1940). 5 G.E. Serniuk et al., Study of the Reaction of Buna Rubbers with Aliphatic Mercaptans, 70 J. Am. Chem. Soc. 1804 (1948). 6 See, e.g., C.E. Hoyle et al., Thiol-Enes: Chemistry of the Past with Promise for the Future, 42 J. Polymer Sci.: Part A: Polymer Chemistry 5301, 5324-26 (2004) and references cited therein; C. Decker, Light- Induced Crosslinking Polymerization, 51 Polymer Int’l 1141 (2002) and references cited therein; C. Decker, Kinetic Study and New Applications of UV Radiation Curing, 23 Macromol. Rapid Communications 1067, 1085-91 (Cont’d) Appeal 2013-002146 Application 11/530,338 3 The ’338 Application describes methods for producing cross-linked elastomers through photoinitiated cross-linking of a latex with a thiol or a selenol. Spec. Claim 1 is representative of the ’338 Application’s claims and is reproduced below: 1. Method for producing a cross-linked elastomer comprising: producing a mixture of at least one latex and at least one starter component for initiating a cross-linking reaction or at least one latex that bears at least one starter component or starter group, the starter component or starter group comprising at least one photoinitiator which through irradiation with electromagnetic radiation in the ultraviolet and/or visible spectral range forms a reactive species, and the mixture includes at least one cross-linking auxiliary comprising at least one of a thiol and a selenol; and irradiating the mixture with UV light and/or visible light to at least partially cross-link the at least one latex. App. Br. 40 (Claims App’x) (some paragraphing and indentation added). (2002); C. Decker & T. Nguyen Thi Viet, High-Speed Photocrosslinking of Thermoplastic Styrene-Butadiene Elastomers, 77 J. Applied Polymer Sci. 1902 (2000); C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers IX. Thiol-Ene Polymerization of Styrene- Butadiene-Block-Copolymers, 41 Polymer 3905 (2000); C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers, 8: The Thiol-Polybutadiene System, 200 Macromol. Chem. Phys. 1965 (1999); C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers, 7: Styrene-Butadiene Block Copolymers, 200 Macromol. Chem. Phys. 358 (1999). Appeal 2013-002146 Application 11/530,338 4 REJECTIONS On appeal, the Examiner maintains the following rejections: 1. Claims 1-3, 8-10, 48-51, 64-69, 75, 79, 80, and 82 are rejected under 35 U.S.C. § 103(a) as unpatentable over the combination of Dove7 and Warner.8 FR 2; Ans. 3-4. 2. Claims 4-6 are rejected under 35 U.S.C. § 103(a) as unpatentable over the combination of Dove, Warner, and McGlothlin.9 FR 5; Ans. 3-4. 3. Claim 7 is rejected under 35 U.S.C. § 103(a) as unpatentable over the combination of Dove, Warner, McGlothlin, and Park.10 FR 6; Ans. 3-4. 4. Claims 72-74 and 76-78 are rejected under 35 U.S.C. § 103(a) as unpatentable over the combination of Dove, Warner, and Lipinski.11 FR 7; Ans 3-4. DISCUSSION Each of the Examiner’s rejections is based upon the combination of Dove and Warner as describing or suggesting the elements of independent claims 1 and 9. FR 2-4, 5, and 7. As Appellants argue, App. Br. 18, the 7 US 5,691,446, issued November 25, 1997. 8 US 3,338,810, issued August 29, 1967. 9 US 6,329,444 B1, issued December 11, 2001. 10 US 2005/0171282 A1, published August 4, 2005. 11 US 2006/0253956 A1, published November 16, 2006. Appeal 2013-002146 Application 11/530,338 5 Examiner has not provided sufficient facts and explanations to support the combination of these references. We, therefore, reverse rejections 1-4.12 The Examiner found: Dove teaches a method for forming a cross linked elastomeric material by mixing a natural rubber latex (polyisoprene) with crosslinking starters and accelerators such as thiol compounds and heating (at least partial crosslinking) the compounds to produce a suitable dip forming latex (col[.] 12 ln[.] 33-44) followed by curing the latex after article formation using normal manufacturing procedures (col[.] 13 ln[.] 13-22). Dove does not teach the use of radiation to cure or the use of a photoinitiator for the accelerant. FR 2-3 (emphasis added). The Examiner further found: Warner teaches a method for crosslinking non-conjugated polymers prepared from dienes such as isoprene (col[.] 2 ln[.] 31-40) by a reaction of a thiol with the diene polymer irradiated by UV light (col[.] 2 ln[.] 1-15, col[.] 1 ln[.] 50-66). Warner teaches the inclusion of photoinitiators (col[.] 2 ln[.] 55-60). It would have been obvious to one of ordinary skill in the art at the time the invention was made to use the thiol and photoinitiator of Warner to crosslink the non- conjugated rubber of Dove, because both relate to crosslinking irradiation treatment of nonconjugated rubbers with a thiol compound presenting a reasonable expectation of success, Dove teaches using known processing techniques for curing prompting one of ordinary skill to look to related art, and doing so presents a simple substitution of known curing techniques (UV substituted for heat) yielding predictable results. Id. at 3. 12 In view of the disposition of this appeal, the Examiner may wish to consider the effect, if any, of the references cited in footnotes 2-6 of this opinion on the patentability of the ’338 Application’s claims. Appeal 2013-002146 Application 11/530,338 6 The evidence in Warner does not adequately support the Examiner’s finding that substituting UV curing for heat curing in the process of Dove would have been reasonably expected to be a simple substitution yielding predictable results. Warner describes the material produced using its curing method as producing clear, hard, solid materials. E.g., Warner, col. 1, ll. 29-42; col. 3, ll. 6-11 and ll. 27-28. Indeed, Warner states that “[t]he clear, colorless, solid product of this invention due to its rigidity and inertness is superbly suited for can coatings, pints [sic, paints], organic glass applications and the like.” Id. at col. 1, ll. 39-42. Appellants also argue that a person of ordinary skill in the art would not use a curing method that produces hard, rigid material in a process for production of elastomeric materials. App. Br. 18; Reply Br. 5-7. The Examiner responds: The combination of Dove and Warner is based on the similar reaction taking place in the thiol crosslinking of natural rubber in Dove and the thiol crosslinking of polyunsaturated dienes in Warner. One of ordinary skill presented with the choice of UV curing a polydiene or heat curing a polydiene would find it obvious to perform the crosslinking with either method. That the end use product of Warner is different than the end use product of Dove does not detract from the teachings of the crosslinking reaction. Ans. 11 (emphasis added). We are not persuaded by the Examiner’s argument. The Examiner assumes that the reactions taking place in Dove and Warner are similar. This assumption is not adequately supported. Given the vast difference in the properties of the resulting materials, a person of ordinary skill in the art likely would believe that the curing reactions in Dove and Warner are very Appeal 2013-002146 Application 11/530,338 7 dissimilar. For example, the person of ordinary skill in the art might think that the reactions might be proceeding through different mechanisms.13 The Examiner has provided neither evidence nor a convincing argument to the contrary. Thus, we cannot agree with the finding that a person of ordinary skill in the art at the time of the invention would have used Warner’s cross- linking method in Dove’s process for making an elastomeric material. CONCLUSION For the reasons set forth above, we reverse the Examiner’s rejections of the each of the pending claims in the ’338 Application. REVERSED sl 13 Early work on reactions between thiols and unsaturated compounds suggests that the acid-catalyzed and peroxide-catalyzed reactions have different mechanisms. E.g., R. Brown et al., The Addition of Thioacetic Acid to Unsaturated Compounds J. Chem. Soc. 2123 (1951); V.N. Ipatieff & B.S. Freidman, Reaction of Thiol Compounds with Aliphatic Olefins 61 J. Am. Chem. Soc. 71 (1939); V.N. Ipatieff et al., Reaction of Aliphatic Olefins with Thiophenol 60 J. Am. Chem. Soc. 2731 (1938); S.O. Jones & E.E. Reid, The Addition of Sulfur, Hydrogen Sulfide and Mercaptans to Unsaturated Hydrocarbons 60 J. Am. Chem. Soc. 2452 (1938). Notice of References Cited                            U.S. PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''         ($     ) ($      ($                           ($                           ($                          * ($                          + ($                          , ($                          - ($                          . ($                          % ($                          / ($                          & ($                        FOREIGN PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''  !  #                0                                                                  1                                 2                                 (                                 3                               NON-PATENT DOCUMENTS   - !"4!5 63 6!"5 6  7 !6            7     8     9        # 5    " ! 5:550 (&;<=<=>  &&$''''  !"   #"(     (  3  ?0 30$@AB2C=$B== Notice of References Cited          11/530,338 2013-002146 1700 N.B. Cramer & C.N. Bowman, Kinetics of Thiol-Ene and Thiol-Acrylate Photopolymerizations with Real-Time Fourier Transform Infrared, 39 J. Polymer Sci.: Part A: Polymer Chem. 3311 (2001) (citing Posner, 38 Ber. 646 (1905)). F. Ashworth & G.N. Burkhardt, Effects Induced by the Phenyl Group. Part I. Addition of Polar Reagents to Styrene and the Behavior of the Halogenated Ethylbenzenes, J. Chem. Soc. 1791, 1797 (1928). F.R. Mayo & C. Walling, The Peroxide Effect in the Addition of Reagents to Unsaturated Compounds and in Rearrangement Reactions 27 Chem. Rev.351, 387394 (1940). G.E. Serniuk et al., Study of the Reaction of Buna Rubbers with Aliphatic Mercaptans, 70 J. Am. Chem. Soc. 1804 (1948). Notice of References Cited                            U.S. PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''         ($     ) ($      ($                           ($                           ($                          * ($                          + ($                          , ($                          - ($                          . ($                          % ($                          / ($                          & ($                        FOREIGN PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''  !  #                0                                                                  1                                 2                                 (                                 3                               NON-PATENT DOCUMENTS   - !"4!5 63 6!"5 6  7 !6            7     8     9        # 5    " ! 5:550 (&;<=<=>  &&$''''  !"   #"(     (  3  ?0 30$@A2B=C$==C Notice of References Cited          11/530,338 2013-002146 1700 C.E. Hoyle et al., Thiol-Enes: Chemistry of the Past with Promise for the Future, 42 J. Polymer Sci.: Part A: Polymer Chemistry 5301, 532426 (2004). C. Decker, Light-Induced Crosslinking Polymerization, 51 Polymer Int’l 1141 (2002). C. Decker, Kinetic Study and New Applications of UV Radiation Curing, 23 Macromol. Rapid Communications 1067, 108591 (2002). C. Decker & T. Nguyen Thi Viet, High-Speed Photocrosslinking of Thermoplastic Styrene-Butadiene Elastomers, 77 J. Applied Polymer Sci. 1902 (2000). Notice of References Cited                            U.S. PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''         ($     ) ($      ($                           ($                           ($                          * ($                          + ($                          , ($                          - ($                          . ($                          % ($                          / ($                          & ($                        FOREIGN PATENT DOCUMENTS    ! !"  !  # $!" $%   &&$''''  !  #                0                                                                  1                                 2                                 (                                 3                               NON-PATENT DOCUMENTS   - !"4!5 63 6!"5 6  7 !6            7     8     9        # 5    " ! 5:550 (&;<=<=>  &&$''''  !"   #"(     (  3  ?0 30$@AB2C=D$B==D Notice of References Cited          11/530,338 2013-002146 1700 C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers IX. Thiol-Ene Polymerization of Styrene-Butadiene-Block-Copolymers, 41 Polymer 3905 (2000). C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers, 8: The Thiol-Polybutadiene System, 200 Macromol. Chem. Phys. 1965 (1999). C. Decker & T. Nguyen Thi Viet, Photocrosslinking of Functionalized Rubbers, 7: Styrene-Butadiene Block Copolymers, 200 Macromol. Chem. Phys. 358 (1999). R. Brown et al., The Addition of Thioacetic Acid to Unsaturated Compounds J. Chem. Soc. 2123 (1951). Notice of References Cited                            U.S. PATENT DOCUMENTS        !   " # ! #$   %%#&&&&         '#     ( '#      '#                           '#                           '#                          ) '#                          * '#                          + '#                          , '#                          - '#                          $ '#                          . '#                          % '#                        FOREIGN PATENT DOCUMENTS        !   " # ! #$   %%#&&&&   "                /                                                                  0                                 1                                 '                                 2                               NON-PATENT DOCUMENTS   ,  !3 4 525 !4 5  6  5            6     7     8        " 4    !  4944/ '%:;<;<=  %%#&&&&   !   "!'     '  2  >/ 2/#?@A1B\ a step supposed to be followed by either (1) the addition of water, or (2) isomerisation to styrene, or (3) addition of phenylmethyl- carbinol. He comments on the mechanism previously discussed by Phillips (J., 1925, 127, 2567) dependent on the production of Rl -C- from a halogen compound by the removal of the halogen ion, but he retains the conception that, in such an ionisation, the hydrogen atom It, accompanies the halogen ion. Rl\ 4- R;/' R2 .. R, .. .. K2 .. .. R,:C:H -+ R,:C* + H:C1: not R,:C:K -I- :C1: .. .. .. : Cfl : .. The inversion in the formation of phenylmethylcarbinol could be accounted for (compare Fischer, Ber., 1907,40,489 ; Lapworth and Mottram, Hem. Munchester PhiE. Xoc., 1927, 71, 63) if the removal of the halogen ion were dependent on one of the frequent approaches of the water or other solvent molecules, to the side of the a-carbon atom away from the halogen, occurring simultaneously with a rare Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online 1796 ASHWOR!PE BND BmEfCHARDT : polariaation of the molecule sufficient to ionise the halogen under these conditions. This mechanism would also account for other observations which Ward discusses, namely, that solutions of potass- ium cyanide, ammonia, and potassium acetate in ethyl alcohol all give a-phenyldiethyl ether as the main product. Many investigators (compare Conant and Kerner, J. Amer. Chem. Soc., 1924,46,232) have observed that phenyl and carbonyl groups have similar effects on the halogen atoms in systems of the form R*[CH,],*Cl (R=phenyl, carbethoxyl, etc .), phenyl being less powerful than any carbonyl group, and the results of the present investigation also show such similarities. It is to be noted from a theoretical point of view, however, that while the general effect of phenyl and carboxyl is in the same direction relative to hydrogen, the conjugative effects will depend on the nature of the side chain and the various factors may operate in different ways in different reactions of halogen atoms in the side chain. The a-halogen in alkylbenzenes forms with the nucleus a type of allyloid system - d9--N (V) (Allan, Ozford, Robinson, and Smith, Zoc. cit.) in which the polarisation of the nucleus assists the tendency of the halogen atom to separate with a negative charge. This is in aocorcl with the results of the investigations on the hydrolysis of substituted benzyl halides (Ann. Reports, 1927, 24, 156), and also with the capacity of the phenyl group for stabilising a positively charged a-carbon atom which is clearly shown in the positive triphenylmethyl ion. The opposite polarisation (type A, p. 1793) and the general effect may account for its less powerful effect in stabilising a negative charge in the sodium derivative of triphenylmethyl. I n the case of the p-halogenated ethylbenzenes the effect of the phenyl group on the halogen atom, here separated from it by two single bonds, appears to be slight, judged by the low reactivity to potassium iodide in acetone (Conant and Kerner, Zoc. cit.). Assuming that the separation of the halogen ion is the first process in the production of styrene by the action of aqueous-alcoholic potassium hydroxide on w-bromostyrena, this reaction will be dependent on the ease with which the electrons of the a-carbon atom co-ordinate with the w-atom to form the double bond, as well as on the ease with which the halogen atom is removed. The conjugation with the nucleus (type C, p. 1793) will facilitate such co-ordination and the removal of the proton will follow. On the other hand, this reaction may be dependent on the tendency of the a-hydrogen to ionise, which will be increhsed by the opposite type of polarisation and the general effect. cBr I=/ Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online EFFECTS INDUCED RY THE PHENYL GROUP. PART I. 1797 The repetition of Posner's experiments gave the products which he obtained, and the snlphone from the addition of phenyl mercaptan to styrene was identical with that obtained by the action of sodium benzenesulphinate on p-bromoethylbenzene. The sulphide obtained from the addition was also shown to contain less than 2% of the a-isomeride, by comparing the boiling point and refractive index with those of the two synthetic sulphides. It is therefore clear that phenyl mercaptan adds to styrene in the opposite sense to hydrogen bromide and other polar reagents, and the theoretical considerations which satisfactorily correlate those normal additions will not serve in this case if the reaction takes place between the plarised ethylene bond and the hydrogen and thiophenoxide ions. I n a search for other evidence in favour of a special mechanism for this reaction the course of the addition of phenyl mercaptan to styrene was folIowed by observations of the very large contraction in volume which takes place during the reaction and also bydeter- minations of the depression of the freezing point of benzene caused by iuixtures of phenyl mercaptan and styrene. The rate of addition is greatly dependent on the intensity of illumination, and the increase in rate produced by exposure to sunlight continues for some time after the reaction vessel is returned to diffused light ; but the reaction was not stopped in the dark and the product was always phenyl p-phenyl- ethyl sulphide. A small percentage of piperidine had such a powerful retarding action on the addition that no change in volume was observed in diffused daylight even after several days, although a contraction took place in sunlight. The action of light is well known to produce or enhance abnormal polarisations and depolaris- ations in a. wide range of reactions, and experiments are in progress with a view to obtaining evidence in favour of one or other of the possible alternative mechanisms for the interaction of mercaptans with ethylenic hydrocarbons. A remarkable isomerisation was observed in an attempt to pre- pare phenyl- p-phenylethylsulphone (111) by reducing phenacyl- phenylsulphone (VI) with zinc amalgam and hydrochloric acid : a liquid sulphide was obtained which on oxidation with hot chromic acid was converted into phenyl-a-phenylethylsulphone (I). A similar migration of groups takes place when u-bromoacetophenone is heated with yellow ammonium sulphide (Willgerodt, Ber., 1889, (vT.) COPh*CH,*SO,Ph -+ (11.) or CHPh(CH,)*SPh (VII.1 -3 (I.) E X P I< R I M E N T A L. 22,534). Preparation of the Monohalogen Derivatives of Ethylbenzene.- Phenylmethylcarbinol (b. p. 100"/15 mm. ; 3 : 5-dinitrobenzoate Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online 1798 ASHWORTH AND BURKHARDT : m. p. 95O), made by the action of magnesium phenyl bromide on acetaldehyde, was converted by gaseous hydrogen bromide into a-bromoethylbenzene. This, by the action of silver acetatc suspended in ether, readily yielded phenylmethylcarbinyl acetate which on hydrolysis gave the carbinol free from any detectable proportion of p-phenylethyl alcohol, as was shown by the properties of its 3 : 5-dinitrobenzoate and the entire absence of the fragrance associated with the isomeric compound. The P-chloroethylbenzene used in some of the experiments described later mas made by converting glycol chlorohydrin into its p-toluenesulphonate and leaving this in contact with a cold ethereal solution of magnesium phenyl bromide for 36 hours; about an equal weight of a liquid, which was probably ethylene chlorobromide, was also produced (see Ferns and Lapworth, J., 1912, 101, 273). p-Bromoethylbenzene was made by heating 8-phenylethyl alcohol with constant-boiling hydrobromic acid for 24 hours; it had b. p. 97"/12 mm., and was quite free from the pungent odour characteristic of the a-bromo-compound. Action of Aqueous-alcoholic Potassium Hydroxide on the Isomeric Brmethy1benxenes.-The two bromo-compounds, when heated on the water-bath under precisely similar conditions with aqueous-alcoholic potassium hydroxide (about 0-4N), lost all the halogen (as deter- mined by the decrease in alkaline titres) within 30 minutes. I n the product from the a-bromo-compound, no styrene whatever could be detected. The product from the p-bromo-compound was almost pure styrene and absorbed bromine (1 mol.), yielding stlyrene dibromide. Addition Reactions of #tyrene.-The styrene was made by Howard's method from cinnamic acid (J., 1861, 13, 136). (A.) Addition of hydrogen bromide. A mixture of styrene and a cold saturated solution of hydrogen bromide in glacial acetic acid was kept over-night and then poured into water, and the oil obtained was distilled (b. p. S5O/lS mm.). The product was nearly pure a-bromo- ethylbenzene, as it was converted, by means of silver acetate and alkali successively, into phenylmethylcarbinol which gave the characteristic 3 : 5-dinitrobenzoate (m. p. 95") and had no detectable odour of the isomeric p-phenylethyl alcohol (3 : 5-dinitrobenzoate, m. p. 108"). (B.) Addition of ammonium hydrogen sulphite. An emulsion of styrene (20 g.) with N/Q-ammonium hydrogen sulphite (2 1.) and kieselguhr (100 g.) was kept with frequent shaking for 10 days. After filtration the liquid was boiled with excess of barium hydroxide until ammonia was expelled and worked up for soluble salts (com- pare J., 1925, 127, 307); 8 g . (yield, about 16%) of colourless barium a-phenyZethanesuZphmate were obtained in bundles of small Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online EFFECTS INDUCED BY THE PHENYL GROUP. PART r, 1799 needles [Found: H20, 6.8; Ba (in anhydrous salt), 27.1. (C8H,SO,),Ba,2H2O requires H,O, 6.6%. (C8H,S0,),Ba requires Ba, 27-0%]. No alkyl sulphite could be detected (compare Zoc. cit.). A portion of the salt was converted into the amide, which separated from alcohol in feathery crystals, m. p. 121". The sulphonic derivatives were identical with those obtained from the ammonium salt produced when a-bromoethylbenzene was heated with aqueous ammonium sulphite for several hours (Found for the barium salt : H,O, 6.9 ; Ba, in the anhydrous salt, 2643%. Amide, feathery crystals, m. p. 121", mixed m. p. 121"). As a-bromoethylbenzene shows little or no tendency to become converted into styrene, the foregoing results indicate clearly that the sulphonic salt from styrene and hydrogen sulphite is the ac-deriv- ative of ethylbenzene. The conclusion, however, was checked by preparing the p-sulphonic acid. p-Bromoethylbenzene (8 g.) was boiled with aqueous ammonium sulphite, and the solution worked up as usual with barium hydroxide. Barium p-pheng1ethanesuJphonate (5 g.) was isolated in large, glisten- ing plates [Found : H,O, 3.4; Ba (in anhydrous salt), 26.8. (C8H9S0,),Ba,H,0 requires H@, 3.4%. (C8H,S0,),Ba requires Ba, 27.0%]. ( C . ) Addition of phenyl m,ercaptan. The product obtained from styrene and phenyl mercaptan by Posner's method (Zoc. cat.) gave, on oxidation with cold dilute potassium permanganate solution and excess of sulphuric acid, the pure p-sulphone, m. p. 56-57", identical with that prepared from P-bromoethylbenzene. Phenyl a- and p-phenylethyl sulphides and the sulphide obtained by fractionating the addition product mere unattacked by hot or cold neutral potassium permanganate, even when it contained manganese dioxide in suspeiision. The addition of bromine or of excess of dilute sulphuric acid caused rapid reduction of the per- manganate. If the oxidation was carried out hot, a mixture of phenacylphenylsulphone (VI), m. p. 94-95", and phenyl-p-phenyl- ethylsulphone was obtained. Phenyl-P-phenylethylsulphone (111) was prepared from P-bromo- ethylbenzene by boiling with an alcoholic solution of sodium benzenesulphinate for 12 hours and separated from carbon tetra- chloride in feathery crystals, m. p. 56-57' (Found : s, 13.2. Calc. for Cl4HI4O2S : S, 13.0%). The sulphones could not be prepared from the chloroethylbenzenes, as the a-chloro-compound was con- verted into styrene and the p-isomeride was unchanged. Pure phenyl 01- and p-phenylethyl sulphides were prepared by heating sodium thiophenoxidc (1.1 mols.) in alcoholic solution with 6- or p-bromoethylbenzene (1.0 mol.), pouring the mixture into The sulphonamide melted at 124". Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online 1800 ASHWORTH AND BUILZ(HARDT: water, and extracting the product with carbon tetrachloride. The sulphides were distilled, heated for 3 hour under reduced pressure at 100" in a stream of carbon dioxide, and redistilled. The a-compound had b. p. 163-164"/15 mm. and n, 1.6042; the p-compound, b. p. 188-189"/15 mm. and nD 1.6082. The whole product of the reaction between styrene and pheiiyl mercaptan was washed with alkali, dried, and ftactionated. The distinction between unchanged styrene, sulphide, and a small residue (probably metastyrene) was clearly marked. The middle fraction (yield, go%), b. p. 187-190"/15 mm., heated in a stream of carbon dioxide under reduced pressure a t 100" and refractionatetl, gave the following refractive indices : first 6 drops, 1.6079; 9 c.c., 1.6082; 2 drops, 1.6085; residue, 1.6088. Fractions 1 and 2 after refractionation gave : first 2 drops, 1.6073; 10 drops, 1.6081 ; thirteenth drop and the remainder, 1.6082, This indicates less than 2% of ol-isomeride in the product. Experiments on the velocity of the reaction and phutocatulysts. During the reaction of styrene and phenyl mercnptan a lOy, eontraotion takes place, and the course of the reaction was followed by cooling a mixture of styrene (11 c.c.) and phenyl mercaptan (10 c.c.) in a glass tube (0.2 to 0.5 cm. bore) in running water, marking the level of the meniscus, sealing the tube, and marking the level from time to time. As the velocity is influenced by light, the two reagents were distilled in only the light of a Bunsen flame (the mercaptan was distilled in hydrogen to prevent the formation of disulphide), and two tubes were prepared as described above, marked and sealed in a similar light, and left in a dark room. The part of each tube containing the liquid was kept inside rubber tubing down which water was running and the subsequent marks were made with a minimum of light in the dark room. The two reactions proceeded at the same rate. After 173 hours one tube was taken into sunlight!, the water-cooling being continued. The contraction was nearly doubled in the next 15 minutes and was practically complete after 1 hour. The other tube in the dark showed no signs of reaching the same limit after 3 weeks. Time (hrs.) ............... 0.1 1.5 2.6 14.0 17.5 17-75 18-0 18.95 Contraction (cm.) in tube I in dark ...... 0.5 0.75 1-05 2.16 2.36 4-O* 4.4* 4.7* Contraction (cm.) in tube I1 in dark ...... 04 0.8 1.1 2.16 2.3 Time (days) ............... 1.6 2.5 3.5 4.5 6.5 8.6 22-5 26.5 Contraction (cm.) in Contraction (cm.) in tube I in light ...... 4.9 4.9 4.9 tube I1 in dark ...... 2.6 2.96 3.25 3.4 3.6 3-75 3-8 3.8 * In gunlight. Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online EFFECTS INDUCED BY THE PHENYI, GROUP. PART I. 1801 The phenyl mercaptan remaining in the reaction carried out in the dark was extracted in alkali and the sulphide which was distilled out of the residue was shown to contain no detectable amount of the a-isomeride by determinations of the refractive index and boiling point. A side tube was inserted about 70 em. from the lower end of the reaction tube, the mixture was introduced until it overflowed from the side tube, and a similar experiment was carried out entirely in the dark. The speed was similar to that in the previous experi- ments before taking into sunlight. Cryoswpic evidence on the coume of the reaction [with CECIL ROBINS]. The freezing points were taken of solutions of phenyl mercaptan and of styrene in suitable amounts of benzene, and also the freezing points of the mixed solutions, with a view to deter- mining whether any considerable amount of association took place quickly, followed by a photocatalytic depolarisation. No such association was observed and the rate of combination appeared to be in agreement with t'he velocity measured by contraction. In this case, however, i t was clear that the reaction did not slow down again at once when the tube was returned to diffused light. Action of Alcoholic Potassium Hydroxide on Styrene Dibromide.-- Styrene dibromide (Qlaser, Annulen, 1870, 154, 164) (26.4 g.) was boiled in alcoholic solution with potassium hydroxide (1 mol.), the alcohol removed by distillation, and the oily product distilled in a vacuum. No fraction was obtained between or-bromostyrene and styrene dibromide. The product was almost instantly attacked by boiling formic acid, the odour of acetophenone becoming perceptible. As p-bromo- styrene was found to be very stable towards boiling formic acid, undergoing but slight hydrolysis after 20 minutes, this observation, which fhds a parallel in the ready hydrolysis of a-chlorostyrene by hydrochloric acid, suggested a means of detecting p-bromostyrene when mixed with the cc-isomeride. 0-774 G. of the lower-boiling portion of the product from styrene dibromide was boiled for about 5 minutes with formic acid, and sodium acetate and semicarbazide hydrochloride were then added ; acetophenonesemicarbazone (0.653 g. ; 92 yo of the theoretical yield) was deposited. It is evident that at least 92% of the oil consisted of or-bromostyrene. No trace of p-bromostyrene (easily detected by its powerful hyacinth-like odour) could be found in the liquor after the acetophenone had been removed. A s a small quantity of unchanged styrene dibromide was present, having escaped the fractionation process, it is clear that removal of hydrogen Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online 1802 BRIERS AND CHAPMAN : THE INFLUENCE O F THE bromide from styrene dibromide takes place almost solely in one direction. The authors gratefully acknowledge their indebtedness to Professor Lapworth for suggesting this investigation and for his advice and interest in it. ‘FEE UNIVERSITY, MAWCIFESTER. [Received, March 29th, 1928.1 Pu bl is he d on 0 1 Ja nu ar y 19 28 . D ow nl oa de d by U S Pa te nt & T ra de m ar k O ff ic e on 1 8/ 02 /2 01 5 22 :0 9: 05 . View Article Online THE PEROXIDE EFFECT I N THE ADDITION O F REAGENTS TO UNSATURATED COMPOUNDS AND I N REARRANGEMENT REACTION3 FRANK R . MAY0 AND CHEVES WALLING' George Herbert Jones Laboratory. Universi ty of Chicago. Chicago. I l l inois Received J u l y 1. 1940 CONTENTS I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 I1 . The addition of halogen acids to unsaturated compounds . . . . . . . . . . . . . . . . . 352 A . Normal and abnormal additions of halogen acids . . . . . . . . . . . . . . . . . . . . . . 352 . . . . . . . . . . . . . . . . 352 2 . The peroxide effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 3 . The normal addition of halogen acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 B . The peroxide effect in the addition of hydrogen bromide. . . . . . . . . . . . . . 359 1 . Products of addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 2 . Rates of addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 3 . Catalysts of abnormal addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367 4 . Inhibitors of abnormal addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 5 . The bromine-atom mechanism of abnormal addition . . . . . . . . . . . . . . . . 372 6 . Other mechanisms proposed for abnormal addition . . . . . . . . . . . . . . . . . 377 7 . The influence of experiment. a1 conditions on additions of hydrogen bromide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 8 . The addition of hydrogen bromide to cyclopropane . . . . . . . . . . . . . . . . . 386 I11 . The addition of mercaptans and thio acids to unsaturated compounds . . . . 387 A . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387 B . The abnormal addition of mercaptans and thio acids . . . . . . . . . . . . . . . . . . 388 C . The normal addition of mercapt'ans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391 1 . Status of the subject in 1930 . . . . . . . . . . . . . . . . . . D . Scope of the mercaptan-alkene reaction . . . . . . . . . . . . . A . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394 B . The addition of bisulfites to aliphatic unsaturated compounds . . . . . . . . . 394 C . The reaction of bisulfites with styrene., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397 V . The peroxide effect in rearrangements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399 A . Rearrangement of 1-bromo-2-butene and 3-bromo-1-butene . . . . . . . . . . . . 399 B . Rearrangement of a-bromoacetoacetic esters . . . . . . . . . . . . . . . . . . . . . . . . . . 401 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403 VI . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407 I . INTRODUCTION IV . The reaction of bisulfites with unsaturated compounds . C . Cis-trans isomerizations . . . . . . . The fact that oxygen and peroxides affect the direction of addition of Subse- ' Present address: E . I . du Pont de Nemours and Company, Inc., Wilmington, hydrogen bromide to allyl bromide was discovered in 1931 (53) . Delaware . 351 352 FRANK R. MAY0 AND CHEVES WALLING quent work has demonstrated that the same agents also determine the direction of addition of various addenda to a large number of other ethylene derivatives. Because of the many papers concerned with such phenomena and the variety of interpretations offered, a critical review of the experi- mental work and theoretical conclusions seems desirable. Recently, J. C. Smith (139, 140, 141) has reviewed the field fully, but from a different viewpoint. This review is based on the premise that several types of substitution and addition can proceed in solution by a t least two mechanisms: a chain reac- tion involving free radicals and occasionally atoms, and an ionic (or molec- ular) reaction. The chain hypothesis has shown that several apparently different types of reaction are actually closely related, and has led to the discovery of new syntheses. Although, in most instances, details of the ionic or molecular mechanism are not yet clear, the data available permit prediction of the products of many reactions. This review will be concerned with the addition of hydrogen bromide, mercaptans, and bisulfites to unsaturated compounds, and with certain rearrangements. The discussion will refer almost entirely to liquid-phase reactions. The important effects of oxygen and peroxides on the bromina- tion and chlorination of hydrocarbons, acids, acid halides, and ketones, on chlorinations with sulfuryl chloride, and on carboxylation reactions will be omitted because rapid development of these fields has just begun. 11. THE ADDITION OF HALOGEN ACIDS TO UNSATURATED COMPOUNDS A. NORMAL AND ABNORMAL ADDITIONS OF HALOGEN ACIDS 1. Status of the subject in 1930 Previous to 1930, many hypotheses were current regarding the factors which controlled the direction of addition of unsymmetrical reagents t o ethylene derivatives. Particular confusion existed with respect to some addition reactions which could be used to help in deciding between con- flicting hypotheses. It was with the hope of reconciling some of the dis- cordant data that an intensive investigation of the addition of halogen acids to unsymmetrically substituted ethylenes was undertaken in this laboratory. The addition of hydrogen bromide to allyl bromide seemed of particular interest, as various workers had reported addition products ranging from nearly pure l12-dibromopropane to nearly pure 1 ,3-dibromopropane, even under supposedly identical experimental conditions. These investigators ascribed the discrepancies to variations in t,he temperature, the reaction time, or the concentration of hydrogen bromide, and to the presence of water or light (for references, see 53). Other factors which have been THE PEROXIDE EFFECT 353 thought to direct the addition of a halogen acid to an alkene are solvents (46, 114, 117), their dielectric constants or internal pressures (44), magnetic fields (17), and previous treatment of the alkene (for references on the elec- tromer controversy, see 77). In the vapor-phase reaction, surfaces and metal halides (15, 163, 164), oxidizing atmospheres (ll), and light (10) have also been supposed to play a r61e. It is noteworthy that most of the work on the addition of halogen acids has been carried out with hydrogen bromide, since hydrogen chloride often adds too slowly and hydrogen iodide frequently gives unstable addition products. This choice is unfortunate, because hydrogen bromide is the only halogen acid whose direction of addition to alkenes is affected by air and peroxides, and therefore the only one which can give results of doubtful significance. The following discussion will show the extent of this peroxide effect with hydrogen bromide and the absence of this effect with hydrogen chloride and hydrogen iodide. 2 . The peroxide eflect The peroxide effect was discovered during an investigation of the addi- tion of hydrogen bromide to allyl bromide (53). When these substances are allowed to react a t room temperature in the dark, the reaction may take either one of two courses: CH3CHBrCH2Rr (1) 7 L CHZ=CHCHzBr + HBr CHzBr CHzCHzBr (2) If the reactants are pure and freshly prepared, and if oxygen is excluded from the reaction vessel, reaction 1 takes place exclusively2 and several days are required for substantially complete reaction. If the reaction occurs in the presence of small quantities of oxygen or if peroxides are in- troduced either deliberately or by use of old allyl bromide, then reaction 2 takes place almost quantitatively in a few hours. As reaction 1 is that which takes place with pure reagents in the absence of catalysts, i t has been termed the normal reaction; reaction 2 is abnormal. Such a reversal of addition has been called a “peroxide effect”, and it has been observed in additions of hydrogen bromide to many ethylene derivatives. In order to show that the normal and abnormal reactions take place by different mechanisms, the characteristics of the normal addition reaction will be discussed before the abnormal reaction is taken up in detail. * Slight corrections (87) to the original work (53) justify this statement. 354 FRANK R. MAY0 AND CHEVES WALLING 3. The normal addition of halogen acids The normal product of the addition of hydrogen bromide to an alkene is uniquely and conveniently defined in terms of the addition of hydrogen chloride and hydrogen iodide, since these halogen acids are not susceptible to a peroxide effect and never add abnormally in the liquid phase. The addition of hydrogen chloride or hydrogen iodide and the normal addition of hydrogen bromide to an alkene or a halogenated alkene usually give a single product, the one which would be predicted from Markovnikov's rule (109). That is t o say, the halogen of the halogen acid always becomes attached to the least hydrogenated of the.two doubly bound carbon atoms. The product thus obtained is apparently the thermodynamically more stable one of the two possibilities (2O).3 The hydrogen chloride and hydro- gen iodide additions in tables 1 and 2 have been carried out since the dis- covery of the peroxide effect. In many instances antioxidants or perox- ides have been employed, or else the reaction has been carried out in a solvent with the object of altering the composition of the addition product. The addition of hydrogen chloride to propene has also been carried out in quartz apparatus in the presence of ultraviolet light (94). With three ex- ceptions, each of these additions yields only one addition product. One exception is l-bromopropene, which with both hydrogen chloride and hy- drogen iodide gives a mixture made up of one-third 1,l-dihalide and two- thirds 1,2-dihalide, thus supplying convincing evidence that (within the limits of experimental error) these two halogen acids always add in the same way. Another mixture is that obtained by the action of hydrogen chloride on 2-pentene. As will be shown in table 4, hydrogen bromide behaves similarly to 172-dialkylethylenes. The third example of a mixture is the addition of hydrogen chloride to 1,3-butadiene, resulting in 75 to 80 per cent of the 1,Saddition product and 20 to 25 per cent of the 1,4- addition product. This proportion is not affected by temperature over the range -78" to 25°C. or by the presence of acetic acid. An earlier com- pilation (52) showed that in general the addition of hydrogen chloride or hydrogen iodide to alkenes has been reported to give only one addition product. Michael and Leighton (115) claim to have obtained a small proportion of n-propyl iodide by the addition of hydrogen iodide to Isopropyl and propyl bromides isomerize in the liquid phase a t 250-275°C. to equilibrium mixtures containing about 30 per cent of primary halide (15); isobutyl and tertiary butyl bromides are known to behave similarly. The vapor-phase addition of hydrogen chloride to isobutene (2-methylpropene) at 270°C. gives an addition product con- taining mostly tert-butyl chloride but also about 8 per cent of isobutyl chloride (112). Thus a portion of the product obtained by the addition of a halogen acid a t high temperature disagrees with Markovnikov's rule. The nature of such reactions is now under investigation in this laboratory. 3 These statements do not hold for temperatures above 200°C. THE PEROXIDE EFFECT 355 CHsCHsCH=CHCHa < propene, and Ingold and Ramsden (44) claim to have obtained up to 24 per cent of primary iodide by carrying out this reaction in certain sol- vents. In this laboratory, repeated attempts (84) to confirm these claims have failed. It has been found that peroxides accelerate the rate of addi- tion of hydrogen iodide (62,84), but this effect is explained by the fact that TABLE 1 The addi t ion of hydrogen chloride to unsaturated compounds UNSATURATED COMPOUND Vinyl chloride.. . . Trichloroethylene Propene, . , , . , . , , . 1-Bromopropene.. 2-Bromopropene.. 1-Chloropropene.. 2-Chloropropene.. Allyl chloride. . , . Isobutylene.. . . . . 2-Pentene . . . . . . . . Trimethylethyl- ene. . . . . . . . , . . . . A2-Pentenoic acid A3-Pentenoic acid A4-Pentenoic acid Ala-Undecenoic acid. . , . , , . . . . , . 1,3-Butadiene 1-Hexyne . , . . . . . . 2-Chloro-1-hexene FORMULA CHz=CHCl C C12=CH C1 CH3CHxCHz J 'i CH&H=CHBr CH3CBr=CH2 CHaCH=CHCl CH3CCI=CHz CHzClCH=CHz (CHa) zC=CIIz CHz=CH(CHz) 8COOH CHz=CHCH=CHz I CH=C(CHz)3CHa C H F C C l (CHz)sCH3 ADDITION PRODUCT CH3CHCIz CC13CHzCl CHsCHCICHs CH3CHzCHBrCl (35%) CH3CHClCHzBr (65%) C€13CBrClCH3 CHaCHzCHClz CHaCClzCH3 CHzClCHClCH3 (CH3)aCCI (50%) (50%) CHsCHzCHClCHzCH3 CH3CHzCHZCHClCH3 (CH3)zCClCHzCHS CHsCHzCHClCHzCOOE CH3CHClCH2CH&OOE CHaCHCICHsCHzCOOE CHsCHCI(CH2) 8COOH CHaCHClCH-CHz CHaCH=CHCHzCl (75-80%) (20-25%) CHz=CCl (CIIt) aCH3 CH3CCls(CHz)aCH3 REPEREBCBIB peroxides liberate iodine from hydrogen iodide and that iodine is a catalyst for the normal addition of hydrogen iodide. Because i t has sometimes been difficult to eliminate completely the abnormal addition of hydrogen bromide, the use of hydrogen chloride or hydrogen iodide as standards for the normal addition has been found con- venient and reliable in doubtful cases (83). The normal addition products of many alkenes are listed in table 3. Present indications are that the addition of hydrogen fluoride t o alkenes 356 FRANK R. MAY0 AND CHEVES WALLING also follows Markovnikov’s rule (32, 102, 143). It will be indicated later that a peroxide effect with this reagent is very unlikely. The available data justify only the following general statements as to the relation between the rate of the normal addition and the structure of the alkene. In the case of any one alkene, hydrogen chloride adds more slowly than hydrogen bromide, and the latter usually adds more slowly than hydrogen iodide. In the normal addition of hydrogen bromide (on which the most information is available), substitution of the ethylene hy- drogens by alkyl or phenyl groups increases the rate of addition, whereas substitution by halogen retards the addition. Even substitution of hydro- gen on an adjacent carbon atom by halogen retards addition. Some data to illustrate these statements appear in table 5. TABLE 2 The addit ion of hydrogen iodide to unsaturated compounds UNSATURATED COMPOUND ~~ Vinyl chloride.. . . . . . . . . Propene. . . . . . . . . . . . . . . . l-Bromopropene . . . . . . . . Allyl bromide. . . . . . . . . . l-Butene . . . . . . . . . . . . . . . Trimethylethylene . . . . . 4,4-Dimethyl-l-pentene Undecenoic acid. . . . . . . . Allyl chloride. . . . . . . . . . 8-Iodocrotonic acid. . . . . Tetrolic acid. . . . . . . . . . . FORMULA C H F C H C l CHICH=CHZ CHICH=CHBr CH2=CHCH2Br CHz=CHCH*CI CEIFCHCH~CH~ (CHI)zC=CHCHs (CHI) sCCHZCH=CH: CH?CH(CHz) 8- CHsCI=CHCOOH CHICECCOOH COOH ADDITION PBODUCT CHICHIC1 CHICHICHI CH3CHICHtBr (64%) CHSCHzCHIBr (36%) CH3CHICHzBr CHaCHICHzCl CHsCHICHzCHs (CHI)ZCICH~CH~ (CHI) ICCHZCH ICHs CHsCHI(CHz)&OOH CHICI~CHZCOOH CH,CI=CHCOOH Since a large portion of this review will be concerned with the mechanism of the abnormal addition of hydrogen bromide, it seems appropriate to discuss first the mechanism of normal addition. Maass and his associates have investigated the reaction of hydrogen bromide with propene and of hydrogen chloride with propene and the butenes in the absence of solvents. They find that those alkenes which (as indicated by the melting-point curves of mixtures) form 1 : 1 complexes with halogen acids a t low tempera- tures react around room temperature to give addition products more easily than those which do not form complexes, that the addition iscomplicated by a dimerization reaction so that rate equations could not be established, and that excess halogen acid is more effective than excess alkene in acceler- ating the addition (19, 106, 107). No electrical conductance could be ob- TEE PEROXIDE EFFECT 357 served in these mixtures, and it was found that hydrogen chloride ac- celerated the addition of hydrogen bromide (106). The rate of addition of hydrogen chloride to propene increases with temperature from about -80" to +45"C. Between 45°C. and the critical temperature of the mix- ture (70"C.), the reaction has a negative temperature coefficient which Holder and Maass (39) have ascribed to loss of orientation or "structure" in the liquid. These liquid-phase reactions, as well as the normal addition of hydrogen bromide to allyl bromide (53), are mostly or entirely homo- geneous. Above the critical temperature, a t high pressure, the hydrogen chloride-propene reaction is very slow, and has a positive temperature coefficient (39). The vapor-phase additions of halogen acids to alkenes are heterogeneous, bimolecular reactions (19, 39, 99), accelerated by metal halides (15, 163). Equilibria and activation energies have been reported in a few instances (48, 99). Kinetic investigations of the addition of halogen acids to alkenes (113, 122) in inert solvents indicate that the reaction is largely, if not entirely, of an order higher than the second. Consequently, inert solvents greatly reduce the rate of addition. That the rates are lower in ether and in diox- ane than in hydrocarbon solvents is apparently a consequence of combina- tion of the halogen acid with the solvent (122). Among oxygen-free sol- vents, the rate of addition increases in approximately the same order as the dielectric constants of the solvents, but in the presence of water or car- boxylic acids, the reaction may be faster than in the absence of a diluent (53). The existing data, inadequate as they are, nevertheless indicate that the conventional mechanism of normal addition, as described by Robinson (128) and Ingold (43), is an oversimplification. According to this mech- anism, the proton of the halogen acid (with or without ionization) becomes attached to one of the doubly bound carbon atoms, leaving the other as a positive carbonium ion which subsequently reacts with the halide ion. Ogg (123) has pointed out that such a positive ion would be configura- tionally unstable and therefore inconsistent with known trans additions of halogen acids to some ethylene bonds. He suggests that the negative halide ion adds first to give a configurationally stable negative carbonium ion which later adds a proton. Neither mechanism is satisfactory under conditions where the reaction is of higher than the second order. Sher- man, Quimby, and Sutherland (134) have discussed both a non-ionic bimolecular mechanism and a chain mechanism for the normal addition in the vapor-phase or in inert solvents. The chain mechanism is that to be described in section 11, B, 5 for the abnormal reaction; reasons for rejecting i t as a normal mechanism will subsequently become clear. No homo- geneous bimolecular addition has yet been observed in the vapor phase, 358 FRANK R. MAY0 AND CHEVES WALLING but the possibility of such a reaction cannot yet be wholly excluded for the liquid phase. Urushibara and Sinamura (151) have suggested a different chain mech- anism for the normal addition of hydrogen bromide: RCH=CH2 + H* 3 RCHCH3 (3) RCHCK + HBr -+ RCHBrCH3 + H. (4) Aside from the fact that this chain lacks any experimental support, i t must be discarded on thermodynamic grounds. As has been mentioned else- where (87), reaction 4 is endothermic by about 35 kg-cal., an impossible condition for a step in a rapid chain reaction. An explanation which seems to eliminate all of the above difficulties is that the reaction occurs, not simply between halogen acid and alkene, but between halogen acid and a halogen acid-alkene complex4 which may contain a hydrogen bond. Such a mechanism explains why the rate de- pends more upon the halogen acid than upon the alkene concentration (19, 39, 106). Because the complex would be expected t o be less stable at higher temperatures, the negative temperature coefficients observed in some instances (39, 111) can be accounted for. The increased reaction velocity in hydroxylic and carboxylic solvents may be due to the fact that reaction is easier when either the complex or the halogen acid is ionized. The effects of catalysts are consistent with this idea of the mechanism of normal addition. Anhydrous ferric and aluminum chlorides are the most powerful known accelerators for the addition of hydrogen chloride and hydrogen bromide, and are particularly useful with those alkenes which otherwise react very slowly. Tests have shown that these catalysts affect only the rate and not the product of addition; in their presence only the normal addition product is obtained (83) .5 Zinc, thallous, cobaltous, and ferrous halides are moderate accelerators of the addition of hydrogen bromide to allyl bromide; cadmium, lead, stannous, cuprous, and nickelous 4 Coffin, Sutherland, and Maass (18a) have previously considered this and other possibilities, but have rejected them in favor of the hypothesis that reaction takes place between two molecules of complex. They consider that the latter explanation best interprets the retardation of some additions by excess alkene. This possibility must still be admitted, but in view of the fact that the authors mentioned used no solvents and could obtain no rate constants, the simpler mechanism seems preferable for the present, since i t has some support from additions in solvents (113). 6 An anomalous case is reported by Schjhnberg (131), who considers that ferric chloride is a negative catalyst for the addition of hydrogen chloride to the three isomeric straight-chain pentenoic acids. That this is not a general rule for acids is shoFn by the fact that normal addition to Ala-undecenoic acid is accelerated by ferric chloride (139). Further work may clarify this question. THE PEROXIDE EFFECT 359 bromides (87) and platinum black (152) are weak accelerators of the same reaction. There has been little need to accelerate the addition of hydro- gen iodide to alkenes, but mercuric iodide and free iodine have been found effective (84). tert-Butyl isocyanide accelerates the addition of hydrogen bromide to allyl bromide (53), vinyl bromide (54), and vinyl chloride (57). This effect is probably due to the formation of some kind of an ammonium salt by the reaction of the isocyanide with hydrogen bromide, since di- methylammonium bromide and tetraethylammonium bromide have been found t o accelerate the addition of hydrogen bromide to allyl bromide (112). These various catalysts may alter the uncatalyzed mechanism in several ways. The metal halide (or iodine) may replace one molecule of halogen acid in the halogen acid-alkene complex; or by combining with the halogen acid, it may activate the proton so that the latter more easily forms complexes with alkenes. Whatever complex is formed, the activity of substituted ammonium halides suggests that the alkene complex may react with the halide ion of a salt more easily than with a halogen acid. B. THE PEROXIDE EFFECT IN THE ADDITION OF HYDROGEN BROMIDE I. Products of addition Table 3 lists those ethylene and acetylene derivatives for which the direction of addition of hydrogen bromide is known to be affected by oxygen or peroxides. In many instances, both products indicated have been obtained in a pure condition. In the others, more or less difficulty has been encountered in completely suppressing one of the competing additions, and mixtures have been obtained, their composition depending on the experimental conditions. In the whole of table 3, there is only one instance where it is well established that the normal addition product is a mixture: l-bromopropene gives about one-third 1 , l-dihalide and two- thirds 1 , 2-dihalide, a result already indicated for the additions of hydrogen chloride and hydrogen iodide. In the presence of peroxides, hydrogen bromide adds to give 1,2-dibromopropane exclusively. Table 3 shows that peroxides alter the direction of addition of hydrogen bromide to unsaturated hydrocarbons, halides, acids, and esters. Five examples indicate that the effect apparently applies t o acetylene as well as to ethylene derivatives. In most of the compounds cited, the double bond is a t the end of the carbon chain, but recent work with trimethyl- ethylene, 2-bromo-2-butene, and 2-methyl-2-nonadecene shows that the effect is not confined to terminal double bonds. Table 4 lists alkenes for whichit has not yet been possible to reverse the addition of hydrogen bromide. This table refers only to studies carried out since the discovery of the peroxide effect, where attempts have been made to obtain more than one product. There is some uncertainty about 360 FRANK R . MAY0 AND CHEVES WALLING . . . . . . . . , . . . . . . . . . . . . . . . , . . . . , . . . . THE PEROXIDE EFFECT 361 8 8 8 0 Y 2-A II x - q Y x 2 . . . . . . . . . . . . . . . . I : : : : : : I , . . . . 362 FRANK R. MAY0 AND CHEVES WALLING N 9 x x" E 9 9- E I x h u- . . . . . . . . . I . . . . . . . . . . . . . . . . . . . . . . . . x" u_ h THE PEROXIDE EFFECT 363 364 FRANK R. MAY0 AND CHEVES WALLING placing compounds in this table, because a past failure to obtain a second addition product does not prove that future attempts will likewise be failures. For example, early attempts to reverse the addition of hydrogen bromide to styrene failed, but improved technique gave up to 80 per cent of the abnormal product (162). It seems well established that peroxides do not affect the direction of addition of hydrogen bromide to the ethylene derivatives in section A of table 4. Here the normal addition products are 50-50 mixtures, and com- plete reversal of the addition gives the same result. Note that, as far as directing tendency is concerned, methyl has the same effect as ethyl or a long-chain alkyl with a substituent on the end.6 Thus there are few ex- amples of a peroxide effect upon non-terminal double bonds. It should be pointed out here that symmetrically substituted ethylenes, e.g., 2-butene, can give only one structurally distinct addition product, but this fact does not preclude the possibility that the product in question can be formed by two mechanisms. Alkenes containing a group which inhibits abnormal addition may fail to give an abnormal product. The only known example of such a com- pound is A1o-undecenol (in section B of table 4); primary alcohol groups are weak inhibitors of abnormal addition. Until more data on such compounds are available, i t should be concluded that the presence of an inhibiting group in an alkene makes abnormal addition less likely but not impossible. In section C are listed some doubtful cases in which attempts to obtain two addition products have been made. It is probably because the best experimental procedures have not yet been employed that the second products have not been detected. In the opinion of the reviewers, pro- penylbenzene, safrole, and (if isomerization’ can be retarded) l-bromo-2- butene should give two addition products. Because of the behavior of l-bromopropene, no comment is made on additions to 1 -bromohexene. Camphene is a special case, because its predicted abnormal addition product is unknown and because acids rearrange its carbon skeleton. Section D (table 4) consists of a, p-unsaturated acids and esters. Even when an improved technique was used, crotonic acid and crotonic ester gave only P-bromobutyric acid or ester, with no indication of a-brominated derivatives. Attempts to obtain a second addition product from the other acids listed have failed, but the improved technique has not yet been e The case of A3-pentenoic acid, listed in table 3, is not an exception to this statement, since the carboxyl group is separated from the double bond by only one methylene group. 7 This rearrangement, together with the addition of hydrogen bromide to buta- diene, will be considered in the section on rearrangements (V, A). Kormal addition gives one product. THE PEROXIDE EFFECT 365 employed. Explanations will be considered in the section on the mech- anism of the abnormal addition. A fact pertinent t o bromomaleic and bromofumaric acids is that 1 , l-dibromosuccinic acid is unknown and may be incapable of existence. Although no instance is known of a peroxide effect on an acid containing an a,@-double bond, there are indications of such an effect on the a,P-triple bonds of tetrolic and phenylpropiolic acids (table 3). The present section has thus far dealt only with the effects of oxygen and peroxides on the direction of addition of hydrogen bromide in liquid-phase reactions. The data on vapor-phase reactions are contradictory but of interest because, about ten years before the peroxide effect was reported in the scientific literature, suggestions of such an effect in the vapor phase appeared in Bauer’s patents. He claimed that the addition of hydrogen chloride, bromide, or iodide to acetylene and vinyl bromide, in the presence of an oxidizing atmosphere containing oxygen, ozone, chlorine, or nitrogen oxides (ll), or in the presence of light (lo), gives ethylene halides rather than ethylidene halides. Kharasch and Walker (97) have found that oxy- gen accelerates the addition of hydrogen bromide to propene, 2-pentene, and butadiene, and Ki’stiakowsky and Stauffer (99) mention a similar effect in the addition of “halogen acids” to ‘%obutene.” Brouwer and Wibaut (15) state that oxygen does not affect the direction of addition of either hydrogen bromide or hydrogen chloride to propene. Vapor-phase reactions need further investigation. 2. Rates of addition A consideration of the rates of some normal and abnormal addition reac- tions will be of assistance in showing that the two reactions are independent and competing, and that the susceptibility of an alkene to the peroxide effect is directly related to the rate of its normal addition reaction. No quantitative data are available, but table 5 has been compiled from experiments which have been carried out in the same laboratory under comparable conditions: that is, a t room temperature, with about 1.5 moles of hydrogen bromide per mole of alkene, and (except as noted) in the absence of solvents. The alkenes are listed approximately in order of increasing rate of nor- mal reaction with hydrogen bromide and of decreasing susceptibility t o the peroxide effect. When the normal addition is extremely slow (section A), traces of oxygen and peroxides can exert a maximum effect; unless the normal addition is specifically catalyzed, abnormal addition predominates, even in the presence of an antioxidant. When the normal addition is somewhat faster (section B), it can outrun the abnormal addition only when the latter is inhibited by an antioxidant. With allyl bromide (sec- 366 FRANK R. MAY0 AND CHEVES WALLING UNSATURATED COMPOUND Section A : Trichloroethylene.. . . . . . . . . . . 1-Bromopropene. . . . . . . . . . . . . . tion C), exclusion of air and ordinary purification of materials permit the normal to exclude the abnormal addition, although the latter occurs in the presence of air. With the alkenes in section D, air alone is ordinarily insufficient to influence the course of the reaction and definite admixture of peroxides is required to prevent fairly rapid formation of the normal addition product. In section E, the normal addition is so fast that it must Yield' Time per cent hours 0 1440 -- Very little TABLE 5 Rate of addition of hydrogen bromide to unsaturated compounds 1-Chloropropene . . . . . . . . . . . . . . Vinyl bromide . . . . . . . . . . . . . . . . Vinyl chloride.. . . . . . . . . . . . . . . Section B: Section C : Section D : Allyl bromide.. . . . . . . . . . . . . . . 2-Bromopropene. . . . . . . . . . . . . . 2-Chloropropene . . . . . . . . . . . . . . Propylene . . . . . . . . . . . . . . . . . . . . Isobutylene. . . . . . . . . . . . . . . . . . Section E: 2-Bromo-2-butene . . . . . . . . . . . . . Trimethylethylene . . . . . . . . . . . . Styrene. . . . . . . . . . . . . . . . . . . . . . 1 NORMAL ADDITION 25 79 59 95 92 70 70 100 55 850 1632 336 240 27 40 3 1 2 :ondi- tionst ABNORMAL ADDITION Yield' er ceni 27 84 21 70 76 100 80 > 86 41 Time hours 24 3 1 31 48 : 16 5 7 0.2i % Faster % 4 __ 2ondi- tionst * When a 90 to 100 per cent yield was obtained, the time given represents only an upper limit to the period required for substantially complete reaction. t Conditions: (1) air absent, antioxidant present; (2) air present; (3) peroxide present; (4) dilute pentane solution with a peroxide present (otherwise no solvent was employed). All reactions were carried out near room temperature, usually in sealed tubes. 3 Reaction appeared to occur as rapidly as hydrogen bromide was added. be retarded by the use of a solvent in order t o obtain the abnormal product. Even so, about 20 per cent normal addition occurs. These observations are all consistent with the view that the two types of addition are independent but competing; the structure of the alkene and the conditions of addition determine the product which predominates. This view is supported by the fact that each alkene in sections B and C THE PEROXIDE EFFECT 367 has yielded many different mixtures of the two possible products a t rates intermediate between those of the independent reactions. Similar ex- amples will be indicated in the section on the influence of experimental conditions.8 Evidence showing how the rate of abnormal addition varies with the structure of the alkene is unsatisfactory, but indications are that the al- kenes in section A, particularly trichloroethylene, are also distinguished by slow abnormal additions. 3. Catalysts of abnormal addition Of the materials which catalyze the abnormal addition, the first ones to be detected, and the most important on account of their wide prevalence, are oxygen and peroxides. Most alkenes, on standing, react with air t o form materials which give the familiar test for peroxides; that is, they give an intense red color when shaken with an aqueous solution containing ferrous and thiocyanate ions.Y The ability of these natural peroxides to cause abnormal addition depends on the extent of their formation and on the rate of the normal addition. For allyl bromide, where the most data are available and where the rate of the normal addition is moderate, the quantities of oxygen or peroxide required are small. Urushibara and Takebayashi (154) state that 1.5 cc. of oxygen per 24.2 g. of allyl bromide (0.03 mole per cent) is sufficient to give a product containing 96 per cent of 1 , 3-dibromopropane (abnormal). Because about three times as much 8 Such observations disagree with the opinion of Urushibara and Takebayashi (151) that the normal and abnormal additions so cancel each other that only one can proceed a t a time. This view is based on a chain mechanism for normal addition which the reviewers have questioned in section 11, A, 3. Assumptions about the effect of one mechanism on the other seem unnecessary, because any rate of reaction or any addition product thus far recorded may be accounted for on the following basis: The abnormal addition may have an induction period (140), proceed very rapidly for a time through the action of peroxides, and then more slowly through the effect of oxygen. 9 Urushibara and Takebayashi (153) state that when oxygen was passed through allyl bromide in the dark, no peroxides were formed in one month, whereas a strong test could be obtained after the bromide was kept for a few hours in diffused light. However, Urushibara and Sinamura (151) record that when oxygen and hydrogen bromide together were passed through allyl bromide in the dark, considerable oxida- tion took place, as indicated by formation of water, the liberation of heat, the temporary formation of bromine, and the development of a peroxide test. A good yield of 1,3-dibromopropane was obtained. Such experiments have never been carried out in this laboratory, but i t has been repeatedly observed that, when allyl bromide is stored in the dark in a bottle containing air, and only intermittently exposed to light for the purpose of withdrawing samples, a weak peroxide test cau be obtained after a few days and a very strong test after a few weeks. The 1- and 2-chloro- and bromo-propenes, as well as styrene, form peroxides more rapidly. 368 FRANK R. MAY0 AND CHEVES WALLING peroxide (formed spontaneously and determined by its ability to oxidize iodide ion) was required to produce a similar effect, they suggest that molecular oxygen, rather than organic peroxides, is responsible for the peroxide effect, and that peroxides function by liberating this gas. The reviewers know of no observations that alkene peroxides liberate oxygen. Added peroxides are quite as effective as those formed spontaneously in the alkene. Among those which have been successfully used are ben- zoyl peroxide, lauroyl peroxide, perbenzoic acid, and ascaridole (natural menthene peroxide). The quantities employed have usually been of the order of 1 mole per cent. Smith has observed the rate of addition of hydrogen bromide to unde- cenoic acid in various solvents and under various atmospheres. He found that there is an induction period in the abnormal addition when air or benzoyl peroxide is the catalyst, but not when perbenzoic acid, which liber- ates bromine immediately, is used. In experiments with purified mate- rials in fairly concentrated solution, air, but not perbenzoic acid, gave an abnormal addition reaction. Smith concludes that oxygen is essential for the abnormal reaction and that peroxides serve as subsidiary catalysts (35, 140). Observations in this laboratory (53, 55, 61, 63, 70) show that both benzoyl peroxide and ascaridole can serve as catalysts for the abnor- mal reaction in the absence of air, but that the rate and extent of abnormal addition are often more unpredictable with peroxides'*, particularly perbenzoic acid, than with oxygen. A simple explanation of these results, as well as those of Urushibara, Takebayashi, and Smith, is as follows: Either oxygen or peroxides can serve as catalyst for the abnormal addition reaction by reacting with hydrogen bromide. If the peroxide reacts rapidly, as do ascaridole and perbenzoic acid, the catalyst is quickly destroyed before much abnormal addition has taken place. On the other hand, benzoyl peroxide" and oxy- gen react very slowly with hydrogen bromide under the usual experimental conditions, and thus exert a catalytic effect over a comparatively long period. The observations of other workers that peroxides may be less effective than oxygen in catalyzing abnormal addition are not questioned ; nevertheless their conclusion that molecular oxygen is necessary seems to the reviewers to be unjustified. Other gases than oxygen, and other oxidizing agents than peroxides, have been investigated for a possible effect on the abnormal addition. Nitrogen, hydrogen, nitric oxide, and nitrogen dioxide were tried with 1 0 With trichloroethylene (74) benzoyl peroxide caused abnormal addition when 11 This statement applies particularly t o experiments without solvents, when the air did not. benzoyl peroxide dissolves slowly. THE PEROXIDE EFFECT 369 allyl bromide and found to be without effect (53). Neither lead dioxide nor bromine (in the dark and in the absence of air) had any effect on the addition to allyl bromide (53), nor did N-bromobenzamide and iodine af- fect addition to propene (111). Smith, however, working with diffused light and in a hydrogen atmosphere, found that intermittent additions of bromine caused an abnormal addition to undecenoic acid in carbon tetrachloride solution (140). Harris and Smith (35) report that a-hep- tenylheptaldehyde and 10,ll-epoxyundecanoic acid are catalysts for the abnormal addition of hydrogen bromide to undecenoic acid in the presence of air, but have only a small effect in its absence. In the faster addition of hydrogen bromide to propene (without a solvent), both propylene oxide and propionaldehyde, when peroxide-free, were ineffective even in the presence of air (111). Therefore such catalysts probably function because they, or impurities which they contain, are oxidized by air t o peroxides.12 Finely divided iron, cobalt, and nickel are strong catalysts for the ab- normal addition of hydrogen bromide to allyl bromide and undecenoic acid; these reactions will be considered later. 4. Inhibitors of abnormal uddi t ion It has been found that small proportions, usually 1 to 5 mole per cent, of certain substances prevent abnormal addition when this reaction would otherwise predominate. These inhibitors are commonly called antioxi- dants because they overcome the effect of oxygen, but many of them are not inhibitors of autooxidation reactions. Table 6 summarizes some of the available information on inhibitors of abnormal addition, and permits rough evaluation of their effectiveness. The alkenes in section A are those which put antioxidants t o the most severe test. For these com- pounds no antioxidant has been found which completely prevents abnormal addition, except in the presence of a catalyst for the normal reaction. I n section B, the inhibitors are eflective for propene only when this com- pound is carefully and freshly purified and when a clean vacuum line is used. The other sections bring out further differences in the effectiveness of antioxidants, and it can be concluded that compounds with sulfhydryl groups are the best inhibitors; some phenols and aromatic amines are less effective. It should be noted that the effectiveness of the inhibiting sub- stances is related t o their solubility. This fact may account for the moder- ate efficacy of diphenylamine (the hydrobromide of which is rather soluble l2 Yote added August 8, 19.10: This conclusion is supported by the paper of M. Talrebayashi (Bull. Chem. SOC. Japan 16, 116 (1940)). He found that the effect of aldehydes on the addition of hydrogen bromide to undecenoic acid depended on the peroxide contcnt of the aldehydes. 370 FRANK R. MAY0 AND CHEVES WALLING I- + THE PEROXIDE EFFECT 37 1 3’72 FRANK R. MAY0 AND CHEVES WdLLING in the reaction mixtures), for the fact that catechol is often better than hydroquinone, and for the differences between some of the metal halides. Ethanol has been found to inhibit abnormal addition to AI0-undecenoic acid (35) and to trimethylethylene (162). In a later section i t will be shown that this inhibiting effect decreases a t lower temperatures. 5 . The bromine-atom mechanism of abnormal addition At this point i t seems desirable to correlate the data so far presented by means of a mechanism for the abnormal addition. Since this reaction can be caused by relatively small quantities of oxygen and peroxides and inhibited by equally small quantities of antioxidants, it is established that the abnormal addition is a chain reaction. In choosing a mechanism for the reaction, i t must be explained how oxygen and peroxides start reaction chains, how the chains give the abnormal addition product, and why hydro- gen bromide is the only halogen acid capable of reacting through such a chain. In order to meet the last requirement, it seems necessary that the oxygen or peroxides function through the hydrogen bromide rather than through the alkene; otherwise it would be difficult to explain why the addition of other unsymmetrical reagents is not also reversed by these reagents. If oxygen or peroxides attack hydrogen bromide, bromine should result; but molecular bromine has no effect on the reaction in the dark. It has therefore been suggested that the slow oxidation of hydrogen bromide in dilute solution gives bromine atoms. In adding to a double bond, these atoms need not attack the same doubly bound carbon atom which would be attacked by the proton in the normal addition. Although oxidation of hydrogen bromide might conceivably give positive bromine ions, and although these ions should add abnormally to double bonds, for- mulation of the mechanism of the abnormal addition in terms of bromine atoms (’70)13 is preferred, because more energy would be required to sepa- rate charged particles in non-polar solvents. According to this mechanism, the reaction should proceed as follows: HBr + O2 ( 5 ) (6) (7) (a1 kene) +H-O--O. + Bra RCH=CH2 + Br* + RCHCH2Br RCHCHzBr + HBr + RCH2CH2Br + Br* The individual steps suggested will next be discussed. 13 The suggestion that the abnormal addition is a chain mechanism involving bromine atoms was made simultaneously and independently, but without details, by Hey and Waters (37). Previously, Burkhardt (16) had suggested that free radicals might account for the abnormal addition of thiophenol to styrene; cf. section 111, B. THE PEROXIDE EFFECT 373 The oxidation of hydrogen bromide by oxygen is ordinarily slow, but Urushibara and Sinamura have shown that in the presence of an alkene such as allyl bromide, peroxides are very rapidly formed (151). These organic peroxides react slowly with hydrogen bromide. The same authors also showed that the interaction of hydrogen bromide, oxygen, and stilbene gives stilbene dibromide, thus proving the oxidation of hydrogen bromide. Similarly, the abnormal addition of hydrogen bromide to allyl bromide is accompanied by the formation of small amounts of a high-boiling liquid which is apparently 1 ,2,3-tribromopropane (94, 151). The essential feature of reaction 5 is that a t least a small portion of the bromine formed must be liberated in the atomic state. This requirement is consistent with the fact that those peroxides which react more slowly with hydrogen bromide are usually more effective; i t also agrees with the observation of Smith (140) that bromine is a catalyst for the addition of hydrogen bromide in diffused light. It may react with hydrogen bromide to give another bromine atom and hydrogen peroxide, or it may combine with alkene. In the nor- mal addition to the type of alkene chosen as an example, the proton be- comes attached to the terminal carbon atom, showing that this atom is the relatively negative carbon atom of the double bond. In abnormal addition the bromine atom becomes attached to this terminal carbon atom. It has already been suggested (70) that the oxidizing properties of the bromine atom cause it to attack the carbon atom with the higher electron density. This explanation is simple and plausible, but the following one has addi- tional advantages: The point of attack by the bromine atom is little af- fected by the polarity of the double bond, but depends upon the relative stability of the two bromoalkyl radicals which may be formed. If the directions of all additions by the chain mechanism are to be explained on this basis, the following orders of decreasing stabilities of free radicals (each formed by the addition of a bromine atom to a double bond) are required : The fate of the HOP radical is unimportant. If reaction 5 is assumed, reaction 6 can follow very easily. Radicals from hydrocarbons : tertiary > secondary > primary H H H H I I I I Radicals from vinyl-type halides : R-C-C-X > R-C-C-X * I I * Radicals from acids, esters: Br Br H H H H I I I I . I I . R-C-C-COOR’ > R-C-C-COOR’ I3r Br 374 FRANK R. MAY0 AND CHEVES WALLING Radical stability in this discussion is intended in the sense of higher heat of formation; no reference to the mean life of the radical is implied. If these relative stabilities can be proved correct, then the hypothesis of radical stability explains why addition of hydrogen bromide by the abnor- mal rather than the normal mechanism gives complete reversal of addition with hydrocarbons, only partial reversal with 1-bromopropene (to give exclusively 1 , 2-dibromopropane), and no change when the ethylene bond is conjugated with the carboxyl group. This hypothesis suggests that the addition of hydrogen bromide to such conjugated double bonds may take place through two mechanisms to give only one product. Addition by the normal mechanism must be fast as compared with addition by the chain mechanism, in order to agree with the following qualitative rate indications. The rate of addition of hydrogen bromide to crotonic acid and its ethyl ester is unaffected by peroxides, even when the normal addition is retarded by the use of an inert solvent (162). These conclusions as t o the relative rates of addition by the two mechanisms are in agreement with experiments (to be described in section V, C) on the relative rates of isomerization of maleic acid to fumaric acid by corresponding mechanisms. Smith (141) has suggested that since ethylene bonds conjugated with carboxyl groups should be represented by the formula and that since there is no “accumulation” of electrons on the a-carbon atom, there is no tendency for the bromine atom to attack this position or for abnormal addition to occur. This explanation is in good agreement with the qualitative observations on the rate of addition of hydrogen bromide, but not with the isomerization studies mentioned. In the opin- ion of the writers, no explanation yet proposed for the absence of abnormal addition to conjugated systems is wholly satisfactory. In order to obtain the addition product from the free radical formed in reaction 6, reaction 7 is necessary. This latter step regenerates a bromine atom, so that the chain reactions (6 and 7) can continue indefinitely. The inhibition of the abnormal reaction by small quantities of antioxidants is excellent evidence that this reaction is of the chain type. The small amounts of antioxidants usually needed to inhibit and the small amounts of peroxides necessary to cause abnormal addition indicate that the chains must be very long. If 0.03 mole per cent of oxygen causes nearly com- plete abnormal addition to allyl bromide (154), and if each molecule of THE PEROXIDE EFFECT 375 8TW IN CHAIN BDACTION ( 6 ) RCH=CH2 + F+ RCHCHzX., . . . . . . . (7) RYHCH2X + H X -+ RCHzCHzX + X'. oxygen generates four bromine atoms by oxidizing hydrogen bromide, then the average chain length is a t least 1000. If the efficiency in generating bromine atoms is low, as seems likely, then the chains must be much longer. These chain lengths, together with the observed high velocity of the ab- normal addition and the known instability of aliphatic free radicals, in- dicate that both steps in the chain reaction must take place very rapidly and with little or no activation energy. The inhibition of the reaction by antioxidants suggests that these function by reacting with bromine atoms. It is also possible, as originally suggested (53), that antioxidants to some extent destroy organic peroxides. Other observations on the abnormal addition are consistent with the chain mechanism suggested. The ab- normal reaction is retarded by glass wool (139), although apparently not by coarser packing (155). When non-polar solvents are used, the absence AH (IN PILOQRAM-CALORIEE PEE MOLD) X = F X - C 1 X = B r 1 X = I -60 -15 0 TABLE 7 Heats of addit ion of halogen acids to double bonds* * When two values are given for AH, the first is based on the bond-energy estimates of Sherman and Ewe11 (31), the second on the estimates of Pauling (124). One value for AH indicates that both sources give the same result. of a large dilution effect (113) shows that the abnormal reaction cannot be of second or third order. It will now be indicated why hydrogen bromide is the only halogen acid capable of giving an abnormal addition. The exposition assumes that a rapid chain reaction is impossible when any step is appreciably endother- mic. Table 7 gives estimates of the heats evolved in reactions 6 and 7 for various halogen acids; it is assumed that AH for each reaction is simply the difference between the energies of the bonds formed and the energies of the bonds broken. The heats indicated are only approximate, for the bond energies are none too well established for the molecules to which they are meant to apply and are still less reliable when applied to free radical^.'^ Table 7 shows that reaction 6 for hydrogen bromide is definitely exothermic and that the heat of reaction 7 is very close to zero. However, the addition l4 Actually, the activation energy rather than the AH will determine the prob- Because long chains occur in some additions, i t is assumed here ability of reaction. that the activation energy is small when AH is positive or zero. 376 FRANK R. MAY0 AND CHEVEB WALLINQ of hydrogen chloride by the same mechanism encounters difficulty in two places, reactions 5 and 7. Since the oxidation of hydrogen bromide is ordinarily slow (reaction 5), the oxidation of hydrogen chloride is probably slower, possibly negligible. In reaction 7 the carbon-chlorine bond formed is weaker than the hydrogen-chlorine bond broken ; the reaction is endo- thermic by about 15 kg-cal. per mole, and unlikely to occur rapidly enough to support long chains a t ordinary temperatures. Both of these difficulties apply also to hydrogen fluoride and to most other acids. Attempts to reverse the direction of addition of sulfuric acid to propene, 1-pentene, and 2-pentene have failed (14). In the addition of hydrogen iodide, reactions 5 and 7 should proceed easily, but reaction 6 may not. Another difficulty with the addition of hydrogen iodide is that the normal addition is ordinarily rapid and is further catalyzed by molecular iodine. Any reagent which generates iodine atoms must necessarily generate also the catalyst for the normal addition. Still another possibility is that iodine, since it does not add readily to alkenes, may accumulate in the reaction mixture and inhibit any possible abnormal addition of hydrogen iodide by an iodine-atom mech- anism. According to the chain mechanism proposed, the unsymmetrical reagents, HX, which can add by a chain mechanism are restricted to narrow limits. The requisites are as follows: the radical (or atom) X must be able to break a carbon-carbon double bond and add to one carbon atom; the free radical formed must be able to take a hydrogen atom away from H X to regenerate another X radical. It will be shown later that this mechanism can also be applied to additions of mercaptans and bisulfites. The next abnormal additions to be considered are those occurring when air and peroxides have been excluded from reaction mixtures (135,136) and when antioxidants have been added (25, 70, 117). The discussion of anti- oxidants showed that these substances exert effects ranging from complete to barely perceptible repression of the abnormal addition. Since different antioxidants have variable effects under the same experimental conditions, the occasional failure of the best antioxidants and the more frequent failure of inferior inhibitors indicate only that the proportions used or the inhibit- ing qualities of the compounds were inadequate; they do not indicate a difference in the nature of the abnormal reaction. It is thus proved for the weaker antioxidants, and inferred for the rest, that only a small proportion of the collisions of a bromine atom with an antioxidant molecule are effec- tive in terminating chains. The smaller this proportion, the greater must be the chain lengths and the fewer the number of bromine atoms necessary to cause abnormal addition. Since there is no basis for estimating an upper limit on chain lengths, i t can be argued that traces of peroxides and oxygen which would defy elimination or detection by any known method THE PEROXIDE EFFECT 377 may be wholly responsible for all abnormal addition reactions. On the other hand, infinitesimal quantities of atoms or radicals may appear spon- taneously in the solution with the result that chains are initiated. The dissociation of hydrogen bromide into atoms would require 87 kg-cal. per mole (31, 124), but the dissociation of a carbon-bromine bond in an alkyl halide requires on the average only 50 t o 60 kg-cal. Such a dissociation might initiate two chains if the free radical reacted with hydrogen bromide to give an alkene and a bromine atom. Sherman, Quimby, and Sutherland (134)16 have utilized such a chain-initiating step in calculating the activa- tion energy of the addition of hydrogen bromide to vinyl bromide by the chain mechanism just described, and have arrived a t a value of 29 kg-cal., one-half the estimated strength of the carbon-bromine bond. If this value is approximately correct, the abnormal addition can be initiated without the assistance of oxygen or peroxides. Such an activation energy, how- ever, is probably higher than that for most normal addition reactions; hence appreciable abnormal addition of spontaneous origin could be expected only when the normal addition is very slow. Although oxygen and perox- ides are usually responsible for the abnormal addition of hydrogen bromide to alkenes, i t is futile a t present either to assert or to deny that they are always entirely responsible. The significant aspects of the peroxide effect in additions of halogen acids are that abnormal addition takes place only with hydrogen bromide, and then by a chain mechanism, and that oxygen and peroxides are largely, if not entirely, responsible for this reaction. 6. Other mechanisms proposed for abnormal addi t ion The first mechanism to be proposed for abnormal addition was that of Urushibara and Takebayashi (155). Because iron and certain other metals also caused abnormal addition, they suggested that the effect of oxygen was due to its paramagnetic properties. The oxygen or metal was supposed to exert a physical influence on the surrounding alkene molec- ules such that the polarity of the double bond was affected. The short- comings of this hypothesis have been mentioned elsewhere (87), and are admitted by a t least one of the above workers (151), who now favors the bromine-atom chain mechanism. Winstein and Lucas (165) have proposed an explanation for abnormal addition t o a double bond the polarity of which (as indicated by the nor- mal addition) is -C-C-. They suggested that oxygen forms with 16 These workers proposed the chain mechanism for both the normal and the abnormal additions of all halogen acids to alkenes a year before investigators in this laboratory suggested i t only for the abnormal addition of hydrogen bromide. I / + - 378 FRANK R. MAY0 AND CHEVES WALLING double bonds of this type a complex represented by the following resonance forms: I / -c-c- - I O+ I 1 + I -c-c- 0 0- II 0 If the second form makes the larger contribution to the complex, the pro- ton of the attacking halogen acid attaches itself to the carbon atom which would have been relatively positive in the pure alkene. The oxygen is then displaced by a bromide ion, and abnormal addition results. Conn, Kistiakowsky, and Smith (20) concur essentially in this suggestion. How- ever, such behavior of an oxygen molecule in forming successive loose complexes with several hundred molecules of alkene is not wholly consist- ent with the rapid irreversible reaction of oxygen on alkenes in the presence of hydrogen bromide (151). Michael (114, 117) has recently attacked all previous mechanisms for abnormal addition and has suggested one of his own. On the basis that oxygen in the atmosphere or in the molecules of the solvents is negative, he proposes that oxygen from either source may become associated with the more positive of the doubly bound carbon atoms, thus making this atom relatively negative. If the effect of oxygen (or oxygen compound) is sufficient to reverse the polarity of the double bond, then the addition of halogen acid is also reversed. All of these hypotheses may be said to account for an abnormal addition of hydrogen bromide, but all fail t o indicate why hydrogen chloride and hydrogen iodide do not give a similar reaction. Further, all of them except that of Urushibara and Takebayashi fail t o explain why some finely divided metals cause abnormal addition. This latter effect will soon be considered in the light of the chain mechanism. 7. T h e in$uence of experimental conditions o n additions of hydrogen bromide The following discussion of the effects of solvents, temperature, light, and metals on the liquid-phase addition of hydrogen bromide to alkenes will assume that all of the abnormal product results from addition of hydro- gen bromide by the bromine-atom chain mechanism, whereas all of the normal product results from addition by other mechanisms. The reaction product obtained is the result of competition between the two mechanisms; hence its composition may vary from one extreme to the other. Experi- mental conditions may favor one mechanism and hinder the other, but the fact that the direction of addition of hydrogen chloride and hydrogen iodide has not yet been certainly (44, 84) altered by any agent indicates THE PEROXIDE EFFECT 379 that these conditions affect the halogen acid or its mechanism of addition, and not the unsaturated compound. Solvents principally affect the rate of the normal addition reaction. Because of the high order of this reaction, its rate is greatly decreased by dilution with inert solvents, and the abnormal addition may then outrun the normal reaction. This effect is strikingly demonstrated in the addition of hydrogen bromide to propene. When no solvent is used, the normal addition requires only a few minutes and abnormal addition is difficult t o obtain, except in the presence of added peroxides (54). But when the reaction mixture is diluted with about ten volumes of pentane, abnormal addition is substantially completeI6 within half an hour a t O'C., even when the reaction mixture is prepared from purified materials in the absence of air (113). However, the addition of an antioxidant prevents abnormal addition; then several weeks are required for the normal reaction. Smith has shown that, a t high dilution, the addition to undecenoic acid is highly sensitive to the presence of air (140). Both observations show clearly how dilution affects the competition between the two mechanisms, why past claims that a direct solvent effect is responsible for abnormal addition are open to serious doubt, and how a peroxide effect may be obtained with alkenes to which normal addition is rapid. Small amounts of polar solvents such as water and acetic acid may increase the rate of the normal addition, probably because of dielectric effects and the change from a molecular t o an ionic mechanism. This acceleration is partly responsible for the small amount of abnormal addi- tion in polar solvents. Some solvents may, however, be weak inhibitors of abnormal addition, as will be indicated in the discussions of the tem- perature and light effects. Table 8 summarizes work on the addition of hydrogen bromide to alkenes in solvents. All the experiments listed fulfill the following conditions: (a) they have been carried out since the discovery of the peroxide effect; ( b ) both products have been obtained in the same solvent by the same workers; (c) either addition product can be made to predominate by the suitable use of oxygen, peroxides, or antioxidants. The facts that a large variety of alkenes and solvents have been employed, and that many dif- ferent workers have participated in obtaining these results, are convincing evidence that the mechanism of addition is the factor which determines the composition of the addition product. l6 Such a velocity would lead one to expect considerable abnormal addition in the absence of a solvent and of air, but no such reaction has actually been observed. If some association product of hydrogen bromide or propene or both is a weak in- hibitor of abnormal addition, then this reaction, in agreement with qualitative in- dications, would be accelerated by dilution. 380 FRANK R. MAY0 AND CHEVES WALLING Some workers in the field disagree with the above conclusion, and main- tain that solvents immediately affect the direction of addition, presumably through their effect on the polarity of the double bond. Gaubert, Lin- stead, and Rydon base their contention on the inability of diphenylamine to prevent formation of primary bromides from terminally unsaturated acids in hexane solution (25). Sherrill and coworkers (135, 136) insist that their 1-alkenes were peroxide-free and that the formation of primary Solvents an which ALKENE Allyl bromide. . . . . . . . . . Vinyl bromide, . . . . . . . . Vinyl chloride.. . . . . . . . . Propene . . . . . . . . . . . . . . . 1-Butene . . . . . . . . . . . . . . . I-Pentene . . . . . . . . . . . . . . Isobutylene. . . . . . . . . . . . Trimethylethylene Styrene. . . . . . . . . . . . . . . . 2-Bromo-2-butene.. . . . . 2-Bromo-1-hexene.. . . . . Allylacetic acid.. . . . . . . AlO-Undecenoic acid. . . Ethyl AIo-undecenoate . Undecenyl acetate. . . . . Undecynoic acid.. . . . . . TABLE 8 drogen bromide yields either of two addi t ion products 80LVENTS E W L O Y E D Water, 90% acetic acid, propionic acid, ligroin, carbon disulfide, chloroform, carboii tetrachloride, 1,2-dibromopropane, 1,3-dibromopropane, acetyl bromide* Acetic acid, nitrobenzene Acetic acid, nitrobenzene, mesitylene Pentane, acetic acid Ligroin Pentane, propionic acid Pentane, carbon disulfide, propionic acid, ethyl bromide, benzonitrile, nitrobenzene, water Pentane, ethyl bromide, ether, methanol, ethanol, acetic acid* Pentane Acetic acid Benzene Hexane Ligroin, hexane, toluene, benzene, carbon disulfide, carbon tetrachloride, chloroform Benzene, ligroin Benzene, ligroin Benzene REFERENCES * In the experiments indicated in these two solvents, less than 50 per cent ab- normal addition in the presence of air or peroxides was observed. bromides in solvents must therefore be a solvent effect. They used no antioxidants. Michael and coworkers claim to have demonstrated a solvent effect in the addition of hydrogen bromide to tetrolic and phenyl- propiolic acids (table 3) and to trimethylethylene (table 8). In the first instance, they employed no antioxidants; in the second, the “solvent effect” was diminished by antioxidants and vanished in experiments with hydrogen chloride (table 1) or hydrogen iodide (table 2). The reviewers question not the experimental results of these workers, but their conclusions. THE PEROXIDE EFFECT 381 An effect of the solvent on the polarity of the double bond, and thus on the direction of normal addition, would be most likely in instances where the normal addition product is a mixture, and where the directing effects of the substituents on the doubly bound carbon atoms are nearly balanced. No such effect has yet been reported with this type of ethylene derivative. The addition of hydrogen bromide to 2-pentene has been carried out without a solvent and in acetic acid solution (77). With A9-undecenoic acid the reaction has been run in ligroin, hexane, benzene, and acetic acid (2,34). In all cases, the addition product consisted, within experimental error, of equal proportions of the two possible halides. The fact that no solvent effect on the direction of normal addition has yet been established does not prove that none will ever be found, but the existence of such an effect should be demonstrated, not with hydrogen bromide (which is the only halogen acid capable of addition by an abnormal mecha- nism) but with hydrogen chloride or hydrogen iodide. Increasing the reaction temperature generally increases the difficulty of eliminating abnormal addition. Air does not cause much abnormal addition to pure allyl bromide a t O"C., but i t does a t room temperature (53). Removal of air prevents abnormal addition a t room temperature, but this procedure becomes ineffective a t 76°C. However, the addition of an anti- oxidant reduces the proportion of abnormal addition product to about 10 per cent, even a t lOO"C., indicating that the effect of temperature is on the relative rates of the normal and abnormal addition reactions, and not on the allyl bromide (see footnote 3). Vinyl bromide (54) and vinyl chloride (57), to which the normal additions are slower, present similar difficulties beginning a t lower temperatures, for antioxidants are always required to eliminate abnormal addition to these substances at room temperature, and partial failure of antioxidants has been observed a t 46" and 76"C., respectively. Abnormal addition to allyl bromide (53) and trimethylethylene (117) has been observed a t -78°C. with small quantities of peroxides, showing that long chains are formed and that their propagation requires a negligible activation energy. Differences in the effectiveness of different anti- oxidants indicate that a t least some, and probably all of them, require a t least a small activation energy for reaction with'bromine atoms. It follows that the proportion of effective collisions between a bromine atom and an antioxidant must increase as the temperature rises.I7 Therefore, the increasing predominance of abnormal addition a t high temperatures in bhe presence of antioxidants cannot be ascribed to increased chain lengths, but must be due t o the origin of many more chains. Although the l7 This phenomenon has been observed when alcohols, acetic acid, or acet3-1 bromide have been used as solvents; it will be described shortly. 382 FRANK R. MAY0 AND CHEVES WALLING decomposition of peroxides or their reaction with hydrogen bromide may have a high temperature coefficient, either reaction should lead to rapid exhaustion of traces of peroxides in experiments from which air has been excluded. Here the spontaneous origin of chains (i.e., without the influ- ence of oxygen and peroxides), a reaction of high activation energy and high temperature coefficient, may become significant. The normal addi- tion of hydrogen bromide to allyl bromide and vinyl bromide has a tem- perature coefficient of about 2 for 10°C. That of the abnormal addition is much larger, a t least over certain ranges of temperature. The temperature effect in solvents fits into the scheme already outlined. If the solvent (ligroin, chloroform, carbon tetrachloride with allyl bromide (53), ether with trimethylethylene (117)) decreases the rate of normal reaction, then the abnormal addition becomes more prominent and appears a t a lower temperature. If the solvent (acetic acid, water, acetyl bromide with allyl bromide) increases the rate of the normal reaction, then abnormal addition becomes less prominent. The fact that abnormal addition is easily obtained in many solvents indicates that their inhibiting properties are usually negligible. Never- theless, almost any of the aliphatic solvents in table 8 can react with bromine atoms, as shown by the mechanism of aliphatic substitution. In the solvents which inhibit, the effect increases with increasing tempera- ture. In the presence of peroxides, no abnormal addition to trimethyl- ethylene takes place in ethanol solution a t 20°C. or in methanol a t O"C., but the proportion of abnormal addition increases a t lower temperatures (117). Acetic acid and acetyl bromide seem to have definite inhibiting properties in additions to allyl bromide a t 76°C.; moreover, in the light, acetic acid seems to be a better inhibitor of abnormal addition a t 25°C. than a t 10°C. Ordinarily, little difficulty is encountered in obtaining abnormal addition in glacial acetic acid (cf. table S), but experiments with allyl bromide in this solvent show that small amounts of admixed water favor abnormal addition. Light accelerates both the normal and the abnormal additions of hydro- gen bromide to allyl bromide (53) and vinyl bromide (54), but the effect on the abnormal addition is usually far greater than on the normal reaction. The rate of addition to allyl bromide in the light is greatest in the presence of air, slower in the absence of air, and still slower in the presence of added antioxidants. In the first two instances, the abnormal addition product is formed exclusively; in the last, predominantly. These observations are corroborated by others made on vinyl chloride (57) and trichloro- ethylene (74). In additions to allyl bromide in the light, the effects of various spectral regions and of solvents have been investigated. In the absence of air THE PEROXIDE EFFECT 383 and antioxidants, abnormal addition was substantially complete in 1 to 4 hr., regardless of whether the light supplied was near ultraviolet, red and infrared together, or a combination of either with visible light.ls In the presence of 1.4 mole per cent of diphenylamine, a combination of red and infrared light was clearly shown to accelerate the normal addition by a factor of 5 to 10 without causing much abnormal addition. Visible and near ultraviolet radiation gave increasing proportions (up to 74 per cent) of abnormal addition product a t about the same total reaction rate. Thus the shorter wave lengths make repression of abnormal addition more difficult. In the absence of air and antioxidants, abnormal addition occurs exclusively when heptane, carbon disulfide, acetyl bromide, or benzoyl chloride are used as solvents. In acetic acid solution, only very small proportions of abnormal addition product are obtained; none of this product is formed in the presence of hydroquinone. Comparison of these experiments with those carried out in the absence of a solvent indicates that acetic acid has some inhibiting properties for the abnormal addition, and that the effect of light on the competition between the normal and abnormal reactions is much like that of increased temperature. Inasmuch as the individual effects of light and temperature on the rates of the normal and abnormal additions are known only qualitatively, and since the effects of temperature on the inhibiting properties of solvents and antioxidants have not been isolated, the combined effects of light and temperature cannot yet be completely resolved. The effect of lowering the reaction temperature in the photochemical addition of hydrogen bro- mide to vinyl bromide (54) is impressive. Although no normal addition occurs a t room temperature in the presence of antioxidants, a 50 to 60 per cent yield of an 80 per cent normal product was obtained in 16 hr. a t 5°C. Thus not only was the proportion of abnormal addition greatly reduced by lowering the temperature, but the normal addition product was formed a t the rate of about 3 per cent per hour, as against 1 per cent per day in the dark a t room temperature,-an increase by a factor of about 70 in spite of a 20°C. temperature decrease. The case of allyl bromide is less clear-cut. The effect of light in accelerating the normal addition is smaller, and lowering the reaction temperature in the presence of antioxidants does not significantly increase the proportion of normal addition product.lg The available data do not permit estimation of the 1s Comparisons of rates are not possible, because the amount of light transmitted by the filters is not known. 19 Kharasch and Mayo (53), using thiocresol as an antioxidant, observed 71 per cent and 30 per cent normal addition a t 5°C. and a t room temperature, respectively, but in view of the fact that the combined yields of both addition products were 54 per c-nt and 91 per cent, respectively, and that normal addition is fastest during 384 FRANK R. MAY0 AND CHEVES WALLING effect of light on the rates of the two possible additions in glacial acetic acid solution, but they show conclusively that in the light more abnormal addition takes place a t 5°C. than at room temperature. This fact can be explained by the assumption that acetic acid is less effective as an inhibitor a t lower temperatures. Such a conclusion is consistent with observations made a t higher temperatures or in the presence of alcohols. It may be noted that a t 5°C. a reaction mixture containing diphenylamine reacted more slowly than one containing oxygen and gave slightly less abnormal addition product,-a fact which may be taken to indicate that the effect of light in accelerating the abnormal addition is large compared with that of oxygen. The acceleration of both the normal and abnormal addition by light, the increased light effect in the presence of oxygen, some relations between the combined effects of light and temperature, and the partial and variable effects of antioxidants in repressing abnormal addition in the light all show the close analogy between the effect of light and that of increased temperature. These observations are consistent with the concept that all abnormal addition products result from a reaction by the chain mechanism. Since the effect of light is greatest in the presence of oxygen, illumination apparently accelerates peroxide formation and oxidation of hydrogen bro- mide, but it apparently also serves to initiate chains without the assistance of oxygen. The fact that light accelerates the normal (as well as the abnormal) addition, and that the effect is much greater with vinyl bromide than with allyl bromide, shows that either the alkene, its complex with hydrogen bromide, or some impurity absorbs visible radiation. The greater effectiveness of the shorter wave lengths in accelerating abnormal addition to allyl bromide is consistent with the greater activation energy of this reaction. Although in the absence of air and peroxides, reaction mixtures usually remain colorless in the dark, the presence of oxygen, peroxides, or light usually leads to the development of variable quantities of dark-colored materials. These may assist in the transference of energy to substances which otherwise absorb only weakly. It was found by Urushibara and Takebayashi that, if the addition of hydrogen bromide to allyl bromide takes place in the presence of finely divided and freshly reduced iron, nickel, or cobalt, varying quantities of abnormal product are formed, even in the presence of some antioxidants. the first part of the reaction when the concentrations of reactants are highest, the conclusion that temperature lowering increases the yield of normal addition product is not justified. A recent experiment with catechol at 5-10°C. (112) gave in 65 hr. a 92 per cent yield of a mixture containing only 26 per cent of the normal addition product, again indicating that temperature lowering may have no marked effect. THE PEROXIDE EFFECT 385 They observed similar phenomena when hydrogen bromide was added t o undecenoic acid in toluene solution. Their work has been reviewed elsewhere (151) and that portion of i t concerned with the effect of reduced iron on allyl bromide has been confirmed, extended, and explained in a paper from this laboratory (87). The complete inhibition of the iron- promoted abnormal addition by some antioxidants, and its partial inhibi- tion by others (cf. table 6 ) , are strong indications that abnormal addition is the result of the bromine-atom chain mechanism. The fact that some hydrogen and metal bromide are formed suggests that interaction of the metal with hydrogen bromide or allyl bromide yields some hydrogen atoms or free radicals. Reaction of hydrogen atoms with hydrogen bromide or allyl bromide would yield free radicals or bromine atoms, either of which might initiate chains for the abnormal addition. Fe + HBr -+ FeBr* + H* H* + HBr ---f Hz + Br* Fe + R’Br ---f FeBr* + R’* Thus far, iron-promoted abnormal addition has not been found with any alkene which gives a rapid normal addition with hydrogen bromide, but i t has been observed with allyl chloride and possibly with vinyl bro- mide. Other metals have failed to cause abnormal addition to allyl bro- mide, because they do not react with anhydrous hydrogen bromide, because the bromides formed are strong catalysts for the normal addition (sec- tion I, A, 3), or because their bromides are strong inhibitors for the abnormal reaction (table 6). The success of the bromine-atom chain mechanism in correlating the metal effect with the peroxide effect, an advantage not possessed by any other mechanism yet proposed for either reaction, is a strong point in favor of this interpretation. However, the value of a hypothesis lies in its ability t o predict new reactions as well as t o explain known ones. At the end of this paper, it will be shown how many new applications of this concept have been developed since it was first formulated. In concluding this section dealing with the influence of experimental conditions on the direction of addition, the conditions which should be chosen to promote either normal or abnormal addition of hydrogen bro- mide are summarized. To favor normal addition, this reaction may be accelerated by the use of high concentrations of reactants, particularly hydrogen bromide, or by the use of fairly small proportions of polar solvents (acetic acid), or of catalysts (ferric and aluminum bromides); and the abnormal addition can be inhibited by the use of antioxidants. To favor (8 ) (9) (10) (11) RCH=CHZ + H* * RCHCH3 0 386 FRANK R. MAY0 AND CHEVES WALLING abnormal addition, this reaction may be accelerated by the use of oxygen, peroxides, certain metals (iron, nickel), light, or elevated temperatures; the normal addition (unless it is naturally slow) should be retarded by dilution with inert, non-polar solvents. It should be added that any atom or free radical which can react with the hydrogen atom or the bromine atom of hydrogen bromide or with an alkene may be capable of starting a reaction chain. 8. The addition of hydrogen bromide to cyclopropane Cyclopropane is closely related to propene, in that both react with some acids and oxidizing agents to give propane derivatives. They differ in that cyclopropane gives 1 , 3-derivatives of propane, whereas propene gives 1,2-derivatives. In an investigation of the addition of hydrogen bromide to cyclopropane (78), although the product was always n-propyl bromide, striking analogies with addition to alkenes were found in rates of reaction. When equimolecular mixtures of cyclopropane and hydrogen bromide were allowed to react in sealed tubes a t room temperature in the absence of air and light, 50 to 60 per cent reaction took place in 4 hr. If 3 mole per cent of water or acetic acid was added to the mixture, about 90 per cent reaction took place in the same period, indicating that addition oc- curred largely through a polar molecular or ionic mechanism like that for the normal addition of halogen acids to alkenes. Similar proportions of catechol or thiocresol acted like water or acetic acid, but oxygen or visible light had only a small accelerating effect on the reaction. However, if the reaction mixture was made up of 10 moles of cyclopropane to 1 mole of hydrogen bromide, thus in effect diluting the previous reaction mixtures with 9 moles of hydrocarbon, then characteristics of the abnormal addition of hydrogen bromide to alkenes appeared. In the absence of oxygen and light, only 8 per cent reaction took place in 2 hr. Light alone increased this yield t o 11 to 12 per cent; oxygen alone, to 81 per cent; oxygen and light together, t o 99 per cent; peroxides and light, to 74 per cent. Catechol and diphenylamine had small t o moderate inhibiting effects on the accel- erated reaction. That the effects were not larger is due to the fact that the antioxidants (and also water) accelerated addition by favoring the competing normal mechanism, as in concentrated solution. These phenomena suggest that oxygen and peroxides react with hydrogen bromide to give bromine atoms, or with cyclopropane to give free radicals, and that addition of hydrogen bromide niay then take place through the following chain mechanism: H2C\ H2C/ 1 CH2 + Br. +. CH2BrCH&H2* (12) (13) CH*BrCH&Hz* + HBr +. CH2BrCHtCK3 + B r THE PEROXIDE EFFECT 387 In additions of hydrogen bromide to pure alkenes, light is more powerful than oxygen in promoting abnormal addition, but the reverse is true for cyclopropane. Hence the analogy between alkenes and this compound is not complete. Possibly the chain may require modification to include oxygen (78). 111. THE ADDITION OF MERCAPTANS AND THIO ACIDS TO UNSATURATED COMPOUNDS A. INTRODUCTION A mercaptan, like a halogen acid, when it is added t o an unsymmetrically substituted ethylene bond, can yield either of two products: RCH=CHz + R’SH + RCHZCH2SR’ (abnormal addition) RCH=CHz + R’SH + RCH(CH3)SR’ (normal addition) (14) (15) These two possible reactions are here designated “normal” and “abnor- mal” additions on the assumption that mercaptans should add like halogen acids. Actually, both additions have been recorded, but when the absorb- ing compound is a hydrocarbon, the abnormal addition reaction is the one commonly observed when no catalysts for the normal addition are em- ployed. This fact accounts for the acceptance of the rule prematurely proposed by Posner (125) for the addition of mercaptans to alkenes: namely, that the sulfur becomes attached to the carbon atom holding the most hydrogen atoms. The products and mechanism of the abnormal addition of mercaptans to alkenes will be first discussed; then those of the normal addition reaction. It will be shown that both have much in common with the additions of hydrogen bromide. Finally, the range of applicability of both reactions will be indicated. Although mercaptans usually add abnormally, there is no evidence of such an addition for hydrogen sulfide, which adds only a t fairly high tem- peratures. Under pressure and below 200”C., all the mercaptans and sulfides obtained are normal addition products, and their formation is catalyzed by sulfur (45, 49). The vapor-phase reaction at atmospheric pressure has been studied at 200-300°C. over a nickel-kieselguhr catalyst (9) and a t 300°C. over silica gel (108). In the first instance, 5 to 25 per cent of a mixture containing about 65 per cent of isopropyl mercaptan and 35 per cent of n-propyl mercaptan was obtained from propene. This reac- tion product was considered to be an equilibrium mixture of the two possible addition products (cf. footnote 3). In the second instance, mer- captans, sulfides, and thiophene derivatives of unknown structure were obtained. 388 FRANK R. MAY0 AND CHEVES WALLING B. THE ABNORMAL ADDITION OF MERCAPTANS AND THIO ACIDS It has been clearly shown (6, 16, 49, 76) that the abnormal addition of mercaptans to alkenes is catalyzed by oxygen and peroxides, that it is inhibited by hydroquinone and piperidine, and that it is accelerated by light. The peroxides formed when an alkene is exposed to air are sufficient to catalyze the abnormal addition; careful purification of the reactants and exclusion of air prevents, or greatly retards, any addition. Table 9 summarizes the evidence that the abnormal addition of mercaptans to alkenes can be inhibited by antioxidants, and initiated by oxygen or per- oxides, particularly in the presence of light. It is therefore a chain reaction. The structures of the products formed also supply excellent evidence that the addition does not proceed through a polar or ionic mechanism. Table 10 lists other reactions which show that abnormal addition products are commonly formed when air is not excluded. Table 9 records two instances in which both normal and abnormal addition products are formed, although the latter predominate. In every other addition listed in either table, the abnormal addition product is formed exclusively. The known examples include both aliphatic and aromatic mercaptans, mercaptoacetic acid, and thioacetic acid; the alkenes contain aromatic and aliphatic substituents, but all except trimethylethylene have terminal double bonds. In 1934, Burkhardt (16) mentioned that the abnormal addition of thio- phenol to styrene might be due to the presence of “sulfur in a positive ion or in an oxidizing form,” possibly as a free radical, and possibly through a chain reaction. A more definite proposal that the abnormal addition takes place through a chain reaction involving free radicals was subsequently made from this laboratory (76) : + RS* + HO2. (1.6) (alkene) RSH + 0 2 (or peroxide) RS* + R’CH=CH2 ---f R’CHCH2SR (AH = 13 kg-cal. per mole) (17) R’CHCHZSR + RSH + R’CH2CHtSR + RS* (AH = 0) (18) 0 The effects of oxygen, peroxides, light, and antioxidants and the nature of the addition product so resemble those observed in the abnormal addition of hydrogen bromide that little further comment on these points is neces- sary. In both instances, the heats of reaction of the corresponding steps, as calculated from estimated bond energies (124), are almost identical (cf. table 7). The well-known easy oxidation of mercaptans to disulfides and the suggested dissociation of disulfides into free radicals (132) further support the mechanism suggested. Substantial addition of mercaptoacetic acid to styrene in the absence THE PEROXIDE EFFECT 389 $ a x % B 00- II 390 FRANK R. MAY0 AND CHEVES WALLING THE PEROXIDE EFFECT 391 f CsHs . \ m- It / C-S- CH //* C6H4 / , 2 .C2HsOOC CH I m- II C-S- CaHd / ,C2HsOOC J 2 392 FRANK R. MAY0 AND CHEVE8 WALLING c C r- Y m.- 42 A d U n a P P v d d sd n THE PEROXIDE EFFECT 393 but several analogies to the normal additions of halogen acids can be pointed out: (a) The catalysis of mercaptan addition by sulfuric acid resembles the participation of two molecules of halogen acid in the addi- tion of one such molecule, and suggests that the alkene-acid complex reacts with mercaptan. (b) The catalysis of mercaptan addition by bases recalls the accelerating effect of ammonium salts on halogen acid additions, and suggests that mercaptide ion may sometimes participate in the reac- tion. (c ) Catalysis by sulfur resembles the catalysis of hydrogen iodide addition by iodine. (d) Catalysis by metal sulfides is somewhat analogous to catalysis of halogen acid additions by metal halides. The fact that an alkene is symmetrically substituted does not prevent the addition of a mercaptan from taking place through two distinct mecha- nisms. The addition of ethyl mercaptan to cyclohexene is catalyzed by either peroxides or sulfur, although cyclohexyl ethyl sulfide is the only possible product. Peroxide catalysis is prevented by hydroquinone, with the result that no addition takes place (49). D. SCOPE OF THE MERCAPTAN-ALKENE REACTION The following work is cited to indicate the applicability of the mer- captan-alkene reaction, but, except in additions to ethylene, the structures of the addition products have not been established. Posner (125) found that thiophenol or benzyl mercaptan would add to a large number of solid and liquid hydrocarbons, the only exceptions noted being stilbene and 1,4diphenylbutadiene. He obtained no significant addition to ethylene or propene at ordinary temperatures, but Jones and Reid (49) added ethyl mercaptan and trimethylene dimercaptan to ethylene a t high tem- peratures and pressures in the presence of sulfur. Von Braun and Plate (13) explained the ready polymerization of allyl, crotyl, furfuryl, and cinnamyl mercaptans by the interaction of sulfhydryl groups with double bonds. Holmberg (40) found that mercaptoacetic acid added readily to cinnamyl alcohol, its acetate, and its benzoate. The ready addition of mercaptoacetic acid to many unsaturated compounds and the easy estimation of the sulfhydryl group have led to the use of this reaction in the determination of the degree of unsaturation in oils, fats (7), and gasoline (42). It is stated that peroxides in the latter instance retard the reaction of 2-octene1 a claim not in accord with any other available information. Morgan and Friedman (118) have studied the rate and extent of addi- tion of mercaptoacetic acid, cysteine, and glutathione to maleic acid. They used evacuated reaction tubes and buffered aqueous solutions of sodium salts a t a pH of 7.4 and a temperature of 37°C. Part of the maleic acid which did not react was isomerized to fumaric acid, a point to be 394 FRANK R. M A Y 0 AND CHEVES WALLING considered in a later section. However, none of these mercaptan deriva- tives was found to add to fumaric, citraconic, mesaconic, or a-phenyl- /3-styrylmaleic acids or t o cis- and trans-cinnamic acids. IV. THE REACTION OF BISULFITES WITH UNSATURATED COMPOUNDS A. INTRODUCTION It has long been known that ammonium and alkali-metal bisulfites in aqueous solution add to carbon double bonds, thus forming alkyl sulfonates: R1R2C=CR3R4 + NaHS03 --+ R1R2CHCR3R4(SO&a) (21) Until recently, the best-known of these additions were those to aldehydes, ketones, and acids unsaturated in the a,P-positions (144). Here the products were exclusively those which correspond to the ones obtained by normal addition of halogen acids,-the proton going to the a-carbon atom, the sulfonate group to the P-carbon atom. It has also been shown that bisulfites sometimes add to double bonds in alcohols, in aldehydes, and in liquid and gaseous hydrocarbons where these bonds are not conjugated with carbonyl groups (for references see 75, 144), but the structures of most of the products are unknown. Ethylene and cyclohexene give only one addition product; allyl alcohol was concluded to yield 3-hydroxy- propyl-1-sulfonate ; the structure assigned to the styrene addition product (6 ) was in error (75). Kolker and Lapworth (100) shook their reaction mixtures with kieselguhr in order to maintain contact between hydrocarbon and water solution, and thus obtained addition products from several hydrocarbons. They noted that dilution of the bisulfite solution favored reaction. Other work- ers found that refluxing the reactants was sometimes successful; still others reported that reaction often failed to take place in sealed tubes, even at higher temperatures. Some of these observations suggested that the addi- tion of bisulfite t o unconjugated double bonds may take place through a radical-chain mechanism, and subsequent work agrees with this hypothe- sis. Since the reaction with aliphatic double bonds seems to consist mostly, if not entirely, of simple addition, i t will be discussed first. The more complicated reaction with styrene will be considered later. B. THE ADDITION O F BISULFITES TO ALIPHATIC UNSATURATED COMPOUNDS The work dealing with the effect of oxidizing agents on the addition of bisulfites t o alkenes has been carried out largely in this laboratory; the three different techniques employed will be indicated. Except with allyl alcohol, all the additions of bisulfites were carried out a t room temperature. Bisulfites were added to ethylene, propene, and isobutylene (75, 29) by shaking an aqueous solution of the salts with the gas under pressures THE PEROXIDE EFFECT 395 of 15 to 40 pounds per square inch. Under these conditions the gases in question did not react with bisulfites in the absence of oxygen. Admission of a little air permitted reaction to proceed only temporarily, and repeated intermittent admissions were required to obtain substantially complete reaction of the bisulfite. Based on the amount of bisulfite consumed, propene and isobutylene gave as much as 55 per cent and 62 per cent, respectively, of organic sulfonates, the remainder of the salt being oxidized to bisulfate. Ethylene gave lower yields. The products isolated were exclusively the primary sulfonates, corresponding to an abnormal addition of hydrogen bromide or mercaptan. No normal addition of bisulfite is known where the double bond is not conjugated with a carbonyl group. When such conjugation occurs, there is no peroxide effect; experiments with crotonic acid (96) show that, just as in hydrogen bromide additions, oxygen here has no effect either on the rate or the direction of addition. Liquid cyclohexene, 2-pentene, trimethylethylene, 2 , 4,4-trimethyl- 2-pentene, isoprene, and pinene were shaken with bisulfite solutions under constant oxygen pressures (130). Oxygen was consumed slowly during the course of the reaction, and, if the supply was interrupted, interaction of the hydrocarbon with bisulfite also stopped. The first three of the substances named gave as much as 90 per cent or more of sulfonates, the yields being greatest a t 30 mm. of oxygen and decreasing progressively a t 152 or 760 mm. The last three alkenes reacted less easily and gave yields of only 15 to 20 per cent. Cyclohexene gave a cyclohexylsulfonate; no other addition products were identified. The product from isoprene still contained one double bond; that from pinene was unstable. Trimethyl- ethylene and 2-pentene seemed to give mixtures of sulfonates. Additions to allyl alcohol were carried out in sealed tubes a t 100°C. (75). Evacuation of these tubes did not prevent addition, but when 10 mole per cent of hydroquinone was added to the reaction mixture before evacua- tion, no reaction occurred. In the presence of oxygen, up to 65 per cent of sodium 3-hydroxypropane-l-sulfonate mas obtained, as demonstrated by conversion of the product t o 3-chloropropane-l-sulfonamide. On the assumption that unsymmetrical reagents add to allyl alcohol in the same way that they add to the allyl halides, the above addition to allyl alcohol is abnormal. In additions to both propene (75) and the liquid alkenes (130), sodium or ammonium nitrites, which are also capable of oxidizing bisulfites, have been found to exert an effect like that of oxygen. In experiments with cychlohexene, 0.06 mole of nitrite per mole of bisulfite, introduced slowly over a long period, gave 84 per cent sulfonate, whereas a larger proportion of nitrite (0.1 mole), introduced all a t once, gave only 55 per cent yield. In an experiment with trimethylethylene, 0.005 mole of sodium nitrite 396 FRANK R. MAY0 AND CHEVES WALLING per mole of bisulfite gave 90 per cent yield of addition product, indicating that about 180 molecules of organic sulfonate were formed per molecule of nitrite reduced. It has already been shown by Franck and Haber (23) and by Back- strom (8) that the oxidation of sulfite and bisulfite are chain reactions involving the *SO3- ion radical and the *HS03 radical. The necessity for using oxygen or other oxidizing agents in the addition reaction, the small proportion of agent required, and the advantage of introducing this agent gradually, as well as the inhibition of the reaction by hydroquinone, and the fact that the product corresponds to an abnormal addition, all suggest a chain reaction involving free radicals (75): SO3-- + oxidant + 4 0 3 - + oxidant- *S03- + RCH=CHz + RCHCHzS03- RCHCHzS03- + HSOa- + RCH2CHzS03- + *S03- The extent to which the sulfite-ion radical and the sulfonate-ion radical may be associated with a proton is not known. 'The range of pH over which addition can take place suggests that either or both charged and uncharged radicals may participate. The oxidation of bisulfite to bisulfate during oxygen-catalyzed additions causes an increase in the acidity of the solution and a retardation of the addition. The increase in acidity can be overcome by substituting sulfite for that part of the bisulfite oxidized, so that normal sulfate rather than bisulfate is formed. If too large a proportion of sulfite is used, some of this substance adds to the alkene, thus liberating an equivalent of alkali which also retards the addition reaction. If the ratio of sulfite to bisulfite in the initial solution is the same as the ratio of sulfate t o sulfonate in the reaction products, then the pH of the sulfite-bisulfite buffer remains practically constant and the addition of bisulfite proceeds a t a maximum rate and to a maximum extent. This optimum proportion of sulfite t o bisulfite varies with both the rate of oxidation and of addition, and there- fore depends upon the concentration of sulfite, the alkene used, and the other experimental conditions. The pH of the sulfite-bisulfite buffer depends upon the sulfite-bisulfite ratio and on the cation. Although the pH of such buffers is thus of little theoretical significance, some representative data are cited here. In the addition to cyclohexene, using oxygen a t 1 atm. pressure, the optimum pH of a sodium salt buffer was about 6.4 and that of a similar ammonium salt buffer about 6.0 (130). In additions to propene, oxygen was supplied intermittently; the optimum pH of a sodium salt buffer was 5.8 (76.5 per cent bisulfite, 23.5 per cent sulfite), and that of an ammonium salt buffer was 6.0 (55 per cent bisulfite, 45 per cent sulfite). With propene, the pH of these solutions remained (22) (23) (24) THE PEROXIDE EFFECT 397 constant while 95 per cent of the available sulfite was consumed, and the proportions of sulfonate and sulfate formed corresponded closely to the bisulfite-sulfite ratio (29). It follows that if an alkene reacts sluggishly with bisulfite, the proportion of oxidation increases and a higher proportion of sulfite should be used. Although the only products isolated from the reaction of bisulfites with simple alkenes are the addition products, there is evidence that some by- products are formed. Kolker and Lapworth (100) observed the formation of variable, but usually very small, proportions of what they thought t o be sulfite esters. These by-products were easily oxidized by bromine and permanganate. When hydrolyzed with acid, they yielded sulfur dioxide. By hydrolysis with acid, the formation of small amounts of sulfur dioxide has in some cases been confirmed qualitatively in this laboratory, but since most crude addition products do not reduce iodine (29), the ability of these products to reduce bromine and permanganate is more likely due to unsaturation rather than to the presence of sulfite esters. Attempts have been made, by bromide-bromate titration (105) and by permanganate assay (81), to estimate the degree of unsaturation in the crude sulfonic acids. Since the two methods do not agree and since the existence and proportions of unsaturated sulfonate, hydroxysulfonate, and sulfite esters in the addition products have not yet been demonstrated, the results cannot be considered reliable. In additions to propene (29), about 15 per cent unsaturation in the products is claimed when the buffers are of nearly ideal composition. With more acid or more basic buffers, the yields of sulfonic acids are much lower, but the proportion of unsaturation increases with the acid concentration. In additions to liquid alkenes using oxygen a t 1 atm. pressure, 1.5 per cent to 18 per cent unsaturation has been reported in the product. With nitrite instead of oxygen, i t was shown conclusively that unsaturation was absent (130). The significance of unsaturation will be considered in the case of styrene, where various reaction products have been isolated and analytical methods have been tested. The yields of addition products from the less reactive alkenes are decreased by the use of either alcohol or hydrocarbon solvents and slightly increased by the use of ethylenediamine. Since ethylenediamine has little effect in nitrite-promoted additions, i t probably serves mostly to inhibit the oxidation of bisulfite (29, 130). Attempts to add bisulfites t o acetylene yielded only traces of unidenti- fied sulfonic acids (29). C. THE REACTION O F BISULFITES WITH STYRENE Since the reaction of styrene with bisulfites is fully described in a recent The reactions paper (Sl), the work need be only briefly summarized here. 398 FRANK R. MAY0 AND CHEVES WALLING were carried out under constant oxygen pressure. At 1 atm. of oxygen, sodium and ammonium sulfite-bisulfite mixtures gave 30 per cent or less of organic sulfonates (the remainder of the sulfite being oxidized to sulfate), but yields up to 68 per cent were obtained a t lower oxygen pressures. The reaction of styrene was about as sensitive to excess acid and base as the reactions of those aliphatic olefins which are liquids at room temperature. In the presence of oxygen, the organic sulfonates always consisted of three types of salts. Where sodium salts were used, the final mixture contained about 25 per cent of addition product (I) corresponding to an abnormal addition reaction, 10 per cent of substitution product (11), and 65 per cent of hydroxysulfonate (111) : CeHaCHzCHzSOaNa CeH&H=CHSO3Na CeH6CHOHCH2S0&a I I1 I11 The proportions of these products varied slightly. Ammonium salts (from ammonia or the methylamines) gave somewhat more substitution product (11) and correspondingly less addition product (I) than sodium salts; ethylene diammonium sulfite gave a higher proportion (82 per cent) of hydroxysulfonate (111). Additions of aqueous sulfurous acid to styrene in the presence of a large excess of dimethylaniline or pyridine gave about 60 per cent yields of total sulfonates. This increase is probably due to the fact that these bases increase the miscibility of bisulfite and hydrocarbon, and that they maintain the pH a t a favorable level. Sodium and ammo- nium nitrites and ammonium persulfate can replace oxygen in promoting the reaction of bisulfite with styrene but they are no more efficient, 30 mole per cent or more of these substances being required for a 15 to 25 per cent yield of total sulfonates. With these oxidizing agents, no unsatu- rated sulfonate (11) whatever was found, the proportion of both addition product and hydroxysulfonate being increased. All three types of sulfonates were isolated in the pure state. No one of them is converted into any other under the conditions of the styrene- bisulfite reaction, and all are stable to hot dilute acids and bases. There- fore all three are thought to be primary products. Since the presence of an oxidizing agent is necessary for the formation of all three sulfonates, and since in all of them the sulfur is attached to the terminal carbon atom, the three reactions are easily correlated by the assumption that the sulfonate-ion radical formed by reactions 22 and 23 is an intermediate common to all. If such is the fact, then the low yields of addition product (I) prove that only a small proportion of the sulfonate- ion radicals formed undergo reaction 24, whereas the large amount of oxidizing agent consumed in the styrene reaction suggests that this agent reacts with the sulfonate-ion radical. The unsaturated sulfonate may be THE PEROXIDE EFFECT 399 formed through reaction 25, oxygen being the only oxidizing agent known t o give this result: RCHCR2S03- + 0 2 + RCH=CHS03- + HOz* (25) The H02* radical may react with bisulfite. is formed in the presence of oxygen, nitrite, or persulfate: The hydroxysulfonate (111) RCHCH2S03- + oxidant .--f RCHCH2S03- + oxidant- (26) + RCHCHzS03- + HzO + RCHOHCHzSO3- + H+ (27) + The difference between styrene and the alkenes apparently lies in the ease with which the sulfonate-ion radical reacts with bisulfite (reaction 24). The aliphatic free radicals apparently need little or no activation energy in order to undergo this reaction, and since the concentration of bisulfite is much higher than that of oxygen, the simple alkenes give long chains. The substituted benzyl radical formed from styrene obviously reacts slug- gishly with bisulfite, but much more easily with oxygen, considering the low concentration of the latter in solution. If the slow reaction of bisulfite with the substituted benzyl radical is due to the stabilization of the latter by resonance, then this stabilization has little or no effect on the ability of the radical t o react with Oxidizing agents. Except for compounds containing carbonyl groups, cinnamyl alcohol seems to be the only styrene derivative whose reaction with bisulfite has been investigated. The reaction depends upon the presence of oxygen (75), but no products have been identified. V. THE PEROXIDE EFFECT IN REARRANGEMENTS A. REARRANGEMENT O F l-BROMO-2-BUTENE AND 3-BROMO-l-BUTENE The peroxide effect in the rearrangement of l-bromo-2-butene and 3- bromo-l-butene was discovered (66) during a study of the addition of hydrogen bromide to butadiene, of which the bromides in question are the two addition products: p XCH&H=CHCH3 (“crotyl halide”, IV) CHFCHCH=CHZ + H X . (28) ~r ’ CH2=CHCHXCI& (“secondary halide”, V) (29) It was shown by Winstein and Young (166) that either bromide when pure undergoes, on standing a t room temperature, an allylic rearrangement to an equilibrium mixt’ure of 85 per cent crotyl bromide (IV) and 15 per cent secondary bromide (V). Later, it was shown in this laboratory (66) 400 FRANK R. MAY0 AND CHEVES WALLING that under the combined influence of hydrogen bromide and ascaridole, rearrangement, even a t - 12"C., is rapid and complete. Under the same conditions, ascaridole alone has no effect; hydrogen bromide alone causes only slight rearrangement. Young and Nozaki (169) have since employed hydrogen bromide and benzoyl peroxide to accelerate these rearrange- ments, but the reviewers know of no other application of the early finding. The rearrangements of some homologs of these bromides are reported to be highly susceptible to the effects of traces of unspecified catalysts (170). It is probable that the peroxide effect is widespread in allylic rearrange- ments of bromides, and that therefore the additions of hydrogen bromide and bromine to conjugated systems need reinvestigation. That the allylic rearrangement sometimes has a molecular or ionic mech- anism is shown by a study of the behavior of l-chloro-2-butene (IV) and 3-chloro-l-butene (V) (71). As expected, peroxides and air have no effect on their rearrangements, which are very slow except in the presence of a catalyst. Small amounts of anhydrous ferric chloride or of a mixture of cuprous and hydrogen chlorides cause isomerization of either chloride to an equilibrium mixture containing about equal proportions of each isomer. This isomerization is rapid with the former reagent, somewhat slower with the latter. A large proportion of hydrogen chloride alone causes slow rearrangement to a different equilibrium mixture, indicating that this acid forms a complex with one or both of the organic halides. These phe- nomena show that the rearrangements of these chlorides have much in common with the normal addition of halogen acids to alkenes. The facts that small amounts of both hydrogen bromide and peroxides are required for a very rapid rearrangement of bromides, and that allyl- type chlorides are not susceptible to corresponding influences, suggest that the peroxide-catalyzed rearrangement of bromides proceeds by a chain mechanism involving bromine atoms or free radicals. On this meagre basis, two such mechanisms are tentatively suggested : CH2=CHCHBrCH3 + Br* F? CHZBrCHCHBrCH3 F? CH2BrCH=CHCH3 + Br* (30) CH&=HCHCHB (indistinguishable from CH&H=CHCHa) + CHZBrCH=CHCH3 Ft CH2=CHCHBrC& + CH&H=CHCHa (3 1) There are two difficulties with the first mechanism. One is that in the addition of a bromine atom to l-bromo-Zbutene the bromine atom might not attach itself t o the 3-carbon atom,20 as required for rearrangement. 1 0 This difficulty may not arise if i t is assumed that the bromine atom adds to the ethylene bond to give the more stable free radical. THE PEROXIDE EFFECT 401 The other difficulty is that, wherever a bromine atom adds to either isomer, subsequent separation of a bromine atom from the free radical formed may be too endothermic for a chain reaction. The second mechanism avoids both of these objections. The energy change involved in reaction 31 is very close to zero, and if the exchanges proceed through a chain reaction, the activation energies must also be small. The free radical necessary to start the chain may be formed as follows: CH2BrCHCHBrCH3 (cf. reaction 30) + CH2BrCH=CHCH3 --$ CHzBrCHBrCHBrCHa + #CHzCH=CHCH3 (32) The conditions for the rearrangement of the products having been es- tablished, the addition of hydrogen chloride and hydrogen bromide to butadiene may be considered. As mentioned in the discussion accom- panying table 1, the addition of hydrogen chloride to butadiene gives, over a wide range of temperatures (71), 75 to 80 per cent of secondary chloride (V) and 20 to 25 per cent of crotyl chloride (IV). Isomerization of the products under the conditions of addition is negligible. Addition of hydro- gen bromide under conditions most favorable for the normal addition (pres- ence of an antioxidant, absence of air, temperature -78°C.) gives within experimental error the same proportion of isomers. As the temperature of addition is raised to 25"C., the proportion of crotyl bromide formed in- creases to 56 per cent. If the addition is carried out in the presence of a peroxide, only 40 t o 45 per cent of crotyl bromide is formed a t -78"C., but 70 to 80 per cent of this bromide (approximately the equilibrium mixture) is formed a t -12°C. Since the addition of hydrogen chloride shows no temperature effect, and since a t low temperatures in the presence of antioxidants essentially the same results are obtained with hydrogen bromide, it is concluded that the normal addition of a halogen acid t o 1,S-butadiene gives about 80 per cent 1 ,&addition and 20 per cent 1,4-addition. In additions of hy- drogen bromide a t room temperature, or in the presence of peroxides, some or all of the additional crotyl bromide found is due t o isomerization of secondary bromide first formed by 1 ,Zaddition. The possibility that addition by an abnormal mechanism increases the proportion of direct 1,4-addition has not been ruled out, but the failure to find any 4-bromo-1- butene excludes the possibility that appreciable abnormal 1 ,2-addition has occurred in any experiment made to date. B. REARRANGEMENT OF a-BROMOACETOACETIC ESTERS It has been shown by Hantzsch and coworkers (33) that a-bromo- acetoacetic ester rearranges slowly a t room temperature to ybromoaceto- acetic ester, that this change is accelerated by hydrogen bromide, and that 402 FRANK R . MAY0 AND CHEVES WALLING i t is inhibited by water. However, the a-bromo ester, in spite of its instability a t room temperature, can be distilled four or five times in vacuo at 100°C. without change. The effect of hydrogen bromide explains the differences between the methods for preparing the two esters directly (21). When acetoacetic ester is brominated in the presence of ice and water, pure a-bromo ester is immediately obtained. When slow bromination lasting several hours takes place in carbon disulfide solution, the y-bromo ester is formed. If, in the latter preparation, hydrogen bromide is removed by intermittent washings with water, mixtures are obtained. This result suggests that the y-bromo ester is formed by rearrangement of the a-bromo ester. The a- and y-bromo derivatives of methyl a-methylacetoacetate have been similarly prepared. Ethyl a-chloroacetoacetate has no ten- dency to rearrange, even in the presence of hydrogen chloride (21). This difference between the a-chloro and a-bromo esters suggested the possi- bility of a peroxide effect in the rearrangement of the bromo esters. By a series of experiments, each lasting only a few hours and carried out in glacial acetic acid solution, it was found that air, a peroxide, or light greatly accelerated the rate of rearrangement of the a-bromo ester by hydrogen bromide (68). In the absence of hydrogen bromide, no re- arrangement took place in the presence of a peroxide and light, either with or without hydrogen chloride. In glacial acetic acid, the bromination of ethyl acetoacetate in the absence of air, peroxides, and light gave more than 90 per cent a-bromo ester; in the presence of any one of these agents, 80 per cent or more of the y-isomer was formed. The bromination of ethyl a-methylacetoacetate was very similar, except that rearrangement was more rapid. The effects of oxygen, peroxides, and light on the rearrangement of a-bromoacetoacetic esters by hydrogen bromide and the stability of the corresponding chloro esters suggest that the rearrangement has much in common with the rearrangements of the butenyl halides and the addition of hydrogen bromide to alkenes. Probably a part, if not all, of the rear- rangement takes place through a chain mechanism in some stage of which bromine atoms are involved. Stability of the bromo esters in thepresence of a suitable inhibitor would indicate whether the isomerization is ex- clusively a chain reaction and whether the chain carrier is the same as in the abnormal addition of hydrogen bromide to alkenes. Such experiments have not yet been performed; consequently any discussion of a mechanism is admittedly speculative. Only in order to show that a simple chain can be written is the following mechanism ventured: *CHZCOCHZCOOCZH~ 4- CH3COCHBrCOOC2H6 CH2BrCOCH2COOC2H6 + CH3COCHCOOC2H6 (33) CH~COCHCOOCZH~ $ *CHzCOCHzCOOC2Hs (34) THE PEROXIDE EFFECT 403 The mode of formation of bromine atoms has been indicated in an earlier section. These atoms may attack ester molecules to yield the free radicals necessary to start chains. Since the bromination of acetoacetic ester is reversible, the bromine atoms may attack a bromo ester molecule to give either bromine or hydrogen bromide. Like the second mechanism pro- posed for the rearrangement of the butenyl bromides, the chain here sug- gested consists of the transfer of bromine atoms from a molecule to a radi- cal; the chain carrier is not a bromine atom, but a radical which can iso- merize. Whatever the mechanism of the rearrangement, the product should be an equilibrium mixture of the y-bromo ester with a small pro- portion of the a-isomer. C. CIS-TRANS ISOMERIZATIONS A thorough discussion of the rearrangement of geometrical isomers is beyond the scope of this review on the peroxide effect. Except for recent work, the subject has been covered by both R. Kuhn (24) and Dufraisse (28). In order that isomerization may occur, the resistance of the double bond to free rotation must be overcome. The assumption by some in- vestigators that all such isomerizations take place by a single mechanism has led to some confusion, for a t least four distinct mechanisms will be cited for the conversion of a cis-ethylene derivative to its trans-form. The first mechanism is associated with the simplest reaction, repre- sented by the homogeneous, apparently unimolecular, rearrangement of dimethyl maleate in the vapor phase (119, 147). This reaction has a fairly high activation energy and requires a temperature of around 300°C. The uncatalyzed liquid-phase isomerizations of isostilbene and its a-chloro derivatives (149) are also unimolecular. They require a temperature of a t least 200°C. and have activation energies of about 35 kg-cal. These isomerizations apparently depend on violent collisions to effect rotation about the double bond. Ultraviolet radiation is known to cause isomerization of ethylene deriva- tives in the absence of other catalysts. According to Mulliken (118a), absorption of such radiation by an ethylene derivative causes a transition to an excited electronic state in which the perpendicular configuration of the groups placed about the double bond is more stable than the planar configuration characteristic of the unexcited state. A third type of mechanism is associated with a catalyst which can donate a proton or accept a pair of electrons. Mineral acids (24, 28, 137, 138, 149) are known to be effective in many instances and ineffective in others; primary and secondary, but not tertiary, amines rapidly isomerize di- methyl maleate (18); aluminum, ferric, and zinc chlorides have been effective with the same ester (26) ; boron trifluoride has been found to rear- range isostilbene, but not dimethyl maleate (127). Various workers have 404 FRANK R . MAY0 AND CHEVES WALLING suggested that one of the doubly bound carbon atoms shares a pair of electrons with the catalyst, leaving the other previously doubly bound carbon atom with a positive charge and free to rotate about the axis of the former double bond. After such rotation, dissociation from the catalyst permits reestablishment of the double bond and consequent formation of the geometrical isomer. If a carbonyl group is conjugated with the double bond, association of the catalyst may take place a t the carbonyl group, but then the double bond is shifted and rotation between the carbon atoms previously doubly bound becomes possible : \ I \ I / C=C-C-+HX* C-CzC- .. /(+) * ' .o. :0: .. WHX Most of the remaining observations on the effects of catalysts suggest that a fourth type of mechanism involves catalysts with two unpaired or an odd number of valence electrons. For the sake of brevity, it will be assumed that all such catalysts act through a single mechanism, although future work may show that this class of substances should be subdivided. The general scheme is represented by reaction 36: VI VI1 As in the proton-catalyzed mechanism, rotation about the double bond in the intermediate is possible, but the intermediate here is a free radical. I t s stability is unknown. The catalysts include univalent atoms, free radicals, molecules with odd electrons, and paramagnetic substances in general. One class of catalysts which has been assumed to function in this manner consists of the alkali metals (24); these cause isomerization without much reaction on the part of the metal. That the isomerization of isostilbene by stilbene disodium (in the absence of free metal) involves free radicals or metal atoms is possible, but doubtful in view of the state- ment by Ziegler and Wollschitt (171) that isomerization takes place on regeneration of the ethylene compound (after exchange of metal), but not on the appearance of a single free valence. Platinum and palladium blacks have been found to isomerize maleic acid (148) and its methyl ester (24). Paramagnetic metal ions are also said to isomerize the acid (148). The weak effect of oxygen in accelerating both the vapor-phase (147) and liquid-phase (146, 148) isomerizations may be partly due to the paramag- netism of this substance, but in the isomerization of isostilbene (149) the THE PEROXIDE EFFECT 405 effect of oxygen has been ascribed to the fact that i t causes the formation of catalytically active acids. Since nitrogen oxides (24, 27, 146) are better catalysts than oxygen for some isomerizations, the probability that they function through a free radical addition product is somewhat greater. A combination of hydrogen sulfide and sulfur dioxide causes isomerization of maleic acid, although neither gas alone is effective. Heating with aqueous bisulfite has caused rearrangement of erucic acid. In both cases the effect has been attributed (24) to the colloidal sulfur formed. More recent work has shown that a combination of aqueous sulfur dioxide and manganese dioxide (120) rearranges maleic acid, its esters, and citraconic acid. In additions of mercaptoacetic acid and glutathione to maleic acid (118), part of the unreacted maleic acid was isomerized to fumaric acid. Glutathione would neither add to, nor rearrange, cis-cinnamic acid; hence it was concluded that the sulfhydryl group must be able to add to a double bond in order t o cause isomerization. The close analogy between all of of these observations and those on the addition of mercaptans and bisulfites suggests that free radicals (formed as intermediates in the oxidation of bisulfites or hydrogen sulfide by air, peroxides, or manganese dioxide) are the active agents for the observed isomerizations. This fourth type of mechanism is closely related to the peroxide effect, because halogen atoms can cause the reactions in question. Wachholtz (159) found that the photochemical isomerization of dimethyl maleate by bromine in carbon tetrachloride solution depended only on the quanta absorbed by the bromine, and that quantum yields as high as 600 were ob- tained. The isomerization was thought to proceed according to reaction 36, but an exchange of bromine atoms between the free radicals (VII) and the unsaturated molecules (VI) seems more likely than a dissociation. Even higher conversions were obtained when bromine atoms, the pre- sumably active agents, were generated by chemical means (160). In water solution, the action of ferrous sulfate on bromine, hypobromous acid, or bromic acid isomerized 10,000, 1000, and 500 molecules, respec- tively, of maleic acid per atom of bromine formed. Iodine in the light (24, 28) and a t elevated temperatures (167) has been found to catalyze other isomerizations; here iodine atoms seem to be the active agent. Since it is well established (24, 37) that bromine atoms can cause cis- trans isomerizations proceeding through chain reactions, the fact that hydrogen bromide can cause similar isomerizations under conditions favor- able for previously described peroxide effects is strong support for the con- tention that, where hydrogen bromide is thus effective, bromine atoms are involved. The peroxide effect in the rearrangement of geometrical isomers was observed in this laboratory in 1937, but only the first portion of this work has yet appeared. Urushibara and Sinamura have subsequently 406 FRANK R. MAY0 AND CHEVES WALLING published several papers in this field; their observations agree with, and extend, those made here. The following description is intended to bring out the relations indicated above and to show how the principal mech- anism of isomerization changes with the structure of the unsaturated compound. In the absence of air and light, the isomerization of stilbene (93, 67, 150) by hydrogen bromide or hydrogen chloride in benzene solution is very slow, requiring several days. The isomerization by hydrogen bro- mide is accelerated by light, air, peroxides, or by reduced iron or nickel. The accelerating effect of these agents can be overcome by catechol, less effectively by diphenylamine. The isomerization by hydrogen chloride is unaffected by light or peroxides. These results show that there is a slow isomerization by acids through a molecular or ionic mechanism, but that the isomerization by hydrogen bromide proceeds through a much more rapid chain mechanism. That this chain mechanism involves bro- mine atoms is suggested by the known ability of such atoms to cause iso- merizations, together with the fact that small proportions of stilbene dibromide are formed as a result of the oxidation of hydrogen bromide in the presence of air and light (93, 150). a,a-Dichlorostilbene is isomerized by a combination of hydrogen bromide and oxygen, but not by halogen acids alone (149). In the isomerization of maleic acid and some of its derivatives when the halogen acids are present, rearrangement by the bromine-atom chain mechanism is negligible compared with that by the ionic or molecular mechanism. Sinamura (137) found that the isomerization of dimethyl maleate by hydrogen bromide is unaffected by oxygen or antioxidants and that hydrogen chloride is nearly as effective as hydrogen bromide in pro- ducing the reaction. Kharasch, Mansfield, and Mayo (93) found that isomerization of maleic acid, maleic ester, and bromomaleic acid in air was catalyzed to about the same extent by both hydrogen bromide and hydro- gen chloride. Further work by Sinamura (138) shows that the behavior of methyl allocinnamate is intermediate between that of maleic ester and stilbene with respect to both mechanisms. With this compound, hydrogen chloride is distinctly less effective than hydrogen bromide, and the isomerization by hydrogen bromide is only moderately accelerated by oxygen or partially inhibited by catechol. Work in this laboratory (93) shows that isomeriza- tion of the labile form of a-phenylcinnamic acid to the stable form by hydrogen bromide is accelerated by air and light, whereas the apparently slower isomerization by hydrogen chloride is not. From the foregoing discussion it seems probable that acids usually cause isomerization of a cis- to a trans-ethylene derivative. This mechanism is THE PEROXIDE EFFECT 407 most important when the ethylene bond is conjugated with one or more carbonyl groups. Probably any type of ethylene derivative can also iso- merize by the atom or radical type of mechanism. Examples of the isomerization of maleic acid derivatives by the bromine-atom mechanism have been cited, but two groups of workers have failed to observe any evidence of this mechanism in the presence of hydrogen bromide. It may therefore be concluded tentatively that, with maleic acid derivatives, the action of hydrogen bromide or hydrogen chloride through the polar mech- anism is large compared with the action of the former reagent through the atom mechanism. The reverse is true for isostilbene, whereas the behavior of allocinnamic ester is intermediate. Such considerations show also that the varying effectiveness of different reagents, which has often in the past been described as anomalous, may easily be fitted into a general scheme. Some cis-derivatives should be expected to rearrange easily by several mechanisms, whereas others may rearrange with difficulty by some or all mechanisms. Accordingly, it is not a t all surprising that one group of investigators found no correlation between catalytic activity and magnetic susceptibility (26). In the large amount of experimental work required to permit comparisons between different alkenes or different catalysts, it will be necessary to establish the mechanism of each isomerization. VI. CONCLUSION In all the studies discussed in the foregoing review the experimental facts seem to be best explained by the hypothesis that the striking effects of oxygen and peroxides arise out of their ability t o initiate chain reactions in which atoms or free radicals act as chain carriers. So-called solvent effects and inhibitory effects of traces of various materials are best inter- preted as the result of their effects on chain reactions. Many discrepancies in observations recorded in the earlier literature are explained ; many hitherto isolated phenomena are correlated; and many supposed abnormal- ities are reduced to parts of a logical pattern. It is perhaps of even greater significance that a new and broader outlook on organic reactions in general is opened up by this hypothesis. The con- cept of chain reactions in solution, involving atoms or free radicals, will doubtless in many cases supersede earlier limited and inadequate notions which ascribed all reactions to simple unimolecular or bimolecular mech- anisms. In any event, it now seems a necessary supplement t o such ideas. Even though future work may necessitate changes in the theoretical concepts of chain reactions, the present ideas have served as a powerful working hypothesis. Already they have been applied in this laboratory 408 FRANK R. MAY0 AND CHEVES WALLING to several classes of reactions other than those here discussed. Such applications are the following: 1. The bromination of phenanthrene (72), toluene (73), cyclopropane (78), cyclohexane, methylcyclohexane, isobutane (go), aliphatic acids, acid halides, and anhydrides (91). 2. The use of sulfuryl chloride (in the presence of a peroxide) as 8 chlorinating agent for aliphatic compounds (79, 82, 86). 3. The use of sulfuryl chloride (under illumination in the presence of pyridine) as a sulfonating agent for aliphatic compounds (80, 89). 4. The introduction of the -COC1 group into aliphatic compounds by the use of a peroxide and oxalyl chloride (92), or by the use of light and either oxalyl chloride or phosgene (85). 5 . The use of peroxides in accelerating the chlorination of hydrocarbons in the dark (88). It is hoped that a review of these and related phenomena will be under- taken when sufficient data become available. The principles discussed in this paper have been developed in collabora- tion with Prof. M. S. Kharasch over a period of several years. He has assisted in determining the scope of this review and has offered many helpful criticisms of its content. Dr. James K. Senior has greatly assisted the authors in the revision of the manuscript. REFERENCES (1) ABRAHAM, E. P., AND SMITH, J. C.: J. Chem. SOC. 1936, 1605. (2) ABRAHAM, E. P., MOWAT, E. L. R., AND SMITH, J. C.: J. Chem. SOC. 1937,948. (3) ALLEN, C. C.: U. S. patent 2,051,807; Chem. Abstracts 30, 6760 (1936). (4) ASHTON, R., AND SMITH, J. C.: J. Chem. SOC. 1934, 435. (5) ASHTON, R., AND SMITH, J. C.: J. Chem. SOC. 1934, 1308. (6) ASHWORTH, F., AND BURKHARDT, G. N.: J. Chem. SOC. 1928, 1791. (7) AXBERG, G., AND HOLMBERG, B.: Ber. 66, 1193 (1933). (8) B ~ C K S T R ~ M , H. L. J.: Z. physik. Chem. B26, 122 (1934). (9) BARR, F. T., AND KEYES, D. B.: Ind. Eng. Chem. 26, 1111 (1934). (10) BAUER, W.: German patent 368,467; Friedlander 14, 104; Chem. Zentr. 1923, (11) BAUER, W.: German patent 394,194; Friedlander 14, 106; Chem. Zentr. 1924, (12) BOORMAN, E. J., LINSTEAD, R. P., AND RYDON, H. N.: J. Chem. SOC. 1933, 568. (13) BRAUN, J. v., AND PLATE, T.: Ber. 67, 281 (1934). (14) BROOKS, B. T.: J. Am. Chem. SOC. 66, 1998 (1934). (15) BROUWER, L. G., AND WIBAUT, J. P.: Rec. trav. chim. 63, 1001 (1934). (16) BURKHARDT, G. N.: Trans. Faraday SOC. 30, 18 (1934). (17) CLARK, R. H., AND GRAY, K. R.: Trans. Roy. soc. Can. 111, [3] 24, 111 (1930). (18) CLEMO, G. R., AND GRAHAM, S. B.: J. Chem. SOC. 1930, 213. (Ma) COFFIN, c. c., SUTHERLAND, H. s., AND MAASS, 0.: Can. J. Research 2, 11, 1218; U. S. patent 1,414,852; Chem. Abstracts 16, 2150 (1922). 11, 1022; U. S. patent 1,540,748; Chem. Abstracts 19, 2210 (1925). 267 (1930). THE PEROXIDE EFFECT 409 (19) COFFIN, C. C., AND MAASS, 0.: Can. J. Research 3, 526 (1930). (20) CONN, J. B., KISTIAKOWSKY, G. B., AND SMITH, E. A.: J. Am. Chem. SOC. 80, (21) CONRAD, M., et al.: Ber. 29, 1042 (1896). (22) FINZI, C., VESTURINI, G., AND SARTINI, L.: Chem. Abstracts 26, 1526 (1931). (23) FRANCK, J. AND HABER, F.: Sitzber. preuss. Akad. Wiss. Physik. math. (24) FREUDENBERG, K. : Stereochemie, p. 913. Franz Deuticke, Leipzig and Vienna (25) GAUBERT, P., LINSTEAD, R. P., AND RYDON, H. N.: J. Chem. SOC. 1937, 1974. (26) GILBERT, W. I., TURKEVITCH, J., AND WALLIS, E. S.: J. Org. Chem. 3, 611 (27) GRIFFITHS, H. E., ASD HILDITCH, T. P.: J. Chem. SOC. 1932, 2315. (28) GRIGNARD, V., AND BAUD, P.: Trait6 de Chimie Organique, Vol. 1, p. 1038. (29) GRIGORIEFF, W. W.: Dissertation, University of Chicago, 1939. (30) GRIMSHAW, D. C., GUY, J. B., AXD SMITH, J. C.: J. Chem. SOC. 1940, 68. (31) GROGGINS, P. H.: Unit Processes in Organic Syntheses, p. 160. McGraw-Hill (32) GROSSE, A. V., AND LISN, C. B.: J. Org. Chem. 3, 26 (1938). (33) HANTZSCH, A., et al.: Ber. 27, 355, 3168 (1894). (34) HARRIS, P. L., AND SMITH, J. C.: J. Chem. SOC. 1936, 1108. (35) HARRIS, P. L., AND SMITH, J. C.: J. Chem. SOC. 1936, 1572. (36) HENNION, G. F., AND WELSH, C. E.: J. .4m. Chem. SOC. 62, 1367 (1940). (37) HEY, D. H., AND WATERS, W. A.: Chem. Rev. 21, 169 (1937). (38) HOBBS, L. M.: Dissertation, University of Chicago, 1938. (39) HOLDER, C. H., AND MAASS, 0.: Can. J. Research lSB, 453 (1938). (40) HOLMBERG, B.: J. prakt. Chem. 141, 93 (1934). (41) HOLMBERG, B.: Chem. Abstracts 32, 4151 (1938). (42) HOOG, H., AND EICHWALD, E.: Rec. trav. chim. 68, 481 (1939). (43) INGOLD, C. K.: Chem. Rev. 16, 225 (1934). (44) INGOLD, C. K., . ~ N D RAMSDEN, E.: J. Chem. SOC. 1931, 2746. (45) IPATIEFF, V. K., AND FRIEDMA?;, B. S.: J. Am. Chem. SOC. 61, 71 (1939). (46) IPATIEFF, V. N., AXD OGONOTVSKY: Ber. 36, 1988 (1903). (47) IPATIEFF, V. K., PINES, H., AND FRIEDMAN, B. S.: J. Am. Chem. SOC. 60, (48) JONES, J. L., AND OGG, R. A., JR.: J. Am. Chern. SOC. 69, 1943 (1937). (49) JOSES, S. O., AND REID, E. E.: J. Sm. Chem. SOC. 60, 2452 (1938). JONES, (50) KAKEKO, T.: Chem. Abstracts 33, 2105 (1939). (51) KANEKO, T., AND MII, S.: Chem. Abstracts 33, 2106 (1939). (52) KHARASCH, M. s., AND REINMUTH, 0.: J. Chem. Education 8, 1703 (1931). (53) KHARASCH, M. S., AND MAYO, F. R.: J. Am. Chem. SOC. 66, 2468 (1933); 80, (54) KHARASCH, M. S., MCNAB, M. C., AND MAYO, F. R.: J. Am. Chem. SOC. 66, (55) KHARASCH, M. S., MCNAB, M. C., AND MAYO, F. R.: J. Am. Chem. SOC. 66, (56) KHAR.4SCH, &f. s., HAxNuAf, c. W., AND GLADSTONE, 31.nd.: J. Am. Chem. Soc. 2764 (1938). Klasse, p. 250 (1931). (1933). (1939). Masson et Cie., Paris (1935). Book Co., Inc., New Pork (1938). 2731 (1938). S. 0.: Dissertation, The Johns Hopkins University, 1936. 3097 (1938) ; slight correction in reference 87. 2521 (1933). 2531 (1933). 66, 244 (1934). 410 FRANK R. MAY0 AND CHEVES WALLING (57) KHARASCH, M. S., AND HANNUM, C. W.: J. $m. Chem. SOC. 66, 712 (1934). (58) KHARbSCH, M. s., .4ND HINCKLEY, J. -4.: J. Am. Chem. SOC. 66, 1212 (1934). (59) KHARASCH, M. s., AND HIKCKLEY, J. A.: J. Am. Chem. SOC. 66, 1243 (1934). (60) KIIARASCH, M. S., AND MCNAB, M. C.: J. Am. Chem. SOC. 66, 1425 (1934). (61) KHARASCH, hl. S., HINCKLEY, J. A., AND GLADSTONE, PI. M.: J. Am. Chem. (62) KHARASCH, I f . S., AXD HANNUM, C. W.: J. .4m. Chem. SOC. 66, 1782 (1934). (63) KHARASCH, M. S., MCNAB, J. G., AND MCSAB, M . C.: J. Am. Chem. SOC. 67, (61) KHARASCH, hf. S., AND MCNAB, &I. C.: Chemistry & Industry 64, 989 (1935). (65) KHARASCH, M. s., .4ND POTTS, W. AI.: J. Am. Chem. SOC. 68, 57 (1936). (66) KHARASCH, M. S., MARGOLIS, E. T., AND MAYO, F. R.: J. Org. Chem. 1, 393 (67) KHARASCH, M. S., MANSFIELD, J. V., AND MAYO, F. R.: J. Am. Chem. SOC. (68) KHARASCH, M. s., STERNFELD, E., AND h f A Y o , F. R.: J. Am. Chem. soc. 69, (69) KHARASCH, M. S., ASD POTTS, W. &I.: J. Org. Chem. 2, 195 (1937). (70) KHARASCH, M. S., ENGELMANN, H. AND MAYO, F. R.: J. Org. Chem. 2, 288, (71) KHARASCH, M. S., KRITCHEVSKY, J., AND MAYO, F. R.: J. Org. Chem. 2, 489 (72) KHARASCH, M. S., WHITE, P. C., AND MAYO, F. R. : J. Org. Chem. 2,574 (1938). (73) KHARASCH, M. S., WHITE, P. C., AXD MAYO, F. R.: J. Org. Chem. 3, 33, 192a (74) KHARASCH, M . S., NORTON, J. A., AND MAYO, F. R.: J. Org. Chem. 3,48 (1938). (75) KHARASCH, I f . S., MAY, E. M., AND MAYO, F. R.: J. Org. Chem. 3, 175 (1938). (76) KHARASCH, M. s., READ, A. T., AND MAYO, F. R.: Chemistry & Industry 67, (77) KHARASCH, M. s., WALLING, c., ASD MAYO, F. R.: J. Am. Chem. soc. 61, (78) KHARASCH, M. S., FINEMAN, At . Z., AND MAYO, F. R.: J. Am. Chem. SOC. 61, (79) KHARASCH, hf. s., AND BROWN, H. e.: J. Am. Chem. soc. 61, 2142 (1939). (80) KHARASCH, M. S., ANDREAD, A. T.: J. Am. Chem. SOC. 61, 3089 (1939). (81) KHARASCH, M. S., SCHENCK, R. T., AND MAYO, F. R.: J. Am. Chem. SOC. 61, (82) KHARASCH, M. S., AND BROWN, H. C.: J. Am. Chem. SOC. 61, 3432 (1939). (83) KHARASCH, M. S., KLEIGER, S. C., AND MAYO, F. R.: J. Org. Chem. 4, 428 SOC. 66, 1642 (1934). 2463 (1935). (1936). 69, 1155 (1937). 1655 (1937). 400, 577 (1937). (1937). (1938). 752 (1938). 1559, 3605 (1939). 2139 (1939). 3092 (1939). (1939). 81 (1940). (84) KHARASCH, M. s., ORT TON, J. A., .4ND MAYO, F. R.: J. Am. Chem. sot. 62, (85) KHARASCH, M. S., AND BROWN, H. C.: J. Am. Chem. SOC. 62,454 (1940). (86) KIIARASCH, M. S., AND BROWN, H. C.: J. Am. Chem. SOC. 62, 925 (1940). (87) KHARASCH, M. S., HAEFELE, W. R., AND MAYO, F. R.: J. Am. Chem. SOC. 62, (88) KHARASCH, M. s., BERKMAN, M., AND MAYO, F. R.: Unpublished work. (89) KHARASCH, M. S., CHAO, T. H., AND BROWN, H. C.: J. Am. Chem. SOC. 62, (90) KHARASCH, M. S., HERED, W., AND MAYO, F. R.: Unpublished work. 2047 (1940). 2393 (1940). THE PEROXIDE EFFECT 41 1 (91) KHARASCH, M. S., AKD HOBBS, L. M.: Unpublished work. (92) KHAR~SCH, &‘I. s., KANE, s., AND BROWN, H. C.: Cnpublished work. (93) KHARASCH, M. S., MAKSFIELD, J. V., . ~ S D MAYO, F. R.: Unpublished work. (94) KHARASCH, &I. S., ASD ~ I A Y O , F. R.: Unpublished work. (95) KHARASCH, M. s., AND hfcr\TAB, M. C.: Unpublished work. (96) KHARASCH, hf. s., - ~ N D SCHENCK, R. T.: Unpublished work. (97) KHAXASCH, hl. S., ASD WALKER, A. 0.: Unpublished work. (98) KHAR-ISCH, h1. s., WHITE, P. C., AND MAYO, F. R.: Unpublished work. (99) KISTIAKOWSKY, G. B., AND STAUFFER, C. H. : J. Am. Chem. SOC. 69, 165 (1937). (100) KOLKER, I., AND LAPWORTH, A.: J. Chem. SOC. 127, 307 (1925). (101) KOZACIIC, A. P., ASD REID, E. E.: J. Am. Chem. SOC. 60, 2436 (1938). (102) LAZIER, W. A . : British patent 406,284; Chem. Zentr. 1934, 11, 132. (103) IAN, K. H., AND ROBISSON, R.: J. Chem. SOC. 1938, 2005. (104) LIKSTEAD, R. P., AND RYDON, H. N.: J. Chem. SOC. 1934, 2001; Chemistry & (105) ILJCAS, H. J., AND PRESSMAN, D.: Ind. Eng. Chem., Anal. Ed. 10, 140 (1938). (106) MAASS, O., A K D SIVERTZ, C.: J. Am. Chem. SOC. 47, 2883 (1925). (107) Maass, O., AND WRIGHT, C. H.: J. Am. Chem. SOC. 46, 2664 (1924). (108) MAILHE, A., AND RESAUDIE, M.: Compt. rend. 196, 391 (1932). (109) MARKOWNIKOFF, V.: Compt. rend. 81, 670 (1875). (110) MASUDA, E.: Chem. Abstracts 33, 131 (1939). (111) hf.4~0, F. R.: Cnpublished work. (112) MAYO, F. R., AND HAEFELE, W. R.: Unpublished work. (113) ~ I A Y O , F. R., . ~ N D SATOYIAS, Ivl. G.: Unpublished work. (114) MICHLEL, -4.: J. Org. Chem. 4, 519 (1939). (115) MICHAEL, A,, ASD LEIGHTON, V. I,.: J. prakt. Chem. I21 60, 445 (1899). (116) MICHAEL, .Q., ASD SHADISGER, G. H.: J. Org. Chem. 4, 128 (1939). (117) ~ ~ I C H A E L , ak., AND F ~ E I X E R , s.: J . Org. Chem. 4, 531 (1939). (118) MORGAN, E. J., A N D FRIED~IAXS, E.: Biocheni. J. 32, 733 (1938). (118a) MULLIKEN, R. S.: Phys. Rev. 41, 751 (1032). (119) KELLES, M., ASD KISTIAKOWSKY, G. B.: J. A4m. Chem. SOC. 64, 2208 (1932). (120) NEOGI, P., AND ;\IITRA, 8. K.: J. Indian Chem. soc. 6, 969 (1929). (121) XICOLET, B. H.: J. Am. Chem. SOC. 67, 1098 (1935). (122) O’CONNOR, S. F., BALDIXCER, L. H., VOGT, R. R., AND HENNION, G. F.: J. Am. (123) OGG, R. A., JR.: J. Am. Chem. SOC. 67, 2727 (1935). (124) PAULIKG, L. : T h e Nature ojthe Chemical Bond , pp. 53, 123. (125) POSN-CR, T.: Ber. 38, 646 (1905). (126) PRICE, C. C., AND COYNER, E. C.: J. Am. Chem. SOC. 62, 1306 (1940). (127) PRICE, C. C., AKD ~IEISTER, M.: J. Am. Chem. SOC. 61, 1595 (1939). (128) ROBINSOS, It. : Outline of a n Electroclkemical (Electronic) Theory of the Course The Institute of Chemistry of Great Britain and Industry 64, 1009 (1935). Chem. SOC. 61, 1454 (1939). Cornel1 University Press, Ithaca, New York (1939). of Organic Reactions. Ireland, London (1932). (129) RUDKOVSKI~, M. AKD TRIFEL, A.: Chem. Abstracts 31, 1004 (1937). (130) SCHENCK, R. T. : Dissertation, University of Chicago (1940). (131) SCHJKNBERG, E. : Ber. 70, 2385 (1937). (132) SCHBNBERG, A., RUPP, E., AXD GUMLICH, W.: Ber. 66, 1932 (1933). (133) SHAKE, R. S. : Dissertation, University of Chicago, 1933. (134) SHERMAN, A., QUIMBY, 0. T., AND SUTHERLAND, R. 0.: J. Chem. Phys. 4, 732 (1936). 412 FRANK R. MAY0 AND CHEVES WALLING (135) SHERRILL, M. L., MAYER, K. E., AND WALTER, G. F.: J. Am. Chem. SOC. 66, (136) SHERRILL, hf. L.: J. Am. Chem. SOC. 66, 1645 (1934). (137) SINAMURA, 0.: Bull. Chem. SOC. Japan 14, 22 (1939). (138) SINAMURA, 0.: Bull. Chem. SOC. Japan 14, 294 (1939). (139) SMITH, J. C.: Chemistry & Industry 66, 833 (1937). (140) SMITH, J. C.: Chemistry & Industry 67, 461 (1938). (141) SMITH, J. C.: A n n u a l Reports o n the Progress of Chemistry (for 1939) 36, 219 (142) SOCIkT6 DES USINES CHIMIQUES RH6NE-POULENC: British patent 438,820; (143) SOLL, J.: German patent 641,878; Chem. Abstracts 31, 5809 (1937). (144) S T o E R m R , R.: In Houben's Die Methoden der Organkchen Chemie, Vol. 2, 926 (1934). (1940). Chem. Abstracts 30, 2993 (1936). p. 982. George Thieme, Leipzig (1925). (145) STOERMER, R.: Reference 144, p. 1008. (146) TAMAMUSHI, B., AND AKIYAMA, H.: Z. Elektrochem. 43, 156 (1937). (147) TAMAMUSHI, B., AND AKIYAMA, H.: Z. Elektrochem. 46, 72 (1939). (148) TAMAMUSHI, B., AND AKIYAMA, H.: Bull. Chem. SOC. Japan 12, 382 (1937). (149) TAYLOR, T. W. J., AND MURRAY, A. R.: J. Chem. SOC. 1938, 2078. (150) URUSHIBARA, Y., AND SINAMURA, 0.: Bull. Chem. SOC. Japan 13, 566 (1938). (151) URUSHIBARA, Y., AND SINAMURA, 0.: Bull. Chem. SOC. Japan 14, 323 (1939). (152) URUSHIBARA, Y., AND TAKEBAYASHI, M.: Bull. Chem. SOC. Japan 11, 692 (153) URUSHIBARA, Y., AND TAKEBAYASHI, M.: Bull. Chem. SOC. Japan 11, 798 (1%) URUSHIBARA, Y., AND TAKrnBAYASHI, M.: Bull. Chem. SOC. Japan 12, 138 (155) URUSHIBARA, Y., AND TAKEBAYASHI, M.: Bull. Chem. SOC. Japan 12, 173 (156) URUSHIBARA, Y., AKD TAKEBAYASHI, M.: Bull. Chem. SOC. Japan 13,331 (1938). (157) URUSHIBARA, Y., AND TAKEBAYASHI, M. : Bull. Chem. SOC. Japan 13,400 (1938). (158) URUSHIBARA, Y., AND TAKEBAYASHI, M.: Bull. Chem. SOC. Japan 13, 404 (159) WACHHOLTZ, F.: Z. physik. Chem. 126, 1 (1927). (160) WACHHOLTZ, F.: Z. Elektrochem. 33, 545 (1927). (161) WALLING, C., KHARASCH, M. S., .4ND MAYO, F. R.: J. Am. Chem. SOC. 61, (162) WALLING, C., KHARASCH, M. S., . ~ N D MAYO, F. R.: J. Am. Chem. soc. 61, (163) WIBAUT, J. P., et al.: Rec. trav. chim. 60, 313 (1931). (164) WIBAUT, J. P., DIEKMANN, J. J., . ~ N D RUTGERS, A. J.: Rec. trav. chim. 47, (165) WINSTEIN, S., AND LUCAS, H. J. : J. Am. Chem. SOC. 60,836 (1938). (166) WINSTEIN, S., AXD YOUKG, W. G.: J. Am. Chem. SOC. 68, 104 (1936). (167) WOOD, R. E., AND DICKINSON, R. G.: J. Am. Chem. SOC. 61, 3259 (1939). (168) YOUKG, C. A., VOGT, R. R., AND NIEUWLAND, J. A.: J. Am. Chem. SOC. 68, (169) Y O U N G , W. G., AND NOZAKI, K.: J. .4m. Chem. SOC. 62, 311 (1940). (170) YOUNG, W. G., RICHARDS, L., AKD AZORLOSA, J.: J. Am. Chem. SOC. 61, 3070 (171) ZIEGLER, K., AND WOLSCHITT, H.: Ann. 479, 129 (1930). (1936). (1936). (1937). (1937). (1938). 1711 (1939). 2693 (1939). 477 (1928). 1806 (1936). (1939). 1x04 G. E. SERNIUK, F. W. BANES AND M. W. SWANEY Vol. 70 and on occasions the temperature was allowed to climb to 160' in order to remove all of this fraction a t atmos- pheric pressure. The pressure was subsequently reduced to 50 to 60 mm. and n-butanol (+OD 1.390 to 1.398) was removed. After removing unreacted ethyl phosphate a t pressures varying between 3 and 10 mm. of mercury (b. p. 66-69' at 3 to 4. mm.; n% 1.404-1.4055), diethyl n-butyl phosphate (b. p. 82-87" a t 3 to 4 mm.) and ethyl di-n-butyl phos- phate (b. p. 95-96' at 3 to 4 mm.) were fractionated out of the reaction mixture. Anal. Calcd.for C8HlnP04: P, 14.7; M ~ 5 1 . 1 . ~ Found: P, 14.9; MR 51.5. Calcd. for C ~ O H ~ ~ P O ~ ; P, 13.0; X ~ 6 0 . 5 . Found: P, 13.2; M ~ 5 9 . 8 . ~ The material balances in all runs including residues amounted to 97 to 99%. Isolation of Ethyl %-Butyl Ether.-Fifty grams of the ethanol-ether fraction (b. p. 36-93') was added to 180 ml. of water. The upper layer, amounting t o 8.9 g., was separated, dried over activated silica gel and distilled. The main fraction boiled at 90-93 and was found to have the following constants: d'4, 0.752; nMD 1.3818; MR calcd., 31.6; found, 31.6. Anal. Calcd. for C6H140: C, 70.5; H, 13.8. Found: C, 70.6; H, 13.9. Treatment of the ether with hydriodic acid yielded ethyl and n-butyl iodides. Alkylation of Morpholine .-An equimolar mixture of morpholine and ethyl phosphate was charged to a round- bottomed flask equipped with a water-cooled reflux con- denser. The mixture was heated by means of a Glas-Col mantle and brought to 150" in fifteen to twenty minutes. At this temperature, vigorous refluxing took place due to heat of reaction, and the mixture changed from a water white to a reddish brown color. If to: well insulated, the reaction temperature may rise to 190 . I t was found, however, that by maintaining the reaction temperature between I57 and 159" good results can be obtained. The product was poured into 500 ml. of water and heated with (8) The molecular refractivities were calculated from that of ethyl phosphate by adding the proper value for the required number of methylene groups to that molecule. 10% excess 144 g.) of sodium hydroxide. The basic aqueous solution was charged to a still and the amine distilled with water as an azeotrope over the range 95- 99.8'. The azeotropic distillate was saturated with po- tassium carbonate whereupon the amine was salted out. After separating from the aqueous layer, the amine was dried over sodium sulfate, filtered and distilled; b. p. 137-138', azo4 0.919, n'% 1.4418; yield 70%. Anal. Calcd. for CBH,,NO: C, 62.2; H, 11.4; N, 12.2. All physical constants given for known compounds agree satisfactorily with those previously published. Acknowledgment.-The authors are indebted to Messrs. N. Beitsch, S . Sass and B. Zeffert of this Laboratory for having performed the neces- sary analytical and physical determinations. Summary Sodium butylate behaves catalytically on a mixture of n-butanol and ethyl phosphate yielding diethyl n-butyl phosphate and ethyl di-n-butyl phosphate. This alcoholysis is accompanied by a side reaction which causes the alkylation of the butylate ion to ethyl n-butyl ether. This behavior indicates that ethyl phosphate under the conditions employed behaves both as a true ester undergoing alcoholysis and as an alky- lating agent. The degree to which each of the products of reaction is produced depends upon the concentration of sodium butylate as well as upon the mole ratio of n-butanol to ethyl phosphate. The alkylation of morpholine to N-ethylmor- pholine by means of ethyl phosphate is also de- scribed. ARMY CHEMICAL CENTER, MD. Found: C,62.3; H, 11.5; N, 11.9. RECEIVED DECEMBER 15, 1947 [CONTRIBUTION FROM THE ESSO LABORATORIES, CHEMICAL DIVISION, STANDARD OIL DEVELOPMENT COMPANY] Study of the Reaction of Buna Rubbers with Aliphatic Mercaptans1 BY G. E. SERNIUK, F. w. B A N E S A N D M. w. SWANEY Introduction The relative proportion of 1,4- versus 1,2-addi- tion of diene units and the elucidation of the par- tial structure of polymers and copolymers of buta- diene have been investigated by various chemical and physical methods such as ozon~lysis ,~-~ per- benzoic acid oxidati~n,~,' potassium permanga- nate oxidation,6 and infrared absorption.* The (1) This paper was presented before the Division of Rubber Chemistry at the American Chemical Society Meeting in Chicago, 1946. (2) Hill, Lewis and Simonsen, Trans. Forodny Soc., 86, 1067 (1939). (3) Yakubchik, Vasiliev and Zhabina, RubEn Chcm. a*d Tech., 16, 780 (1945). (4) Alekseeva nnd Belitzkaya, ibid., 16, 693 (1942). (5) Rahjohn, Bryan, Inskeep, Johnson and Lawson, THIS JOUR- (8) Weidlein, Jr., Chum. Eng. News, 24, 772 (1946). (7) Kolthoff, Lee and Mairs, J . Polymer Science, 2.220 (1947). (8) Rasmussen and Brattain, private communication. NAL, 69, 314 (1947). work presented in this paper was undertaken in an attempt to obtain further information regarding the structure of butadiene polymers and copoly- mers by studying the reaction of these polymers with aliphatic mercaptans. The reaction of mercaptans with unsaturated compounds including natural and synthetic rub- bers is not new. Posner) Gunnar, Axberg and Holmberg, lo Hoag and Eichwald,' Kharasch, Read and Mayo,12 Jones and Reid,13 Cunneen,14 and others have treated mercaptans with various types of unsaturated compounds. Holmberg's treated natural pale crepe rubber with thiogly- (9) Posner, Ber., 85, 646 (1905). (10) Gunnar, Axberg and Holmberg, i b i d . , 66B, 1193 (1933). (11) Hoag and Eichwald, Rec. trav. chim., 66, 481 (1939). (12) Kharasch, Read and Mayo, Chem. and Ind., 67,752 (1938). (13) Jones and Reid, THIS JOURN+ 60,2452 (1938). (14) Cunneen, J . Chem. Soc., 38.134 (1947). (15) Bolmberg, Ber., 66, 1349 (1932). May, 1948 REACTION OF BUNA RUBBERS WITH ALIPHATIC MERCAPTANS 1805 colic acid, and more recently, Kolthoff and co- workers] l6 and Marvel and co-workers1' studied the reaction of aliphatic mercaptans with buta- diene polymers and copolymers in latex form. From a preliminary study of the reaction of mer- captans with model compounds it was found that ethylenic bonds in conjugated] vinyl, terminal butenyl, and in closed ring structures added mer- captans readily, while internal, non-conjugated ethylenic bonds reacted a t a relatively slower rate. Thioglycolic acid added to simple olefins more vig- orously than n-aliphatic mercaptans. Since this paper was originally submitted] Cunneen'4 re- ported the reactions of unsaturated hydrocarbons and various thiols. An apparent order of reactiv- ity was found to be cyclohexene > dihydromyr- cene > squalene > rubber; and for the thiols, thioglycolic acid > thiophenol - isopentanethiol. It is evident that the ease with which an ethylenic bond can add thiols is dependent, in part, upon the structural unit retaining the double bond. It is quite probable that diene polymers contain several types of ethylenic bonds, but from ozonoly- sis data2-5 it must be concluded that a major por- tion of these bonds results from either 1,2- or 1,4- addition of butadiene units to the polymer chains. These two types of ethylenic bonds should exhibit different rates of mercaptan addition and the de- termination of the proportion of mercaptan-re- active units in the polymer chains should represent a measure of the per cent. of ethylenic bonds pres- ent as side vinyl groups. Ethylenic bonds in structures formed by intramolecular cyclization reactions should likewise be mercaptan reactive. Obviously mercaptan addition reactions will not show complete selectivity for side vinyl groups but i t is quite probable that mercaptan-reactive struc- tures other than the side vinyl groups will repre- sent only a very minor portion of the total un- saturation of the butadiene polymers. This paper records the reaction of thioglycolic acid and n-aliphatic mercaptans of CZ to Cle chain length with diene polymers, and the reaction of thioglycolic acid and a n-Clz mercaptan with model compounds. Polymer-mercaptan reactions were effected in solution, mass and in latex form a t various temperatures, in the presence of air, or in the presence of additives which were evaluated as catalysts. The experimental data indicate a pro- nounced difference in the rate and extent of mer- captan addition by the various diene polymers, and the difference in rate and extent of addition has been utilized in estimating the relative pro- portion of external and internal ethylenic bonds in the polymer chains. Experimental Materials Polymers.-The polymers used in studying the polymer- mercaptan reactions were natural rubber, polyisoprene ] polybutadiene, and copolymers of butadiene and acrylo- (16) Kolthoff and co-workers, private communication. (17) Marvel and co-workers, private communication. nitrile, butadiene and styrene, and butadiene and alpha methyl para-methylstyrene. Both emulsion and sodium catalyzed polymers of butadiene were used, but in all other cases, emulsion polymers and copolymers were em- ployed. The emulsion polymers were prepared by the standard technique. The polymerization reactions were discontinued when 75% of the monomers were converted to polymer. The latices were freed of unconverted monomers by steam stripping under a pressure of 50-60 mm. Polymers required for solution and mass reactions were obtained by coagulating the stripped latices with 997, isopropyl alcohol, followed by water washing and drying at 175' F. No attempt was made to fractionate the resulting polymers, or to free them of any developed peroxide materials. Mercaptans.-Mercaptans of G to CC chain length were obtained from Eastman Kodak and were used after distillation. Thioglycolic acid was first dried by removing the water as a benzene azeotrope before distilling under vacuum. Normal mercaptans of CC to CIS chain length were of research grade from the Connecticut Hard Rubber Company. These mercaptans were used directly without further purification. Sharples 3B mercaptan was used after distillation. Procedure.-A modification of the procedure used by HolmbergI6 was used in effecting the reaction of various polymers with thioglycolic acid. A 5% solution of poly- mer in benzene was placed in a flask and agitated while a calculated amount of dry thioglycolic acid was added slowly to the solution at room temperature. The re- actants were allowed free access to air throughout the course of the reaction. With butadiene polymers and co- polymers the reaction was exothermic, and after a short time the solution became cloudy and an insoluble layer separated. The separated product was solubilized by the addition of n-hexanol and the reaction continued. Samples were withdrawn periodically for analysis. The polymer-thioglycolic acid reaction products were purified by water washing the benzene-n-hexanol solutions until no further test for free thioglycolic acid could be obtained by titration with 0.1 N iodine solution. The solvents were then removed by heating the solutions on a steam- bath under high vacuum. The reaction products were further dried in a vacuum oven at 70 '. Sulfur analyses of the products were obtained by combustion in a Parr bomb. Polymers in latex form were treated with mercaptans in 2 oz. and one quart glass reactors which were charged to varying levels and then agitated in a thermostated bath at 50' for varying periods of time. The amount of mercaptan employed corresponded to a 100% excess over the amount theoretically required for complete double bond saturation. Several conditions were employed wherein the free space of the reactors was flushed either with nitrogen, air, or pure oxygen; and the amount of persulfate in the systems was varied. The polymer- mercaptan reaction products were isolated from the emulsions by coagulating in an excess of 99% isopropyl alcohol. The products were thoroughly washed in fres,h portions of alcohol and then dried in a vacuum oven at 70 . Mass reactions of dry polymers and mercaptans were carried out under essentially the same conditions em- ployed by Jones and Reid18 in their study of the reaction of mercaptans with unsaturated compounds. The dry polymer was dissolved in the desired mercaptan, two mols of mercaptan being used per mol of diene in the polymer. The solutions were agitated in a glass reactor, sealed from the atmosphere without displacing the air in the reactor, at 180-200° for varying periods of time. Samples of the reaction mixture were removed at intervals for purification and analysis. The polymer-mercaptan reaction products were isolated by coagulating the reaction mixture with a large volume of 99% isopropyl alcohol followed by re- peated dissolution of the mass in petroleum ether and co- agulation until the mixed solvents showed no trace of free mercaptan as determined by titration with 0.1 N iodine solution. The purified reaction mass y s stripped of solvents and moisture under vacuum at 80 . 1806 G. E. SERNIUK, F. W. BANES AND M. W. SWANEY \'#8
      ! 
              7   %            
8  
           7      8 "  
   
 
 
         
   / ) "           
           7 
   8   %                       "   
             ! 
                   "   
        
           /       ? '   
! 
    79  ''<+8   
      
       @
  7++<, 8    $
  7+A+. 8              $

   7; ">:8      !          ? ' *!

          7#B8 
 
   !    

     #B               
   
 ? )   
  ?"#B 
           
                Scheme 1. Laser-induced polymerization of a diacrylate monomer. Scheme 2. Various deactivation pathways for the excited photoinitiator molecule. Figure 1. Influence of photoinitiator and of laser wavelength on the polymerization of an acrylate photoresist: Ar
, 363.8nm; Kr
, 337.4nm; [Lucirin TPO]=3wt%; [Darocur 1173]=3wt%; light intensity, 30mWcm2. '',)  -.'','(''30 7)00)8      7A0 
 
  8 
 !            !         ' " '0
   
  
            
     #B                           
    #B    %               
 
    % 7B"#B 
 
8 
                     7? '8 $      B"#B 
 
          
 !
                #     ? +    
       &    30   
 /      
    

 !
  6                       -      

      7
'0,  ''8                7 '03  ''8'2          
               
           
     
   "          
         
  
         
   
         !
 !             
 
 '00'   % ! 
 )0 )' ;      7;9#8         
 
                 
  
 
    -           "        
 
 '000' 
           
       
              " 
        
    7#  +A2  1 /
   1 8 
     Figure 2. Time dependence upon UV exposure of the IR band of the acrylate double bond (light intensity, 30mWcm2). Figure 3. Laser-induced polymerization of an acrylate photoresist: Kr
laser, 337.4nm; light intensity, 250mWcm2; N2 atmosphere: -  -  -, Dark polymerization after a 50ms exposure; --, continuous exposure.  -.'','(''30 7)00)8 '',+           !
    !   7= .0  18   
          7  $8 ? ,    !  
    
   
    
         
    

     
   7  $      !   

     8 "                
!       
     
     
   
 )0000     ))                              
           ,00     
     !           
 !             
   $     
         
         7 8  !     -      )) ? 3  
 
   B"#B 
 
  = .0      
     
      
     C                              7' D8       
          ( 
    !  
                   
                
       '.E)   "      
        
   
     
 
               "       
           
      
 
            !      "   
        
                 
                   
   )+ "   
       

 !       %  
 
  
      
        !    7+30(,00 8 :            
           
   
   /        ! 
 
!   )0              
     
  

          

               
  
 
 !  
  
   
    !     
          

           %         6      "!   &             
            
      
 
!! 
                        !            
         7
     8    ; !  
          %! !
  Figure 4. Insolubilization profile of acrylic photoresists upon UV irradiation: curve A, amino-polyester-tetraacrylate; curve B, polyphenoxy-diacrylate. [Irgacure 369]=1wt%; light intensity, 50mWcm2. Figure 5. Influence of the photoinitiator (0.5wt%) on the polymerization profile of a wet acrylic photoresist exposed to visible radiation: I =14mWcm2. '',,  -.'','(''30 7)00)8      
          "       
 
                
    %   
!         "               !          
 7   8    
  "
            !  
                    
  -      
    "
                                       !     
  
  


        !               '
  
  "  

    ! 
!    
 
    
  %                
  
   
  
   
              ! PHOTOPOLYMERIZATION OF BINARY SYSTEMS $   
         
   
  
                   
 
   
            
  "
 
              
  

       
     
   
    
  !  7#>F8    
 " 
         

     
   >    
  
         
   

     
              ), "
 
    
   !           
   !             
     
  

        Acrylate-epoxide blends " #>F 
   / +           
     '6'        
    7= '30  18   
  
 7$   1G!'<2  1 8  
       !    ! 

 !     7

      !    8 ? A    
     B"#B 
 
      
           $ 
     !
   Scheme 3. Polymer networks formed by photoinitiated polymerization of a bisphenol A diacrylate derivative (top) and of a biscycloaliphatic diepoxide (bottom). Figure 6. Photoinitiated polymerization of an acrylate-epoxide blend in the presence of ITX (0.5wt%) and a diaryliodonium salt (2wt%):I =60mW cm2; air.  -.'','(''30 7)00)8 '',3         
      %                    7? A8 "  #>F     +                    70+8 "            
          ! 
       '0!           $  !           
    0+ 7? <8 "
        
   
      7+308 
        !  
 7+'08              7,008 "            #>F           
      

   
              #   
     
          ! #>F        -   
 

     
        /                 
 

    !           03         )3 "  !
     
    
 
     
        
                 !  
      
 
   

     " 
         %      

    
             "       
!    '  
 +A3    
        
    
      
 " ! 
    
  
 

    
          
               Acrylate-vinyl ether blends $              7=8          !   
    
      
   
   !  )A "     
  
   
    
 
 !      C    ! 

     
        
      !

        
              "                 
             

    7">"$8            !     $                                      !
               =    "                     
                   
  
 !      #>F        6  ! 4     
  !      
      / , Acrylate-polyene blends >    
      !        
   

         
          =  
       
       
  
  
  
  
    )< "        !        
 
   
 
   
       
           
                 
 "       / 3 
   ? .  
 
           
  ( Figure 7. Hardening of an acrylate-epoxide blend and of the neat monomers upon UV irradiation in the presence of ITX (0.5wt%) and a diaryliodonium salt (2wt%): I =500mWcm2; air. Scheme 4. Crosslinked polymer formed upon UV irradiation of a mixture of diacrylate and divinyl ether in the presence of both radical and cationic type photoinitiators. '',A  -.'','(''30 7)00)8        
       
     ! 

     $ '0!       
        
    )0 D    7= '308 $ 
   
           
    
     
 
   
 7? .8            !  
   "            
     7> 8  .0 
 +00      !
           "         -              
      #  
         

 !   
  -    / 
!       

      
             4       7+0D8        

 ). Vinyl ether-maleate or maleimide blends $                
   
  
          
     
        )2 +0 C      
        
                       
           
 "
         
!        
                 "          

                               !       "     
        
    ( 
 
    !

    "            
   !     +' "                  
 
  
         
     *               ! 
             
 

   
         "   
   
   
       !          
      
   *               

  
  !
         7? 28          
   !/             
           
      +)      (     
  
 / A             !
!                 
                 Thiol-polyene polymerization >           
          
        
         7B/*8 "
         
   
!     !   

     !      Scheme 5. Photoinitiated copolymerization of acrylate and polybutadiene double bonds. Figure 8. Photocrosslinking of a 20/80 blend of a diacrylate (Ebecryl 150) and a styrene-butadiene block copolymer: - - - -, neat SBS copolymer; [Lucirin TPO]=3wt%; I =500mWcm2; air.  -.'','(''30 7)00)8 '',<                ++ "  
    
   / < 1   %       
          /    73 D8     
 
     
 

       (    
 '0!             +, $       
 
 !  !!
  !!
    7$$8 " 
  
        !
  
           
       -

  
        !         
     !                          
 !   $$    .0°1   'A0°1    
+3 " $$  
 7@   > 8   
          
        7) D  "B#/  =  1 8  

  
     7' D  #  .'2  1 8 " 

    
          #B 
 
 'A+.'            
    7? '08              ?                          
  #            '0000   '  
 
   !     %       '0  
   +A "        ( 
!                     - 7? '08  %   ! 

       Figure 9. Polymerization profiles recorded by RTIR spectroscopy upon UV exposure of a photoinitiator-free maleimide-vinyl ether combination (MI/VE)=1. Scheme 6. Chemical structure of the crosslinked copolymer formed by photopolymerization of a mixture of difunctional vinyl ether, maleate and maleimide monomers. Scheme 7. Thiol-ene polymerization reaction. Figure 10. Photocrosslinking of the ABA-TRIS system upon UV exposure: [TRIS]=2wt%; [Irgacure 819]=1wt%; light intensity, 600mWcm2. '',.  -.'','(''30 7)00)8   "B#/      ' D  
          0) D     
       $$                      
     6     
       !               
!      
       
       " 

    ( 
!             (      

  ? ''      
   $$ $$
   7)0 D8  $$
"B#/ 7) D8  
 
               7' D #  .'28 #  %                    '0    %            "  
           / .             !          
    -
  
          
    !           
      ! 
         
             CONCLUSIONS B              
                 %          
  
 %   C           & 
            
       

    

     
   !     !      ; ! 
    
                          % !   
            !      C   
  !
 
                        
                 :          
     
!                   
  
 '000' /                !     
 
  %      !   
    !     ' "       
       %           
        
 C       !          
      
                      

    
   "
 
         Figure 11. Influence of thiol and acrylate monomers on the photocrosslinking of an ABA rubber: [TRIS]=2wt%; [HDDA]=20wt%; [Irgacure 819]=1wt%. Scheme 8. Crosslinked polymer formed by photoinitiated thiol-ene polymerization of an acrylonitrile- butadiene rubber.  -.'','(''30 7)00)8 '',2                   
            

    !
 >              (
   
 
  
  
    
     


      ( 
  !  
       
  !  "    
 %              
!   
    
    ( 
   "
                       

   !     %  

     
      

             ? 
   
        
    
 
 !
           
            REFERENCES ' 1 E 
  /'6) 7'2.,8 ) @  EH 
  016' 7'22.8 + :  >@"      !  " #$   %& '   "   (       '(3 /#"$ "  ;  7'22'8 , * 1=  @   E? )    "     $1/ /
 / ,'< $ 1  /  C    91 7'2208 3 >

 /> )        !  >  > F G 7'22)8 A ?  E>  B  E? )           !   '(3 1
  *  ;  7'22)8 < ?  E>     (          *    7'2238 . 9 1    '.632+ 7'22A8 2 9 1       !   '. 1*   C 
A'3 7'22<8 '0 B 1     " ) 
  " #$  C F G 7'22<8 '' /  $1  1F  > BC     '         $1/ /
 / A<+ $ 1  /  C    91 7'22<8 ') 9  / %*    ( !       " #$   %&   /#"$ "  ;  7'2228 '+ 9 1   @    .026)+.' 7'2..8 ', 9 1   @   ''6,,33 7'2.28 '3 9 1 +     % '.6',3' 7'2.+8 'A 9 1    36'+' 7'2.+8 '< 9 1       !  " #$   %& '   "   (       3 =  :  >@" /#"$ "  ;  
',3 7'22,8 '. 9 1   @    .2.62A+ 7'2208 '2 9 1 =    9 9 +      
336'<+ 7'22A8 )0 9 1    "   ( )    
   =  "
 ;? C 1H  ? EE $ 1  /  C    91
)0< 7'2.,8 )' 9 1    "    =  "  G @  "
'.< 7'2208 )) 9 1  =   +    /-6.++ 7'22<8 )+ 9 1        .6,< 7'22.8 ), 9 1  1-6'++ 7'22.8 )3 9 1 ; I *  F "  " +     % 316'<<' 7'22A8 )A 9 1  9 9 +   
316A03 7'22<8 )< 9 1  F "  "      '((6+3. 7'2228 ). ; I *  9 1 +     % 3.6

 /> >  > F G
),' 7'22)8 +0 9 1  9 9   306)))2 7'22<8 +'  ? 9 1 EJ  / 1  /1  * 1=   1(6),,< 7'2228      '((6'0A3 7'2228 +) 1  /1 * 1= EJ  /  ?  9 1   1(630A+ 7'2228 ++ E   $?  )           !   + =  ?  E>  B  E? 1
  *  ;  
)'2 7'22)8 +, 9 1  F "  "      '((6'2A3 7'2228 +3 9
   9 @ F >   B " =
 1     
'<, 7'2238 +A 9 1  F "  " +    0'6))0, 7)00'8 ''30  -.'','(''30 7)00)8   Kinetic Study and New Applications of UV Radiation Curing Christian Decker Département de Photochimie Générale (UMR-CNRS N87525), Ecole Nationale Supérieure de Chimie de Mulhouse, Université de Haute-Alsace, 3, rue Werner, 68200 Mulhouse, France E-mail: c.decker@uha.fr Keywords: crosslinking; infrared spectroscopy; interpenetrating polymer networks (IPN); nanocomposites; photopoly- merization 1 Introduction Light-induced polymerization of multifunctional mono- mers or oligomers, also called UV radiation curing, has become a well-accepted technology which has found a large variety of industrial applications because of its unique advantages.[1-4] It is considered to be the most effective way to rapidly transform a solvent-free liquid resin into a solid polymer, at ambient temperature. Under intense illu- mination, the crosslinking polymerization of acrylate- based resins proceeds extensively within a fraction of a second to generate a dense three-dimensional polymer network, showing excellent resistance toward organic solvents, chemicals and heat.[5] Crosslinking can also be readily achieved by UV irradiation of a solid polymer bearing polymerizable functions, as shown in Scheme 1. In both cases, a photoinitiator is needed to generate the active species (free radicals or proton acid), which will initiate the chain reaction. The most important applications of UV radiation curing are to be found in the coating industry for the surface protection of various materials by fast-drying varnishes, paints or printing inks. UV-curable systems are also increa- singly utilized as quick-setting adhesives, sealants, release coatings, as well as to produce composite materials. In Feature Article: Highly crosslinked polymers can be rea- dily synthesized by photoinitiated polymerization of multi- functional monomers or functionalized polymers. The reaction can be followed in situ by real-time infrared (RT- IR) spectroscopy, a technique that records conversion versus time curves in photosensitive resins undergoing ultrafast polymerization upon UVexposure. For acrylate-based resins, UV-curing proceeds with long kinetic chains (7700 mol/ radical) in spite of the high initiation rate. RT-IR spectro- scopy proved very valuable in assessing the influence of various parameters, such as initiation efficiency, chemical structure of the telechelic oligomer, light intensity, inhibitory effect of oxygen, on polymerization kinetics. Interpenetrat- ing polymer networks can be rapidly synthesized by means of UV irradiation of a mixture of difunctional acrylate and epoxy monomers in the presence of both radical and cationic- type photoinitiators. The same UV technology can be applied to crosslink solid polymers at ambient temperature, which bear different types of reactive groups (acrylate and vinyl double bonds, epoxy ring). UV radiation curing has been successfully used to produce within seconds weathering resistant protective coatings, high-resolution relief images, glass laminates and nanocomposites materials. Photoinitiated crosslinking polymerization. Macromol. Rapid Commun. 2002, 23, 1067-1093 1067 Macromol. Rapid Commun. 2002, 23, No. 18
WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, 2002 1022-1336/2002/1812-1067$17.50þ.50/0 Scheme 1. Photoinitiated crosslinking polymerization. photolithography, light-induced insolubilization of photo- resists is being used to produce high-definition images needed for the manufacture of printing plates, optical disks and microcircuits. Besides its great speed and spatial resolution, light-induced polymerization presents a number of other advantages, such as solvent-free formulations, low energy consumption, ambient temperature operations, and tailor-made properties of the photocured polymers. There are two main classes of UV-curable resins, depen- ding whether the chain reaction proceeds by a radical-type or cationic-type mechanism (Table 1). In the first case, aromatic ketones are used to generate free radicals that will initiate the polymerization of acrylate double bonds or copolymerization of styrene with unsaturated polyesters or of the thiol-polyene system by a step-growth addition mechanism.[6] In photoinitiated cationic polymerization, a proton acid is produced by photolysis of triarylsulfonium (TAS) or diaryliodonium salts to initiate the polymerization of epoxides or vinyl ethers.[7] Radical-type systems are by far the most widely used in today’s UV-curing applications, mainly because of their higher reactivity. The large choice of acrylate monomers and telechelic oligomers also allows to adjust more precisely the final properties of the UV-cured polymer for the considered application. During the last decade, a considerable amount of acade- mic work has been devoted to UV radiation curing, with most of the research efforts being focused both on the kinetic and mechanistic aspects of such crosslinking reac- tions and on the design of new photoinitiators, monomers and telechelic oligomers, best suited to producing high- performance polymer networks. A number of textbooks and review articles have reported the progress made in this area over the past few years.[1-19] In this paper, the recent work done in our laboratory, regarding kinetics of ultrafast photoinitiated polymerization of various monomers, func- tionalized oligomers or polymers and monomer blends, is reviewed before presenting some promising novel applica- tions of this environmentally benign technology. Christian Decker has been Research Director at the CNRS (Centre National de la Recherche Scientifique) since 1980 at the Ecole Nationale Supérieure de Chimie de Mulhouse, and head of the Polymer Photochemistry Laboratory at the Department of Photochemistry (University of Haute Alsace). He received his doctoral degree from the University of Strasbourg in 1967 at the Centre de Recherches sur les Macromolécules (CRM). In 1971 he spent a year and a half at Stanford Research Institute in Menlo Park (California) as a post-doc in Frank Mayo’s group, working on the effect of gamma rays on polymers, a topic he continued to investigate at the CRM. In 1975, he joined the Department of Photochemistry to study the reactions induced by light in polymer systems, both photodegradation and photopolymerization. His main research interests include kinetic studies into ultrafast light-induced polymerizations, photocrosslinking of functionalized polymers, characterization of UV-cured polymers and nanocomposites, photostabilization of polymer materials, and laser-assisted chemical processing of polymers. Table 1. Different types of UV-curable resins. Mechanism RADICAL CATIONIC Photoinitiator Monomers and functionalized polymers 1068 C. Decker 2 Experimental Part 2.1 Materials UV-curable formulations are usually made of three basic components: (i) photoinitiator that effectively absorbs the incident light and readily generates reactive radicals or ions, (ii) functionalized oligomer that, upon polymerizing, will constitute the backbone of the three-dimensional polymer network to be formed, and (iii) mono- or multifunctional monomer that acts as a reactive diluent to adjust formulation viscosity and that will be incorporated into the polymer network. The photoinitiator plays a key role in that it governs both the rate of initiation and the penetration of incident light into the sample, and therefore controls the depth of cure. The rate of polymerization depends initially on the reactivity of the functional group, its concentration and the viscosity of the resin. The chemical structure and functionality of both monomer and oligomer are also important, for they will determine the final degree of polymerization, as well as the physical and chemical characteristics of the UV-cured polymer. Chart 1 gives the chemical formulae and trade names of some of the compounds used in our kinetic studies into the polymerization of both radical and cationic-type UV-curable resins. 2.2 Irradiation The liquid resin was applied onto a KBr crystal, a trans- parent poly(propylene) film or a siliconwafer bymeans of a calibrated wire-wound applicator. In some experiments, a second poly(propylene) film was laminated on top of the sample to prevent oxygen diffusion. The thickness of the UV-curable film was set typically between 5 and 25 mm, as determined by IR spectroscopy by means of a calibration curve. For the kinetic analysis, the sample was placed in the compartment of an IR spectrophotometer where it was exposed to UV radiation of a medium pressure mercury lamp (HOYA- SCHOTT 200 U or EFOS Novacure) for a few seconds. The light intensity at the sample position could be set between 5 and 200 mW
cm2, as measured by radiometry (International Light IL-390). Some curing experiments were performed on an industrial-type UV-line (IST Minicure-80 W
cm1), was operated at belt speeds between 5 and 60 m/min, i.e. UV doses of 600 and 50 mJ
cm2 per pass, respectively. 2.3 Analysis Polymerization reactions were followed in situ by RT-IR spectroscopy,[21] bymonitoring the decrease of the IR band characteristic of the reactive functional group upon UV exposure. The sample was exposed simultaneously to the UV beam that induces polymerization and to the IR beam that analyzes its extent, as shown in Figure 1, depicting the Kinetic Study and New Applications of UV Radiation Curing 1069 instrumental set-up. With the FT-IR spectrophotometer used (Bruker IFS-66), up to 57 spectra can be taken per second at a spectral resolution of 4 cm1, thus allowing high speed polymerizations to be accurately followed in real time. The disappearance of the functional groups (acrylate, vinyl ether double bonds, epoxy ring) was monitored continuously by selecting the IR wavenumber where these functional groups exhibit their characteristic absorption bands: 812 cm1 for the acrylate double bond, 1622 cm1 for the vinyl ether double bond, and 795 cm1 for the epoxy ring. The degree of con- version (x) was calculated from the decrease in IR absorbance (A) after a given exposure: x¼ 1 (At/A0). By setting the IR spectrophotometer in the absorbance mode, conversion versus time curves were recorded directly. From the slope of this curve, the actual rate of polymerization (Rp) can be evaluated at any stage of the reaction: Rp¼ [M0] (dx/dt), where M0 is the initial concentration in reactive groups. In most systems Rp reaches its maximum value in the 10 to 30% conversion range. This analytical method also permits to determine the final conversion reached at the end of UV exposure, and thus the amount of unreacted functionalities in the UV-cured polymer, a crucial factor with respect to long-term properties of the materials. The disappearance of the photoinitiator during photopoly- merization was followed by real-time ultraviolet (RT-UV) spectroscopy, by setting the detection wavelength at its maxi- mum absorbance and monitoring continuously its decrease upon UV exposure.[22] The time resolution of the instrument used (Beckman DU-7400) was 0.1 s. Chloroform was used as a solvent to measure the insoluble fraction of the crosslinked polymer. The hardness of the UV- cured polymer was evaluated by monitoring the damping of the oscillations of a pendulum placed onto a glass plate coated with a 30 mm thick film (Persoz hardness). Polymer hardness was shown to be strongly dependent on the glass transition tempe- rature.[23] Persoz values, expressed in seconds, are typically ranging from 30 s for soft elastomeric materials up to 400 s for very hard and glassy polymers. 3 Kinetics of Photoinitiated Polymerization The polymerization of liquid resins, as well as the cross- linking of functionalized polymers, can be readily achieved by UV-irradiation in the presence of a photoinitiator. The various analytical methods commonly used to study the kinetics of photopolymerizations in situ (differential scanning calorimetry,[18,24,25] dilatometry,[26] fluorescence spectroscopy,[27,28] laser interferometry[29,30]) are not cap- able of both following ultrafast reactions in real time and providing quantitative information about the actual degree of polymerization. By these techniques, one monitors the physical effects resulting from polymerization, such as the heat evolved, volume shrinkage, and increase in viscosity or refractive index. A calibration curve is therefore required to correlate these quantities with the actual mono- mer conversion, a curve which is specific for each type of resin formulation. This is not the case for RT-IR spectroscopy, a technique which monitors in situ the chemical changes occurring upon UV exposure, i.e., the disappearance of the reactive group of the monomer.[31] Conversion versus time curves have thus been recorded for polymerizations occurring within a fraction of a second, by radical-type[32] or cationic- type mechanisms.[33] RT-IR spectroscopy proved to be particularly well-suited to study the photopolymerization of monomer mixtures, which leads to the formation of copolymers or interpenetrating polymer networks, as it allows the disappearance of each monomer to be followed in real time.[34] This chapter is focused on the kinetic aspects of both UV radiation curing of different types of liquid resins and photocrosslinking of functionalized polymers, by working under conditions similar to those found in industrial applications, i.e., thin films of solvent-free resins exposed to intense polychromatic radiation in the presence of air. 3.1 Photoinitiated Radical Polymerization RT-IR spectroscopy has been successfully used to evaluate the kinetic chain length and initiation efficiency in UV- curing of acrylate-based resins, as well as to quantify the influence of oligomer structure, light intensity, and atmo- spheric oxygen on the polymerization kinetics. 3.1.1 Kinetic Chain Length Most of the resins employed in UV-curing applications consist of acrylate-based monomers and telechelic oligo- mers, because of the high reactivity of the acrylate double bond. They are associated to aromatic ketone photoinitiators, which undergo fast homolytic cleavage upon UV exposure, according to a Norrish-type I mechanism, to generate free radicals that will initiate the polymerization (Scheme 2). The high value of the propagation rate constant (kp 104 l
mol1
s1)[35] together with a relatively slow termi- nation process (kt 105 l
mol1
s1)[36] are the main reasons why the polymerization of acrylate resins proceeds so fast upon intense illumination. Crosslinking polymer- ization was found to develop with long kinetic chains in Figure 1. Instrumental set-up for real-time infrared spectro- scopy analysis of ultrafast photopolymerizations. 1070 C. Decker spite of the very high rate of initiation.[37] The kinetic chain length (KCL) can be evaluated by monitoring the disap- pearance of both monomer and photoinitiator after short UV exposure. Figure 2 shows the acrylate conversion and photoinitiator decay curves recorded by RT-IR and RT-UV spectro- scopies, respectively, upon continuous UV irradiation of a polyester acrylate resin (Ebecryl 830þHDDA, [M0]¼ 6 mol
l1) containing Lucirin TPO (1 wt.-%) as a photoinitiator, at a light intensity of 20 mW
cm2 under O2-free conditions. After 0.1 s of exposure (dashed lines), polymerization continues to proceed in the dark up to a conversion of 50%, while photolysis of the initiator stops immediately, as expected. This post-polymerization is attributed to the polymer radicals which continue to grow until bimolecular termination. The KCL was evaluated by calculating the ratio of the amount of acrylate double bonds polymerized (3 mol
kg1) to the amount of photoinitiator destroyed (3 104 mol
kg1). For this specific resin, 10 000 acrylate double bonds were found to be polymerized for each photoinitiator molecule destroyed. If each Lucirin TPO molecule gene- rates 1.3 initiating radicals (see below), the value of KCL is in the order of 7 700 mol per radical. This value is surpri- singly high for a polymerization initiated at a very high rate (ri 4 103 mol
l1
s1). Indeed, for radical-induced polymerizations where termination is expected to occur by bimolecular reactions of polymer radicals, KCL should decrease as ri (or light intensity) is increased, in considera- tion of the following equations Rate of polymerization Rp ¼ kp ð2ktÞ0:5 ½Mr0:5i ð1Þ Rate of initiation ri ¼ Fi
Ia ð2Þ Kinetic chain length KCL ¼ Rp ri ¼ kp½MF 0:5 i ð2ktÞ0:5I0:5a ð3Þ where Fi is the initiation quantum yield and Ia the light intensity absorbed. The long kinetic chains observed in UV-curable acrylate resins are attributed, on the one hand, to the high value of ratio kp/(2kt) 0.5 (20 l0.5
mol0.5
s0.5)[36] and, on the other hand, to the fact that each initiating radical is surrounded by a great number of acrylate double bonds. In a typical UV-curable formulation ([M0]¼ 3 mol
l1), each photoinitiator molecule is initially surrounded by 100 acrylate double bonds. After a 100 ms UVexposure, 1% of the photoinitiator (3 104 mol
l1) has been destroyed to generate initiating radicals. Consequently, each initiating radical will be surrounded by more than 5 000 acrylate double bonds (the photoinitiator splits into 2 radical frag- ments), some of these radicals having already disappeared by reaction with the acrylate double bonds. This means that, once the polymerization has been initiated under intense illumination, there is no other choice for the polymer radi- cal than to continue to grow extensively, until it encounters ultimately another polymer radical. The situation remains the same upon further UV exposure, the additional amount of initiating radicals formed being compensated by the amount of polymer radicals getting lost upon bimolecular termination reactions (steady state conditions). The overall effectiveness of photoinitiated polymeriza- tion can be evaluated from the polymerization quantum yield, Fp, i.e. the number of acrylate double bonds polymerized per photon absorbed, by using the following equation[37] Fp ¼ ½Mpmol l1  ‘cm f  1000  ts  I0 einstein cm2s1 ð4Þ where [Mp] is the concentration of acrylate double bonds polymerized after UV exposure during time t at a light intensity I0, with f being the fraction of incident light absorbed by the sample of thickness l. For the most reactive acrylate formulations, Fp values up to 10 000 mol
einstein1 were obtained under O2-free conditions. [37,38] Precisely this large amplification factor is responsible for Scheme 2. Photoinitiation of a radical-type polymerization. Figure 2. Real-time monitoring of the polymerization of the acrylate double bond and photolysis of the initiator upon UV irradiation (20 mW
cm2) under O2-free conditions: continuous UVexposure (solid line), 0.1 s UVexposure (dashed line); [Lucirin TPO]¼ 1 wt.-%. Kinetic Study and New Applications of UV Radiation Curing 1071 the outstanding performance of UV-curable acrylate resins and accounts for the success of this technology in a large variety of applications. 3.1.2 Evaluation of Initiation Efficiency The photoinitiator (PI) plays a key role in UV-curable systems by controlling both the initiation rate and the cure depth. It is essential to select a photoinitiator showing the highest initiation efficiency and undergoing a fast photobleaching upon UV exposure in order to achieve a deep-through cure by frontal polymerization.[13,39] Figure 3 shows the UVabsorption spectra of an acylphosphine oxide photoinitiator (Lucirin TPO) in an acrylate resin,[40] before and after exposure to radiation from a medium pressure mercury lamp for various durations. Initiation efficiency can be evaluated from monomer conversion and PI decay curves recorded by means of RT-IR and RT-UV spectro- scopy, respectively. The various processes occurring upon UVexposure of the liquid resin can be represented schematically by Scheme 3. Key parameters are fraction f1 of the PI excited states that generate free radicals and fraction f2 of these radicals that react with monomer and initiate polymerization. Initiation efficiency is given by the product f1
f2 and corresponds to the number of initiating radicals produced per PI molecule destroyed. The rate of initiation can therefore be expressed as ri ¼ d½PI dt
f1
f2 ð5Þ where d[PI]/dt is the initial loss rate of photoinitiator. From the Equation (1), ri can also be expressed as a function of the rate of polymerization: ri ¼ 2kt
R2p k2p
½M 2 ð6Þ Introducing this expression in Equation (5) leads to the following equation for the initiation efficiency: f1
f2 ¼ 2kt
R2p k2p
½M 2
d½PI dt ð7Þ By measuring initial polymerization rate and PI loss rate, one can evaluate the product f1
f2 from Equation (7). Figure 4 shows polymerization profile and photoinitiator decay curve recorded upon UV exposure of a polyurethane acrylate resin ([M]0¼ 3.2 mol
l1) containing 1 wt.-% Lucirin TPO. By taking kt¼ 105 l
mol1
s1 and kp¼ 104 l
mol1
s1,[36] the initiation efficiency of this acylphosphine oxide was found to be 1.3 radical per TPO molecule, which is not so different from the maximum value of 2. Figure 3. UV absorption spectrum of Lucirin TPO (1 wt.-%) in acrylate resin before and after UV exposures of up to 5 s at a light intensity of 30 mW
cm2. Scheme 3. Photoinitiated radical polymerization. Figure 4. Decay profile of the photoinitiator and polymerization profile of an acrylate resin upon UV irradiation under O2-free conditions (30 mW
cm2). 1072 C. Decker The initial quantum yield of the photolysis of Lucirin TPO, FTPO, was determined by calculating the ratio of the initial PI loss rate to the absorbed light intensity. The fraction of incident light absorbed by the 24 mm thick sample was evaluated by differential radiometry and found to be 15%. From the FTPO value thus obtained (0.62 mol
einstein1) initiation quantum yield Fi of Luci- rin TPO could finally be evaluated to Fi¼FPI
f1
f2¼ 0.8 radical einstein1. This high value of Fi, together with fast photobleaching, are typical of the behavior of acylpho- sphine oxides. Similar results were obtained with a bis(acylphosphine) oxide (Irgacure 819) which generates 4 free radicals per molecule after absorption of 2 photons (Scheme 4).[41] The great reactivity of phosphinoyl radicals toward the acrylate double bond was considered to be mainly res- ponsible for the higher initiation efficiency of this type of compounds.[42,43] These photoinitiators have also the ad- vantage of absorbing in the near UV region (350-400 nm), where the mercury lamps have their strongest emission, and are thus particularly well suited for the UV-curing of formulations containing pigments or light stabilizers.[41,44] While fast photolysis of the initiator is beneficial for increasing the cure speed and lowering the residual PI content, it may yet lead to a premature ending of the poly- merization due to complete consumption of the photo- initiator, if its initial concentration is not high enough. Such an effect is illustrated in Figure 5 for an aromatic polyether- diacrylate (Ebecryl 600 from UCB) containing Lucirin TPO (0.1 and 1 wt.-%) as the photoinitiator. 3.1.3 Influence of the Chemical Structure of the Oligomer Different types of structures can be used for the telechelic oligomer which will constitute the framework of the UV- cured polymer: polyurethane, polyester, polyether and polysiloxane. The final properties of UV-cured acrylate polymers depend primarily on the chemical structure of the oligomer backbone, on its molecular weight (typi- cally between 500 and 2 000), as well as on the cure extent. Low-modulus elastomers are generally obtained with aliph- atic telechelic acrylate oligomers associated to a mono- acrylate reactive diluent, whereas hard and glassy materials are formed when aromatic structures are introduced into the polymer network, together with multifunctional monomers. The glass transition temperature (Tg) of the fully cured polymer has a strong influence on the final conversion reached upon UVirradiation at room temperature. TheTg of UV-cured dimethacrylate polymers was shown to increase steadily with monomer conversion, as well as with irra- diation temperature.[45,46] Figure 6 shows the polymeriza- tion profiles recorded for two types of acrylate end-capped oligomers, an aliphatic polyurethane-acrylate and an aro- matic polyether-acrylate. The chemical structure has a great effect on the final cure extent, while it does not affect resin reactivity significantly. Scheme 4. Photolysis of a bis(acylphosphine oxide). Scheme 5. Thermally induced polymerization upon heating of a polyphenoxy-acrylate UV-cured in the presence of air. Figure 5. Influence of photoinitiator concentration on the photopolymerization of a polyphenoxy-diacrylate. Light inten- sity: 100 mW
cm2, 12 mm thick laminated film. Kinetic Study and New Applications of UV Radiation Curing 1073 High degrees of conversion (close to 100%) are reached with the aliphatic polyurethane-acrylate that gives an elas- tomeric material, well suited for adhesives applications. By contrast, the aromatic polyether-acrylate yields a glassy polymer (Tg 80 8C), hard but brittle. Here polymerization stops prematurely because of vitrification and related molecular mobility restrictions. A more complete polymerization can be achieved by rising the sample temperature to 80 8C, either before or after UV exposure (Figure 7). In the latter case, the additional polymerization and hardening observed upon thermal treatment was attributed to the restored mobility of trapped polymer radicals,[47] as well as to the decomposition of previously formed hydroperoxides, which yields initiating hydroxyl and alkoxy radicals (Scheme 5).[48] Evidence in favor of this process was obtained by performing UV curing under O2-free conditions. The conversion increase obser- ved upon heating at 80 8C in the dark, due to the un-trapping of the occluded polymer radicals, was found to be half as large as the one observed in the presence of air. As expected, the more complete polymerization achieved by operating at elevated temperatures leads to an increase in Tg of the samples (up to 100 8C) and, concomitantly, to an increase in polymer hardness (Figure 8). The lower amount of remai- ning acrylate unsaturations is also beneficial with respect to long-term properties of UV-cured polymers. Another way to achieve more complete polymerization in photopolymer materials is by performing UV irradiation at ambient temperature but with higher light intensity. As the heat of the exothermal polymerization is released in a shorter period of time (less than 1 s), the temperature of the sample will rapidly rise and molecular mobility will be increased. Figure 9 shows the marked influence of light intensity on the final conversion of a UV-cured polyur- ethane-acrylate (Actilane 20). To confirm that the observed increase in conversion was indeed resulting from a rise of the sample temperature, a novel method, based on RT-IR spectroscopy, has been developed to monitor the temperature in thin films under- going high-speed photopolymerization.[36,49] The tempera- ture probe is the poly(propylene) film serving as a support, which exhibits a temperature-sensitive IR band at 842 cm1. Figure 6. Influence of the chemical structure of the diacrylate oligomer (aliphatic polyurethane or aromatic polyether) on poly- merization kinetics. Light intensity: 50 mW
cm2, [Irgacure 651]¼ 2 wt.-%, laminated film. Figure 7. Influence of temperature on the UV-curing of an aromatic polyether-diacrylate (Ebecryl 600) in the presence of air (I¼ 500 mW
cm2). Figure 8. Influence of temperature on the hardening of an aromatic polyether-diacrylate (Ebecryl 600) upon UVexposure in the presence of air (I¼ 500 mW
cm2). 1074 C. Decker Figure 10 shows some typical temperature profiles recorded at various light intensities for a UV-curable polyurethane- acrylate (PUA). Upon intense illumination, the sample temperature reaches values up to 90 8C within a fraction of a second. A linear relationship was found to exist between the rate of polymerization and the rate at which the temperature rises,[49] thus demonstrating that the increase in tempera- ture is indeed caused by the exothermal polymerization. It leads to a more complete polymerization, as shown in Figure 11 where the final acrylate conversion was plotted as a function of maximum temperature of the sample under- going polymerization. The maximum temperature reached upon UV exposure depends on the film thickness, as well as on the substrate on which the sample is coated. Indeed, no significant rise in temperature was observed upon UV-curing of a sample coated onto a silicon wafer which acts as a heat trap. By contrast, for a 300 mm thick PUA sample coated on a glass plate and cured by a short passage under a powerful UV lamp (600 mW
cm2), a thermocouple revealed that the temperature rose to 165 8C, whereas it hardly changed in the absence of photoinitiator (no curing), as shown in Figure 12. An important consequence of this behavior is that irradiation conditions (light intensity, initial temperature, film thickness, support) affect the cure extent, and therefore the physico-chemical properties (Tg) of the photopolymer material (Scheme 6). 3.1.4 Influence of Light Intensity A distinct feature of photoinitiated polymerization is that the initiation rate can be varied in a very large range by changing the light intensity. It should yet be remembered that a 100-fold increase in light intensity will lead to only a 10-fold increase in polymerization rate, because of the square-root dependence of Rp on Ia (Equation (1) and (2)). This relationship holds only in the early stages of polymeri- zation when termination occurs by bimolecular reactions between polymer radicals. As mobility restrictions increase, the crosslinked polymer radicals are less likely to meet by sole segmental diffusion. The radical site will move mainly by reacting with neighboring functional groups until it combines with another such radical. When termination is controlled by reactive diffusion,[36,50,51] the termination rate constant becomes proportional to the propagation rate Figure 9. Influence of light intensity on the final conversion of a polyurethane-diacrylate (Actilane 20) after a UV dose of 1 J
cm2. Figure 10. Temperature profiles recorded by means of RT-IR spectroscopy upon UV-curing of a polyurethane-diacrylate (Actilane 20) at different light intensities. Figure 11. Dependence of the final conversion on the maximum temperature reached for a UV-cured PUA sample. Kinetic Study and New Applications of UV Radiation Curing 1075 constant. For acrylate-based resins, it was found to occur at conversions of 20% and above.[36] Later on, as the concen- tration of acrylate double bonds in the vicinity of the radical site has decreased and gelation has occurred, the polymer chain stops growing and leaves long-lived radicals trapped within the polymer network.[45,52,53] This corresponds to a monomolecular termination process and will lead to a linear dependence of Rp on light intensity. Actually, exponent x of the kinetic lawRp¼KI x was found to increase steadily with conversion as the liquid resin is transformed into a solid polymer network (Figure 13). It shows that, as expected, radical trapping takes an increasingly important part in the termination step as UV-curing proceeds.[54] The overall rate of polymerization can be expressed, in a first approximate, as the sum of two terms which depend on the first power and on the square root of the absorbed light intensity, respectively[55] Rp ¼ a kp k0t ½MFiIa  ð1  aÞ kp ð2ktÞ0:5 ½MF0:5i I 0:5a ð8Þ where kt 0 is the rate constant of the unimolecular termi- nation process, i.e., the reciprocal of the polymer radical lifetime. Coefficient a is ranging between 0 and 1, and reflects the relative contribution of the unimolecular path- way, i.e., the probability for a given polymer radical to become occluded in the polymer matrix. Value a is strongly dependent on the structure and crosslink density of the polymer network. It will be lower for elastomeric materials than for glassy UV-cured polymers where reaches values up to 0.8.[55,56] A direct consequence of the first order variation in Rp with light intensity is that the polymerization quantum yield (Fp¼Rp/Ia) remains constant, instead of dropping with increasing Ia (reciprocity law). When those conditions are reached, i.e., toward the end of the crosslinking polymer- ization, one makes a more efficient use of the energy delive- red by the light beam, specially at high light intensities. 3.1.5 Influence of Atmospheric Oxygen Another advantage of performing the polymerization under intense illumination is to reduce UV exposure time during which atmospheric oxygen diffuses into the film and sca- venges both the initiating radicals and the polymer radicals (Scheme 7). The peroxy radicals are inactive toward the acrylate double bond and disappear by hydrogen abstraction to generate hydroperoxides and alkyl radicals, which are also scavenged by oxygen. The free radicals, produced by photolysis of the initiator during the induction period, serve Figure 12. Temperature profiles recorded with a thermocouple upon a 1.6 s UVexposure in air of a 300 mm thick PUA film, with and without photoinitiator ([Irgacure 651]¼ 2 wt.-%). Light intensity: 600 mW
cm2. Scheme 6. Various factors affecting the properties of UV-cured polymers. Figure 13. Variation of rate exponent xwith exposure time, upon UV-curing of a polyurethane-diacrylate resin at a light intensity of 10 and 100 mW
cm2 under O2-free conditions. [Lucirin TPO]¼ 2 wt.-%. 1076 C. Decker to consume the oxygen dissolved in the resin, which will become chemically bonded to the polymer in the form of hydroperoxides. The higher the light intensity, the faster will oxygen be removed by the high flux of free radicals. Polymerization of the acrylate double bond will only start once the concentration of the dissolved oxygen has de- creased sufficiently (by at least two orders of magnitude) to allow the monomer to compete successfully with oxygen for the scavenging of the initiating radicals.[57] There is a number of factors controlling the importance of the inhibitory effect of oxygen in UV-curable acrylate resins: (i) photoinitiator efficiency in producing free radi- cals, (ii) formulation viscosity, which controls the diffusion rate of oxygen, (iii) sample thickness, thin films being more difficult to cure in the presence of air (Figure 14), and (iv) light intensity, which determines the exposure time during which atmospheric O2 diffuses into the sample. For thick coatings, a deep-through cure can be achieved by low intensity UV radiation, but the surface in contact with air will remain tacky. It may happen in some limiting case, i.e., a few micron thick films of a fluid resin exposed to dim light (e.g., sunlight), that polymerization does not occur at all because the oxygen consumed by the free radicals formed is steadily replaced by the air diffusing into the sample. It is then necessary to perform UV irradiation under O2-free conditions in order to avoid wasting the few initiating radicals formed. To achieve such an inert environ- ment, nitrogen can be advantageously replaced by carbon dioxide (Figure 14), especially for the UV-curing of acrylic resins coated onto three-dimensional objects.[58] As CO2 is heavier than air, gas losses during the process are minimized when conducted in a pool-type photoreactor. 3.1.6 Photopolymerization of the Thiol/ene System There is another type of photoinitiated radical polymeriza- tion which, unlike the acrylate-based resins, is not inhibited by molecular oxygen, and which was successfully used to photopolymerize thin films in the presence of air. The UV- curable resin consists of a mixture of polyene and multi- functional thiol that undergoes copolymerization upon UV irradiation by a step-growth addition mechanism.[6] The polymerization process, which is propagated by chain- transfer reaction involving thiyl radical RS., can be repre- sented by the following set of reactions: Photoinitiator!hn ln ln þ RSH ! ln H þ RS RS þ CH2 CHR0 ! RSCH2C  HR0 RSCH2C  HR0 þ RSH ! RSCH2CH2 R0 þ RS ð9Þ A linear polymer is formed in the case of a diene/dithiol couple. For crosslinking to occur, a polyene must be asso- ciated to a multifunctional thiol. In the presence of air, the alkyl radical will be scavenged by oxygen to yield a peroxy radical, which is also able to propagate the chain reaction by hydrogen abstraction to yield a hydroperoxide: ð10Þ Figure 15 shows some typical conversion versus time profiles recorded by means of RT-IR spectroscopy (vinyl band at 1638 cm1) for a stoichiometric mixture of a tetraene and a tetrathiol exposed to UV radiation in the presence of a hydroxyphenyl ketone ([Irgacure 184]¼ 2 wt.-%). In such thin films (20 mm), similar polymerization profiles were obtained in the presence of air and under O2- free conditions. The low sensitivity to oxygen of the polyene/thiol system can be made even more apparent by stopping UV exposure, and hence the production of free radical, at an early stage of the reaction, e.g., at 20% conversion. After an 0.1 s exposure, polymerization was Scheme 7. Inhibitory effect of oxygen on photoinitiated radical polymerization. Figure 14. Inhibitory effect of oxygen on the photopolymeriza- tion of a polyurethane-acrylate resin UV-irradiated at a light intensity of 30 mW
cm2. [Irgacure 2959]¼ 3 wt.-%. Kinetic Study and New Applications of UV Radiation Curing 1077 found to continue to proceed in the dark up to 50% conversion, even in the presence of air (Figure 15). A quite different behavior was observed with an acrylate resin showing a similar reactivity, where dark polymerization was found to hardly occur in the presence of air, because of an efficient scavenging of the growing polymer radicals by oxygen diffusing into the thin film. In spite of its outstanding reactivity performance, the polyene/thiol system is not used in UV radiation curing as widely as acrylate-based resins. This is partly due to higher costs and unpleasant odor of the formulation, the limited choice of monomers, and the presence of hydroperoxides in the UV-cured polymer, which may affect its long-term stability. This is not the case for acrylate coatings UV-cured in the presence of air, because polymerization starts to proceed only when the sample has been essentially depleted of the dissolved oxygen.[57] Indeed, UV-cured aliphatic PUA coatings proved to be quite resistant to accelerated weathering and were successfully used to increase the light stability of polymeric materials.[59] It should be noted that monomers that polymerize by a cationic mechanism, such as epoxides or vinyl ethers, are not sensitive at all to the presence of atmospheric oxygen. Indeed, the photogenerated proton acid and the propagating polymer carbocations are not reacting with molecular oxygen. Tack-free coatings have thus been obtained upon UV irradiation in the presence of air, even by operating at very low light intensity, in particular with sunlight.[60] 3.2 Photoinitiated Cationic Polymerization While most of the UV-curable resins currently used are based on free-radical polymerization, the less utilized cationic polymerization of multifunctional monomers and oligomers bearing vinyl ether or epoxy groups still present a number of advantages: (i) lack of inhibition by atmospheric oxygen, (ii) post-polymerization in the dark, (iii) low shrin- kage, (iv) high mechanical performance of the UV-cured material, and (v) good adhesion onto various substrates, in particular metals. One of the main limitations to a greater industrial development of UV-curable epoxy resins lies in their relatively low reactivity and the limited choice of com- mercially available monomers and telechelic oligomers. The photoinitiated cationic polymerization of epoxi- des has been thoroughly investigated over the past 20 years,[7,15,33,61-70] in particular by Crivello and co-workers, who pioneered this technology by developing triarylsulfo- nium and diaryliodonium photoinitiators, as well as appro- priate photosensitizers.[71] Most of these studies were focused on cycloaliphatic epoxides which are the most utilized monomers because of their superior reactivity. The addition of a comonomer was shown to speed up the UV- curing process and to lead to a more complete polymeriza- tion, especially for aromatic diepoxides.[72] Promising results have been obtained with an epoxidized soyabean oil (ESO; see below),[73] a cheap compound which acts as reactive diluent, while imparting some flexibility to the UV- cured epoxy coating at the same time.[74] We have studied the photoinitiated cationic polymerization of aromatic and cycloaliphatic diepoxides, neat and in blends with ESO, vinyl ethers or acrylate monomers, by monitoring both the epoxy group consumption and the insolubilization and hardening of the polymer upon UV irradiation. 3.2.1 UV Radiation Curing of Aromatic Diepoxides (ADE) Diglycidyl ethers are poorly reactive monomers which undergo cationic polymerization at a much lower rate than vinyl ethers. Under intense illumination in the presence of an aryliodonium salt, the liquid resin can still be transfor- med within seconds into a hard and insoluble polymer. The crosslinking polymerization process of aromatic diepox- ides (ADE) can be formally written as shown in Scheme 8. Ring-opening polymerization can be quantitatively follo- wed by infrared spectroscopy through either the decrease of the characteristic band of the oxirane at 865 cm1, or the increase of the ether band at 1080 cm1. Figure 16 shows some typical kinetic curves obtained upon intense UV irradiation of a diglycidyl ether derivative of bisphenol A Figure 15. Lack of oxygen inhibition on the photopolymeriza- tion of the thiol/ene system ([Irgacure 184]¼ 2 wt.-%): O2-free conditions (solid line), in the presence of air (dashed line). Light intensity: 50 mW
cm2, film thickness: 20 mm. 1078 C. Decker (Araldite GY-250 from Ciba SC) in the presence of a triarylsulfonium salt (Cyracure UVI-6990). As polymer- ization proceeds, the polymer formed becomes increasingly insoluble in organic solvents (e.g. chloroform); the gel fraction reaches 80% after 3 s of UV exposure (Figure 17). At the same time, the degree of swelling decreases steadily with exposure time, down to a value of about two (the amount of solvent absorbed is twice the amount of dry polymer), which indicates the formation of a rather loose three-dimensional polymer network. A faster and more complete polymerization was achieved by the addition of epoxidized soyabean oil ([ESO]¼ 20 wt.- %), a comonomer which reduces the resin viscosity and will be incorporated into the polymer network, as shown by the decrease of its epoxy IR band at 825 cm1 (Figure 16). Because it contains three epoxy groups, ESO is working as an efficient crosslinker and accelerates the insolubilization of the irradiated sample, with formation of a tighter polymer network, as shown by the lower swelling ratio (Figure 17). An additional advantage of using epoxidized soyabean oil is to reduce the brittleness of the UV-cured epoxy polymer. By combining hardness and flexibility, this material proved to be quite resistant to both scratching and shocks. A distinct characteristic of cationically initiated poly- merization is that the propagating carbocations or oxonium ions are not reacting among themselves. Consequently, once initiated by a short UV exposure, polymerization will continue to proceed in the dark for some time. After 2 h of storage in the dark, the epoxy conversion of an ADE sample UV-irradiated for 1 s was found to increase from 68% to 95%, with a concomitant increase in insoluble fraction and polymer hardness. Therefore, by contrast to acrylate-based polymers, the properties of cured epoxy polymer emerging from the UV oven continue to change steadily until they stabilize, usually after a 1 d storage. 3.2.2 UV Radiation Curing of Bis(cycloaliphatic) Diepoxides Cycloaliphatic diepoxides (BCDE) are the most widely used monomers in photoinitiated cationic polymerization (Scheme 9) because of their high reactivity, attributed to the strain of the cyclohexane ring adjacent to the oxirane ring.[7,15] Their photopolymerization can be easily followed through the decrease of the IR band at 795 cm1 of the epoxy ring.[33] The photopolymerization of a typical BCDE (Araldite CY-179 from Ciba SC) was shown to proceed more rapidly Scheme 8. Photoinitiated cationicpolymerizationofa diepoxide. Figure 16. Decay of the epoxy ring and build-up of the ether group upon UV-curing of an aromatic diepoxide (ADE) and an epoxidized soyabean oil (ESO). Triarylsulfonium (TAS) PF6  photoinitiator: [TAS]¼ 3 wt.-%, light intensity: I¼ 600 mW
cm2. Figure 17. Influence of ESO (20 wt.-%) on the insolubilization of an aromatic diepoxide (ADE) upon UV exposure. [TAS]¼ 3 wt.-%; I¼ 600 mW
cm2. Kinetic Study and New Applications of UV Radiation Curing 1079 than that of ADE upon intense illumination in the presence of a triarylsulfonium salt (Figure 18). A 0.1 s UV exposure proved to be sufficient to polymerize 60% of the epoxy groups. Upon further exposure, the degree of conversion increased slowly to finally level off at a value around 80%, because of vitrification and related mobility restrictions. As a result of the high epoxy content of this BCDE ([epoxy]¼ 8 mol
kg1), insolubilization occurs rapidly upon UV exposure, with formation of a tight polymer network (swelling ratio of 0.3), as shown in Figure 19. The addition of the less reactive epoxidized soyabean oil (20 wt.-%) has hardly any effect on the cure kinetics (Figure 18), while it is slowing down the insolubilization process and generates a looser polymer network, as expected (Figure 19). The photoinitiated polymerization of BCDE was shown to proceed more extensively under high-humidity condi- tions, to yield a polymer containing no residual epoxy groups.[72] While the addition of water to the formulation will stop the polymerization by nucleophilic attack on the oxonium cation, small amounts of water may enhance the epoxide consumption through a chain transfer reaction leading to re-initiation (Scheme 10). The beneficial effect of moisture can also be attributed to its role as plasticizer that allows the trapped polymeric oxonium ions to react further. While insolubilization occurs faster under high-humidity conditions, the cured polymer is not as hard as when it is cured under normal conditions (35% relative humidity): Persoz hardness of 160 s instead of 290 s.[72] The formation of a more elastomeric polymer material under humid conditions was attributed to the moisture-driven chain-transfer reaction. The addition of a vinyl ether (VE) comonomer, such as the divinyl ether of triethylene glycol (DVE-3), known for its high reactivity, has a marked accelerating effect on the polymerization of the epoxide, as shown in Figure 20 for an equimolar mixture of BCDE and DVE-3. As much as 90% of the epoxy and vinyl ether groups were polymerized after 0.1 s of UV exposure, complete polymerization being achieved within 10 min upon storage in the dark of the irradiated sample.[75] A kinetic analysis by means of RT-IR spectroscopy of the UV-curing of a BCDE/DVE-3 blend reveals that the two monomers exhibit very similar polymerization profiles (Figure 21), which is in favor of a copolymerization mech- anism. Polymerization could proceed via a cross-propaga- tion mechanism (Scheme 11), with the oxonium ion Scheme 9. Photopolymerization of a dicycloepoxide. Figure 18. Influence of ESO (20 wt.-%) on the photoinitia- ted cationic polymerization of a bis(cycloaliphatic) diepoxide (Araldite CY-179). [TAS]¼ 3 wt.-%, I¼ 600 mW
cm2. Figure 19. Influence of ESO (20 wt.-%) on the insolubilization of a BCDE resin (Araldite CY-179). [TAS]¼ 3 wt.-%, I¼ 600 mW
cm2. Scheme 10. Moisture-driven chain-transfer reaction. 1080 C. Decker reacting with the VE double bond and the VE carbocation with the epoxy ring. Experimental evidence in favor of such a copolymeriza- tion mechanism has been recently obtained by Kostanski et al.,[76] who studied the photoinitiated copolymerization of this BCDE with a monovinyl ether. The UV-cured polymer was found to exhibit a single Tg, the value of which being located between the Tg values of the two homopoly- mers. Moreover, it was completely insoluble, thus showing that the soluble and extractable homopolymer was not formed. The addition of BCDE to DVE-3 causes a drastic slowing down of the photopolymerization (Figure 21). This effect is attributed to the conversion of the alkoxy- carbenium ion into the less reactive oxonium ion.[77] 3.2.3 Ultrafast Synthesis of Interpenetrating Polymer Networks (IPN) by UV Radiation Curing Photoinitiated polymerization of a mixture of difunctional monomers reacting by different mechanisms is one of the most efficient ways to rapidly generate IPNs. This method was successfully used with mixtures of multifunctional acrylate and epoxy monomers that were exposed to UV radiation in the presence of both radical- and cationic-type photoinitiators.[72,78] By combining the physico-chemical characteristics of the two types of polymer networks, UV- curing of IPNs appears to be a powerful tool to achieve tailor-made properties, in particular to produce hard and flexible polymeric materials, well suited for coating ap- plications. Figure 22 shows the polymerization profiles of the two types of monomers for a 50:50 mixture by weight of hexanediol diacrylate (HDDA) and ESO under intense illumination (600 mW
cm2) in the presence of air. The two polymer networks are produced simultaneously, as photolysis of the two initiators (hydroxyphenyl ketone, Figure 20. Influence of a divinyl ether ([DVE-3]¼ 41 wt.-%) on the UV-curing of a bis(cycloaliphatic) diepoxide (Araldite CY- 179). [TAS]¼ 4 wt.-%, I¼ 600 mW
cm2. Figure 21. Photoinitiated cationic polymerization of an equi- molar mixture consisting of divinyl ether ([DVE-3]¼ 41 wt.-%) and bis(cycloaliphatic) diepoxide ([Araldite CY-179]¼ 55 wt.- %), [TAS]¼ 4 wt.-%, I¼ 600 mW
cm2. Scheme 11. Copolymerization of a vinyl ether and a cycloep- oxide. Kinetic Study and New Applications of UV Radiation Curing 1081 triarylsulfonium salt) generates both free radicals and pro- ton species, respectively. The diacrylate monomer yet polymerizes more rapidly than the triepoxide due to its higher reactivity. The polymerization is also more complete than in neat HDDA (98% final conversion instead of 85%), because of the plasticizing effect of the less reactive ESO comonomer. Similar results have been obtained by UV irradiation of blends of HDDA with an aromatic diepoxide or a bisphenol-A-based diacrylate with BCDE,[78] by using a newly developed diaryliodonium hexafluorophosphate salt as a cationic photoinitiator that undergoes photolysis without the formation of benzene.[20] It should be ment- ioned that the use of such hybrid formulation is an easy way to overcome oxygen inhibition in the UV curing of thin films or under dim-light conditions. While the radical poly- merization of the acrylate monomer is initially inhibited by atmospheric oxygen, the cationic polymerization of the epoxide proceeds effectively upon UVexposure. The build- up of the epoxy network leads to a strong increase in viscosity, which slows down the diffusion of atmospheric O2 into the sample, thus allowing acrylate polymerization to start. An acrylate/epoxide hybrid system has thus been thoroughly cured within minutes by simple exposure to sunlight. A unique advantage of photoinitiation for the synthesis of IPNs is to allow a sequential build-up of the two polymer networks, one after the other, by a proper selection of both photoinitiators and radiation wavelengths. The acrylate/ epoxide blend is first exposed to filtered UV radiation (> 350 nm) which is absorbed only by the acylphosphine oxide photoinitiator. The free radicals thus generated will initiate solely the polymerization of the acrylate monomer, while the photoinitiator is bleached out (Figure 23a). Here again, the presence of the unreacted liquid epoxy monomer allows a more complete and rapid polymerization of the diacrylate monomer. In a second step, the sample is exposed to unfiltered UV radiation which is absorbed by the triaryl- sulfonium salt, thus promoting the cationic polymerization of the epoxide (Figure 23b). As the polymerization of the epoxy monomer occurs within the already formed acrylate polymer network, it proceeds slower and less extensively than in the neat epoxide. This drawback will be partly alleviated by the post-polymerization upon storage in the dark, which can be further enhanced by warming the UV- cured sample. Owing to their performance regarding both processing and properties, UV-cured acrylate/epoxide IPNs are expec- ted to find their main applications as fast-drying protective coatings and for the rapid manufacturing of composite materials and optical components. 3.3 Photocrosslinking of Functionalized Polymers Besides transforming a liquid resin into a solid polymer, UV radiation can also be used to induce the polymerization of Figure 22. Photoinitiated polymerization of a 50:50 blend of ESO with hexanediol diacrylate. [TAS]¼ 1 wt.-%, [Darocur 1173]¼ 1 wt.-%; I¼ 600 mW
cm2. Figure 23. Two-step UV curing of an epoxide/acrylate hybrid upon exposure to filtered (a) and unfiltered (b) light. Light intensity: 30 mW
cm2. 1082 C. Decker reactive groups located on a polymeric chain and thus achieve a rapid insolubilization, as well as modify the physical and mechanical properties of functionalized poly- mers. The high initiation rate provided by intense illumina- tion is beneficial in speeding up the reaction, which still proceeds much slower in the solid state than in liquid resins. UV technology could be successfully applied to crosslink waterborne urethane-acrylate coatings, functionalized polyisoprene rubbers, and polybutadiene-type thermoplas- tic elastomers within seconds. 3.3.1 UV Curing of Water-Based Urethane/Acrylate Coatings The potential of water-based systems and their performance in UV radiation curing have been thoroughly investigated in the past decade.[23,79-84] Most of them consist of a disper- sion or emulsion of acrylate functionalized oligomers that can be photocrosslinked in the solid state after the removal of water. The advantages offered by these environmental friendly systems are partly offset by the required drying step that increase the overall processing time. The water-dispersible polyurethane-acrylate (PUA) oli- gomer was synthesized by reaction of a diisocyanate with a mixture of a telechelic diol oligomer, dimethylol propionic acid, and a hydroxy-acrylate monomer.[84] The carboxylic acid groups were neutralized by an amine, and water was added to form an aqueous dispersion with a 35 wt.-% solid content. The formulation containing a soluble photoinitia- tor ([Irgacure 2959]¼ 0.3 wt.-%) was cast onto a BaF2 crystal so as to obtain a 20 mm thick dry film after water removal. The photopolymerization of the acrylate end groups, monitored by IR spectroscopy, proceeds relatively slowly at ambient temperature and is leveling off at 50% conversion because of severe mobility restrictions of the reactive species (Figure 24). Both polymerization rate and final cure extent were substantially increased by rising the sample temperature to 80 8C, and even more so by adding a diacrylate monomer ([HDDA]¼ 10 wt.-%) to act as a reactive plasticizer (Figure 24). For maximum efficiency, the UVexposure has therefore to be performed immediately after the drying step on the hot sample emerging from the infrared oven. It should be mentioned that the UV-curing of waterborne PUA coatings is not inhibited significantly by atmospheric oxygen, because of the very slow diffusion of air into the solid film.[85] The polymer obtained after UV curing is highly cross- linked, as shown by the schematic 3D representation of Figure 25 and therefore totally insoluble in organic solvents. The crosslink density, typically on the order of 1 mol
kg1, can be easily controlled through the molecular weight of the telechelic PUA oligomer. The build-up of numerous chemi- cal links between the polymer chains makes the coating harder and more resistant to scratching. Figure 26 shows how the pendulum Persoz hardness of the PUA coating increases upon intense illumination to reach values up to 350 s, which are almost as high as for mineral glass. One of the distinct advantages of water-based UV-cured polyur- ethane acrylate coatings is that they combine hardness and flexibility,[23] thus making such materials quite resistant to abrasion and shock (Figure 27). 3.3.2 Photocrosslinking of Functionalized Polyisoprene It is difficult to crosslink polyisoprene by UV irradiation because of the poor reactivity of the amylene double bond. By transforming this function into an epoxy group or acry- late double bond, a great improvement in reactivity was achieved, thus allowing such functionalized polyisoprene Figure 24. Influence of the temperature and of HDDA (10 wt.- %) on the UV-curing of a waterborne PUA coating. [Irgacure 2959]¼ 1 wt.-%, I¼ 500 mW
cm2. Figure 25. Three-dimensional polymer network formed by photopolymerization of a polyurethane-diacrylate. Kinetic Study and New Applications of UV Radiation Curing 1083 to be crosslinked at ambient temperature by a short UV exposure in the presence of an adequate photoinitiator. Epoxidized polyisoprene (EPI) was obtained by treat- ment of natural rubber by peracetic acid for 6 h at 5 8C (Scheme 12).[86] The ring-opening polymerization of the epoxy groups proceeds effectively upon UV irradiation of EPI in the presence of a triarylsulfonium salt, with the formation of ether crosslinks.[87,88] As expected, the photo- initiated cationic polymerization continues to proceed in the dark, so that all epoxy groups had reacted after a 0.5 s UV exposure and a 1 d storage in the dark, as shown in Figure 28. At the same time, the UV-irradiated rubber be- comes harder and insoluble in organic solvents (Figure 29). The high value of the swelling ratio (SR¼ 8) indicates that a very loose polymer network has been formed, most prob- ably because a majority of the epoxy groups tend to polymerize with neighboring groups located on the same polymer chain. The fact that complete insolubilization was achieved only when 60% of the epoxy groups had polyme- rized (73 epoxy groups per chain) argues in favor of such intramolecular process. Its importance was found to rise with the epoxidation extent of EPI, as expected.[87] It is therefore possible to achieve photocrosslinking with a func- tionalized polyisoprene containing only a few epoxy groups, since the intermolecular process is then predominant. Polyisoprene can also be readily photocrosslinked by introducing pendant acrylate double bonds on the polymer backbone, simply by treating epoxidized polyisoprene by acrylic acid (Scheme 13). Acrylated polysioprene (API) was transformed into a hard and insoluble material within a few seconds upon intense illumination in the presence of a hydroxyphenyl ketone photoinitiator.[89] The in-chain amylene double bonds of polyisoprene were shown to copolymerize with the Figure 26. Insolubilization and hardening of a waterborne PUA coating upon UV exposure at ambient temperature. I¼ 500 mW
cm2. Figure 27. Correlation between hardness and flexibility in polymeric materials. Performance of a UV-cured waterborne PUA coating. Scheme 12. Epoxidation of polyisoprene. Figure 28. Photocrosslinking of epoxidized polyisoprene. Influ- ence of storage in the dark on cationic post-polymerization of the epoxy group. I¼ 500 mW
cm2; [TAS]¼ 3 wt.-%. 1084 C. Decker acrylate double bonds, thus leading to the formation of a relatively tight polymer network (SR¼ 1 for [acrylate]0¼ 4.8 mol
kg1).[90] The crosslinking reaction can be markedly accelerated by introducing in the solid API film a di- or triacrylate monomer that acts as a reactive plas- ticizer. UV exposure of 0.2 s proved to be sufficient to transform the low-modulus elastomer into a high-modulus thermoset polymer which is very hard (Persoz hard- ness¼ 330 s), but still flexible (Figure 30). Such perfor- mance was achieved immediately after UV irradiation, as there is no dark post-cure in radical-type polymerizations. By using monomers like epoxides that polymerize by a cationic mechanism as reactive plasticizers, two IPNs have been formed simultaneously upon UVirradiation in the pre- sence of both radical and cationic photoinitiators. A similar IPN has been obtained by photocrosslinking of an epoxidized polyisoprene plasticized with a diacrylate monomer (HDDA).[91] By changing the relative concentra- tion of the monomer and its chemical nature, as well as the degree of functionalization of polyisoprene, it is possible to modify the properties of the photocured elastomer to make it well-suited for given applications (adhesives, protective coatings, composite materials). Owing to their high reac- tivity, such monomer-plasticized functionalized polyiso- prenes were readily crosslinked by a few minute exposure to sunlight,[88,90] which makes this environmentally benign technology (solvent-free, energy-cost-free) particularly attractive for the curing of large-dimension rubber-made materials. 3.3.3 Photocrosslinking of Thermoplastic Elastomers Thermoplastic polymer networks combine the properties of thermoplastic materials and elastomers. They are typically made of block copolymers containing glassy domains (e.g. polystyrene) and elastomeric domains (e.g. polybutadiene) which tend to segregate to form a two-phase morphol- ogy.[92] Such polymers exhibit a high extensibility and resilience, but are still soluble in hydrocarbons and start to flow upon heating above the Tg of the glassy polymer moiety (80 8C). The physical network can be strength- ened by creation of a chemical network through permanent covalent bonds joining together the polymeric chains of the elastomeric phase, thus making the material insoluble and more resistant to heating. Such a crosslinking process was effectively performed in polybutadiene-based thermo- plastic elastomers (styrene/butadiene,[93,94] acrylonitrile/ butadiene (ABA)[95]) through a short UV exposure in the presence of a radical-type photoinitiator. The shear adhe- sion failure temperature was found to increase from 80 8C to over 160 8C upon UV-curing, while the polymer became insoluble in organic solvents, as a result of the intermolecular reaction between the polybutadiene vinyl double bonds.[95] Again, photocrosslinking was greatly accelerated by the addition of a triacrylate monomer (TMPTA) that increases the molecular mobility in the solid polymer and copoly- merizes with the polybutadiene unsaturations. Figure 31 Figure 29. Insolubilization and hardening of an epoxidized polyisoprene exposed to UV radiation (measurements were made after 20 min of storage). I¼ 500 mW
cm2, [TAS]¼ 3 wt.-%. Scheme 13. Acrylation of epoxidized polyisoprene. Figure 30. Influence of a triacrylate monomer ([TMPTA]¼ 50 wt.-%) on the photocrosslinking and hardening of an acrylated polyisoprene. [Lucirin TPO]¼ 5 wt.-%, I¼ 500 mW
cm2. Kinetic Study and New Applications of UV Radiation Curing 1085 shows some typical insolubilization and acrylate conver- sion profiles obtained upon UVexposure of an ABA triblock copolymer containing 20 wt.-% of TMPTA. The UV-irra- diated rubber becomes totally insoluble in toluene within less than 1 s, with formation of a tight polymer network (swelling ratio of 0.5). Similar results were obtained with styrene/butadiene copolymers.[93] An equally important accelerating effect was observed by using small amounts (2 wt.-%) of a trifunctional thiol crosslinker (TRIS) which undergoes a step-growth addition polymerization with the alkene double bonds of poly- butadiene (Scheme 14). After a 0.1 s UVexposure of ABA in the presence of a phosphine oxide photoinitiator ([Irgacure 819]¼ 1 wt.-%), 6% of the butene double bonds had already disappeared (Figure 32), compared to less than 1% in the absence of TRIS. After this initial sharp drop, the remaining unsaturations are hardly reacting anymore because essen- tially all thiol molecules have been consumed.[94] As expec- ted, such an efficient crosslinking process leads to rapid insolubilization of the ABA copolymer, with a moderate hardening of the rubber, as shown in Figure 32. The fact that the elastomeric character was retained is a distinct advan- tage of the ABA/TRIS system, compared to the ABA/ TMPTA or SBS/TRIS systems. The soft and highly flexible polymer thus obtained is particularly well suited for the manufacturing of hot-melt adhesives and sealants, safety glasses and flexible printing plates. To illustrate the superior performance of the thiol/ene system for achieving efficient crosslinking of polybuta- diene-based rubbers, Figure 33 shows the insolubilization profiles of ABA, ABAþHDDA (20 wt.-%) and ABAþTRIS (2 wt.-%) samples, which were exposed to UV radiation under the same conditions (I¼ 500 mW
cm2, [Irgacure 819]¼ 1 wt.-%). After 0.8 s of exposure, gel fraction values were measured to be 22%, 83% and 98%, respectively. The greater efficiency of the thiol/ene system can be accounted for by molecular mobility and mechanistic considerations. In a conventional polymerization where polymer radicals Figure 31. Photocrosslinking of an 80:20 mixture by weight of ABA and TMPTA. [Irgacure 819]¼ 1 wt.-%, I¼ 500 mW
cm2. Scheme 14. Reaction scheme of thiol/ene polymerization of the ABA/TRIS system. Figure 32. Photocrosslinking of an acrylonitrile/butadiene rub- ber in the presence of a trifunctional thiol ([TRIS]¼ 2 wt.-%). [Irgacure 819]¼ 1 wt.-%, I¼ 500 mW
cm2. 1086 C. Decker are reacting with a close-by double bond, mobility restric- tions affect drastically the growing of the polymer chain in a solid medium. By contrast, in a step-growth addition poly- merization, the coupling reaction between polymer radical and thiol is regenerating a mobile thiyl radical that propa- gates chain reaction more easily (Scheme 14). Because of its process facility and efficiency, UV techno- logy has been revealed as a powerful tool to crosslink rapidly thermoplastic elastomers and modify, selectively in the illuminated areas, their physico-chemical characteris- tics. The main applications of these photocurable rubbers are expected to appear in sectors where cure speed and spatial control are a major concern, such as in flexography and photolithography. 4 Applications of UV-Cured Polymers UV radiation curing has found its major openings in various industrial applications where its distinct advantages (fast cure, solvent-free formulation, selective cure) have allowed this environmentally friendly technology to outclass more conventional processing techniques. The most important applications of UV-cured polymers are to be found in the following areas:[11] (i) graphic arts, with the manufacturing of printing plates and ultrafast drying of printing inks and overprint varnishes, (ii) coating industry, to protect all kinds of materials (wood, paper, plastics, metals, optical fibers), (iii) microelectronics, where UV-curable resins serve as photoresists in the imaging step, but also as fast-drying adhesives and conformal coatings, and (iv) bonding indus- try, where solvent-free UV glues allow precise control of the setting time and faster manufacturing of pressure- sensitive adhesives and release coatings. In all of these applications, the physico-chemical pro- perties of the UV-cured material have to meet precise specifications imposed for the considered end-use. This is usually achieved by designing properly both resin formula- tion and processing conditions (light intensity, exposure time, temperature, atmosphere). The mechanical properties of such photopolymers depend primarily on the chemical structure, functionality and concentration of the various constituents of the resin, as well as on the cure extent. Cross- linked polymers containing aliphatic polyurethane chains show usually an elastomeric character and are therefore well suited for adhesives applications and surface protec- tion of flexible supports. By contrast, UV-cured coatings containing polyphenoxy chains are hard and glassy ma- terials that are more appropriate for the treatment of rigid substrates requiring a good resistance to scratching. In this last section, some potential new applications of the UV- curing technology are reviewed, in particular to produce weathering-resistant protective coatings, laser-cured mate- rials, glass laminates and nanocomposite materials. 4.1 UV-Cured Protective Coatings A common characteristic of UV-cured polymers is that they are strictly insoluble in organic solvents and quite resistant to most chemicals. Because of their high crosslink density that can reach up to 5 mol/kg, they undergo hardly any swelling in good solvents like chloroform or tetrahydro- furan and are therefore well suited to protect polymer materials, such as polycarbonates or poly(vinyl chloride) (PVC). By covering both faces of a 2 mm thick PVC plate with a UV-cured coating made of a polyester hexaacrylate (Ebecryl 830), the thermoplastic polymer was rendered perfectly resistant to tetrahydrofuran, as shown in Figure 34. This type of coating proved to be also resistant to strong acids like concentrated HCl, H2SO4 and HNO3. It was successfully used to protect a glass/epoxy composite against hydrofluoroboric acid, which otherwise destroys rapidly the uncoated material. Acrylate-based UV-cured coatings can be made more resistant to staining by using multifunctional monomers like trimethylolpropane triacry- late or pentaerythritol tetraacrylate as reactive diluents.[96] Specially designed UV-cured polyurethane-acrylate polymers (Actilane 20, Ebecryl 284) have been shown to be very resistant to photodegradation when exposed to accelerated weathering.[97,98] The high crosslink density of the photopolymer is favoring cage recombination of the primary radicals formed by UV-irradiation, while it reduces at the same time the extent of chain peroxidation because of the restricted molecular mobility. The residual photoini- tiator is destroyed within minutes by solar radiation and has therefore no detrimental effect on the long-term light stability of the photocrosslinked polymer. Figure 33. Influence of thiol and acrylate monomers on the pho- tocrosslinking of an ABA rubber. [TRIS]¼ 2 wt.-, [HDDA]¼ 20 wt.-%, [Irgacure 819]¼ 1 wt.-%, I¼ 500 mW
cm2. Kinetic Study and New Applications of UV Radiation Curing 1087 The weathering resistance of various organic materials used in outdoor applications (plastics, wood, painted metal) has been greatly improved by means of UV-cured polyur- ethane-acrylate coatings containing light stabilizers.[98-101] The main role of UV absorbers (hydroxyphenylbenzotria- zoles or hydroxyphenyltriazines) is to screen the harmful UV-A and UV-B radiation of sunlight to avoid photodegrada- tionof thecoatedorganic substrate.HALSradical scavengers are also needed to increase the light stability of the coating, as well as the lifetime of the UV absorber.[102,103] The photo- bleaching of a red painted metal exposed to accelerated QUV-A weathering has been effectively prevented by covering its surface with a 50 mm thick UV-cured PUA film (Ebecryl 284þHDDA) containing both types of light stabi- lizers (1 wt.-% each), as shown in Figure 35. After a 5000 h exposure, the coating remained clear and glossy, while one could hardly detect any change in the paint color. This photostabilization process was also successfully applied to prevent the discoloration of transparent PVC plates and wood panels.[98,99] When properly stabilized, photocros- slinked polyurethane-acrylate coatings were found to out- perform the thermosetting melamine-acrylate clearcoats currently used as automotive finishes, with respect to their resistance to weathering, abrasion and scratching.[44,101] 4.2 Laser-Induced Curing In the continuous search for faster photoprocessing and higher image definition, lasers appear to represent the ulti- mate light source to achieve ultrafast polymerization and micronic resolution, because of the great power, concen- trated in the collimated laser beam, which can be focused sharply. Laser-induced polymerization has been applied to different types of multifunctional monomers and oligomers (acrylates, vinyl ethers, epoxides) to produce highly crosslinked polymer materials[32,55,104-106] (for a review cf. to the literature[107]). This advanced technology has found new applications in various industrial sectors, where the improved performance has outweighed the higher cost of laser processing, in particular in photolithography,[108] stereolithography,[109] and holography.[110] In the manufacturing of printing plates or printed circuit boards, the required high resolution patterns can be gene- rated onto a photosensitive resin by laser direct imaging (LDI) without the use of a costly mask.[105,108] With nega- tive photoresists, the illuminated area becomes insoluble and a relief image is obtained after development. By using the highly reactive photocurable acrylate resins described previously, writing speeds up to 100 m/s were reached by focusing a 100 mW Krypton ion UV laser beam down to a 10 mm wide spot, corresponding to an exposure time of 0.1 ms.[106] The LDI technology offers a number of advan- tages when compared to conventional photo-tool imaging technology, such as sharper and well-defined lines, fewer defects, rapid prototyping and process flexibility. Figure 36 shows the insolubilization profiles obtained with a polyester amino-tetraacrylate resin (Ebecryl 80 from UCB) by plotting the thickness of the insoluble polymer films formed as a function of the fluence (or dose) expressed in mJ
cm2. When the photoresist was exposed to UV emission at 337.4 nm of a Krþ laser beam, a UV dose as low as 0.4 mJ
cm2 proved to be already sufficient to achieve insolubilization in the presence of a morpholino ketone photoinitiator ([Irgacure 369]¼ 1 wt.-%). Such a high Figure 34. Influence of a 50 mm thick UV-cured PUA coating on the dissolution profile of a PVC plate in tetrahydrofuran. Figure 35. Photobleaching of a red paint upon exposure to accelerated QUV-A weathering. Performance of a 50 mm thick UV-cured coating containing light stabilizers (1 wt.-%Tinuvin 292 and 1 wt.-% Tinuvin 400) 1088 C. Decker sensitivity is due to the large chemical amplification factor resulting from the chain reaction (KCL 104 mol radical1). Since most of the continuous wave lasers have their strongest emission in the visible range, there has been a growing demand for photocurable polymers which would be sensitive beyond 400 nm radiation. The sensitivity of visible-type photoresists is not as great as that of UV- photoresists because of the lower energy of visible photons and of less efficiency of the suited photoinitiators. For the blue diode laser emitting at 407 nm, bis(acylphosphine) oxide photoinitiators can still be used. With the emission at 488 nm of the argon ion laser, the best results were obtained with a combination of three photoinitiators: xanthone dye (eosin), chloromethyltriazine and aryltitanocene (Figure 36). The photosensitivity of this complex system was still lagging behind that of the morpholino-ketone- based UV photoresist.[105] In a different type of application, laser-curable resins have been successfully used to connect optical fibers. The tips of each fiber, coated by a PUA resin, were put in contact. Bonding was achieved instantly by simply firing the laser through either one of the optical fibers. The steady interest in laser-induced polymerization pro- cesses is expected to grow further, as the potential of this technology becomes more apparent with the development of better performing photosensitive systems and the manufacturing of reliable low-cost lasers and light-emitting diodes. 4.3 UV-Cured Glass Laminates Radiation-curable adhesives are now well-established pro- ducts because of their distinct characteristics: a quick setting of solvent-free glue in well-defined areas, with a perfect control of setting time. The latter factor has resulted in the development of light-curable resins to achieve an accurate assembly of optical components. Other applications of UV- curable adhesives have appeared in various sectors, includ- ing release coatings, sealants, pressure-sensitive adhesives, conformal coatings and hot-melt adhesives.[111] An important limitation of the use of UV-curable glues is that at least one of the two parts of the assembly must be transparent to UV radiation. The two elements to be bonded can be identical or of a different nature, and consist of either flexible films or rigid transparent plates. Safety glasses have been readily produced by photocuring at ambient tempera- ture of an acrylic intercalate consisting either of a liquid resin or a solid thermoplastic functionalized polymer.[112] As the UV radiations reaching the resin are filtered by the glass plate, it is recommended to use a photoinitiator which absorbs above 330 nm, like acylphosphine oxides. Figure 37 shows the polymerization profiles recorded with a lami- nated PUA sample exposed to glass-filtered light of a mercury lamp, using Lucirin TPO or Irgacure 184 as the photoinitiator (2 wt.-%). The concentration of photoinitia- tor has to be kept as low as possible (0.1 wt.-% for 2 mm thick intercalates) to achieve a deep-through cure and mini- mize yellowing, which is more pronounced with Lucirin TPO than with Irgacure 184. It should be emphasized that Figure 36. Insolubilization profiles of an acrylate photoresist (Ebecryl 80) upon UV exposure to a UV (337.4 nm) or visible (488 nm) laser beam under O2-free conditions. Photoinitiator 1: Irgacure 369 (1 wt.-%), and 2: eosine (1 wt.-%)þ triazine (0.5 wt.-%)þ Irgacure 784 (0.5 wt.-%). Figure 37. Influence of photoinitiator (2 wt.-%) on the poly- merization of a liquid PUA resin exposed to glass-filtered UV radiation under laminate conditions. Film thickness: 24 mm, I¼ 50 mW
cm2. Kinetic Study and New Applications of UV Radiation Curing 1089 oxygen inhibition is not a critical factor anymore in this type of application, because the UV-curable resin is not in direct contact with air. After the oxygen dissolved in the resin has been consumed during the induction period, polymeriza- tion proceeds as rapidly as under an inert atmosphere. A liquid formulation containing an aliphatic polyurethane- acrylate oligomer (65 wt.-%), butyl acrylate (25 wt.-%), and acrylic acid (10 wt.-%) was found to give the best performance in terms of reactivity, adhesion, impact resis- tance and transparency.[112] An organosilane (1 wt.-% of g-methacryloxypropyl trimethoxysilane) had to be includ- ed in the UV glue formulation to improve adhesion and ensure a cohesive fracture in the impact resistance test. As required for safety glass applications, all of the broken glass pieces remained stuck on the elastomeric 2 mm thick intercalate. Impact resistance was further increased by introducing a polycarbonate core sandwiched between the two glass plates and bonded on each side by a UV glue. Similar results have been obtained by using a solid aliphatic polyurethane (Mn¼ 125 000 g/mol, Tg¼30 8C), plasticized with an aliphatic PUA oligomer, as an adhesive agent. The 0.8 mm thick photosensitive polymer film was laminated between two glass plates and maintained at 80 8C under 1 MPa pressure. After UV exposure, the softening temperature was increased from 75 8C to above 160 8C, with formation of a tight assembly. The photocrosslinked polymer exhibits the strong elastomeric character required to ensure high-impact resistance, as shown by its tensile properties: breaking occurred under a 22 MPa stress at 300% elongation.[112] Advantages of the UV technology over conventional thermal processing for the manufactur- ing of safety glasses are clearly apparent from Table 2. In addition to their high transparency, these colorless safety glasses appear to be very resistant to photodegrada- tion, specially in the presence of HALS radical scavenger (1 wt.-% of Tinuvin 292). After 2000 h of continuous exposure at 40 8C in a QUV-A accelerated weatherometer, the laminate stayed perfectly transparent and non-colored, while the adhesive strength remained essentially unaf- fected. Similar safety glasses have been recently produced by solar processing, simply by exposing the laminated sample to sunlight for a few minutes.[113] One of the distinct advantages of this technology free of energy costs is to allow large dimension items up to several square meters to be uniformly insolated. Solar curing was also successfully used to assemble mineral glass to a metal or to an organic glass (PMMA, PVC, polycarbonate).[113] The adhesive strength has been greatly enhanced by a styrene/butadiene rubber using as polymer matrix, which copolymerizes with the acrylate monomers and by adding to the formulation nano-sized silica particles grafted with an acrylate function. 4.4 Synthesis of Nanocomposite Materials by UV Curing Mineral charges such as silica particles or glass fibers can be introduced in the formulation of photocurable resins to produce hard and abrasion resistant composite materials by a short UV exposure at ambient temperature. This tech- nology has found its major opening in dentistry, where photocured dental polymeric materials are increasingly used as fast-hardening cements, sealants, adhesives, and dental prosthesis.[114] The inert filler, which can represent up to 80% by weight of the total composition, is usually made of micron-sized particles of quartz or hydroxyapa- tite.[115] Light scattering by the filler particles may cause a rise of the absorbed light through an increase in optical path length, thus allowing up to a few millimeter thick composite materials to be deep-through cured. This also requires to use photobleachable photoinitiators in order to allow the incident UV radiation to penetrate progressively deeper into the sample and thus promote a frontal polymeriza- tion.[13,39] The abrasion and scratching resistance of UV- cured coatings was found to be greatly improved by incorporation into the polymer matrix of some acrylate functionalized colloid silica.[116-118] The mineral particles can be chemically bonded to the polymer matrix by using a silica gel functionalized with various groups (thiol, epoxy, isocyanate) as a filler. UV radiation curing was also shown to be a rapid and environmental friendly process to produce composite membranes and fiber-reinforced plastics.[119,120] In all of these applications, the spatial and temporal control afforded by photopolymerization, together with the short setting time, make this technology attractive for rapid and inexpensive processing of polymeric composites. In conventional composite materials, filler and polymer are combined on a micron scale, which often leads to insuf- ficient adhesion between matrix and reinforcing filler. Composites that exhibit a change in composition and struc- ture on a nanometer length scale have been shown to afford remarkable property enhancement with respect to stiffness and strength, heat resistance and gas barrier properties.[121] Most of the nanocomposite materials are based on linear polymers (polyamides, polyolefins, polystyrene) and there- fore show poor chemical resistance. The mineral filler, usually a layered silicate such as clay, is typically incor- porated into the melted polymer. Thermodegradation may occur during this treatment, depending on the required temperature and swelling time, thus affecting the composite heat stability. Highly resistant nanocomposite materials have been recently produced by UV curing at ambient temperature of Table 2. Manufacturing of safety glasses. Process Thermal UV Temperature (8C) 150 25 Time (min) 60 < 0.1 Pressure (bar) 100 1 1090 C. Decker an acrylate-based resin containing a nano-scale filler (exfo- liated silicate platelets).[122,123] To obtain a true nanocom- posite, the 1 nm thick platelets need to be uniformly dispersed within the polymer matrix. This was achieved by making the clay (bentonite or montmorillonite) organo- philic through treatment with an alkylammonium salt, which leads to a widening of the interlayer and allows the liquid resin to penetrate into the clay galleries.[123] Exfoli- ation of the silicate platelets then occurs, as demonstrated by the disappearance of the characteristic X-ray diffraction peak. The increase in transparency, compared to the micro- composite, as well as the markedly reduced sedimentation rate, further argue in favor of an effective exfoliation of the clay layers. Upon UV exposure in the presence of a photoinitiator (Darocur 1173 and Irgacure 819), the crosslinking poly- merization of a PUA resin containing the nano-filler was found to proceed as fast and as extensively as that of the neat PUA formulation. Figure 38 shows the conversion versus time profiles recorded by RT-IR spectroscopy for 2 mm thick samples, by monitoring the disappearance of the near infrared band of the acrylate double bond at 6160 cm1. Under more intense illumination (600 mW
cm2), 1 s of exposure proved to be sufficient to obtain a hard and insoluble polymer material containing essentially no resi- dual unsaturation. In these UV-cured nanocomposites, 1 nm thick silicate layers are randomly dispersed within the polymer matrix to form a skeleton-like superstructure. This explains why a significant improvement in the polymer properties can be achieved already at low filler content. Such performance, together with cost-effective high-speed processing of a solvent-free resin, are key factors for the success of these newly developed photopolymer nanocom- posites. With respect to the potential hazard of materials containing nanoparticles, an issue recently raised at a symposium on ‘‘Nanotechnology: Environment Friend or Foe’’ (Washington, 11 March 2002), it should be empha- sized that these nanocomposite photopolymers are per- fectly safe, because the mineral filler is a natural, nontoxic product (clay). Moreover, the nanoparticles, present in small amounts, are tightly trapped into the polymer network and are therefore not able to ‘‘penetrate living cells and accumulate in animal organs’’, as was observed for other nano-sized particles. Conclusions Photoinitiated polymerization of multifunctional mono- mers and oligomers can be considered as the most effective method to generate quasi-instantly three-dimensional polymers. This environmentaly benign technology has found its major openings in the coating industry and in graphic arts, because of its many advantages regarding process facility and product performance. Such ultrafast polymerizations are best followed by real- time infrared spectroscopy, a technique that allows the formation of the polymer to be monitored in situ on a milli- second timescale. The influence of chemical and physical factors on the polymerization kinetics has thus been assessed for different types of UV-curable systems, as well as for monomer blends, under the same conditions as those used in most industrial applications, i.e., thin films exposed to intense polychromatic radiation in the presence of air. Significant progress has recently been achieved in UV radiation curing with the development of very efficient photoinitiators and highly reactive monomers and functio- nalized oligomers allowing to produce macromolecular materials with tailor-made properties. It has opened the way to novel potential applications of such photopolymers, in particular for the outdoor protection of organic materials by UV-cured coatings, the production of micron-sized relief images by direct laser writing, the rapid manufacturing of safety glasses and the synthesis of quick-setting nanocom- posite materials. In consideration of its unique features, one can reasonably expect this advanced technology to continue to expand and attract attention in an ever growing number of industrial sectors. Further research on photocrosslinking polymerization is still needed to improve both resin performance and final properties of UV-cured polymers. It should also concentrate on the mechanistic aspects to gain a better understanding and control of the manifold and complex processes invol- ved in ultrafast photocuring reactions. Figure 38. UV curing in air of a 2 mm thick PUA nanocomposite material (solid line) and neat PUA resin (dashed line) monitored in real time by means of near IR spectroscopy (6160 cm1). [Ebecryl 284]¼ 67 wt.-%, [HDDA]¼ 27 wt.-%, [organoclay]¼ 3 wt.-%, [Irgacure 819]¼ 1 wt.-%, I¼ 110 mW
cm2. Kinetic Study and New Applications of UV Radiation Curing 1091 Acknowledgement: The author wishes to thank his coworkers at the Polymer Photochemistry Laboratory,C. Bianchi,D.Decker, L. Keller, F. Masson, T. Nguyen Thi Viet, K. Studer, E. Weber- Koehl, K. Zahouily. Received: October 4, 2002 Revised: November 18, 2002 Accepted: November 19, 2002 [1] ‘‘Chemistry and Technology of UVand EB Formulation for Coatings, Inks and Paints’’, P. K. T. Oldring, Ed., SITA Technol., London 1991, Vol. 1-5. [2] ‘‘Radiation Curing Science and Technology’’, S. P. Pappas, Ed., Plenum Press, New York 1992. [3] ‘‘Radiation Curing in Polymer Science and Technology’’, J. P. Fouassier, J. F. Rabek, Eds., Chapman and Hall, London 1993, Vol. 1-5. [4] C. Roffey, ‘‘Photogeneration of Reactive Species for UV- Curing’’, Wiley, New York 1997. [5] C. Decker, Prog. Polym. Sci. 1996, 21, 593. [6] A. F. Jacobine, in: Radiation Curing in Polymer Science and Technology, J. P. Fouassier, J. F. Rabek, Eds., Chapman and Hall, London 1993, Vol. 3, p. 219. [7] J. V. Crivello, Adv. Polym. Sci. 1984, 62, 2. [8] J. G. Kloosterboer, Adv. Polym. Sci. 1998, 84, 1. [9] ‘‘Radiation Curing of Polymer Materials’’, C. E. Hoyle, J. F. Kinstle, Eds., ACS Symp. Series 417, American Chemical Society, Washington 1990. [10] J. P. Fouassier, ‘‘Photoinitiator, Photopolymerization and Photocuring’’, Hanser, Munich 1995. [11] C. Decker, in: Materials Science and Technology, H. E. H. Meijer, Ed., VCH Verlag, Weinheim 1997, Vol. 18, p. 615. [12] ‘‘Photopolymerization Fundamentals and Applications’’, A. C. Scranton, C. N. Bowman, R. W. Peiffer, Eds., ACS Symp. Series 673, American Chemical Society, Washington 1997. [13] C. Decker, Polym. Int. 1998, 45, 133. [14] S. Davidson, ‘‘Exploring the Science, Technology and Applications of UV and EB Curing’’, SITA Technol., London 1999. [15] J. V. Crivello, J. Polym. Sci., Part A:Polym.Chem. 1999, 37, 4241. [16] N. B. Cramer, C. N. Bowman, J. Polym. Sci., Part A: Polym. Chem. 2001, 39, 3311. [17] [17a] K. A. Berchtold, L. G. Lovell, J. Nie, B. Hacioglu, C. N. Bowman, Polymer 2001, 42, 4925; [17b] K. A. Berchtold, L. G. Lovell, J. Nie, B. Hacioglu, C. N. Bowman, Macromolecules 2001, 34, 5103. [18] E. Andrzejwska, Prog. Polym. Sci. 2001, 26, 605. [19] C. Decker, Opt. Mem. Neural Networks 2001, 10, 125. [20] J. L. Birbaum, S. Ilg, Proc. Rad. Tech. Eur. 2001, 545. [21] C. Decker, K. Moussa, J. Coat. Technol. 1990, 62(786), 55. [22] C. Decker, J. Polym. Sci., Part A: Polym. Chem. 1992, 30, 913. [23] R. Schwalm, L. Haussling, W. Reich, E. Beck, P. Enenkel, K. Menzel, Prog. Org. Coat. 1997, 32, 191. [24] J. E. Moore, in: UV-Curing Science and Technology, S. P. Pappas, Ed., Technology Marketing Corp. Norwalk, Connecticut 1978, Vol. 1, p. 134. [25] C. E. Hoyle, ‘‘Radiation Curing Science and Technology’’, S. P. Pappas, Ed., Plenum Press, New York 1992, p. 57. [26] V. D. Mc Ginnis, D. Dusek, J. Paint. Technol. 1974, 46(589), 23. [27] J. Paczhowski, D. C. Neckers, Macromolecules 1992, 25, 548. [28] T. C. Schaken, J. M. Warman, J. Phys. Chem. 1995, 99, 6145. [29] C. Decker, in:RadiationCuring of Polymers, R. D. Randell, Ed., Royal Soc. Chem. London 1987, Vol. 64, p. 16. [30] O. Dudi, W. T. Grubb, J. Appl. Polym. Sci. 1999, 74, 2133. [31] C. Decker, K. Moussa, Makromol. Chem. 1988, 189, 2381. [32] C. Decker, K. Moussa, Macromolecules 1989, 22, 4455. [33] C. Decker, K. Moussa, J. Polym. Sci., Part A: Polym. Chem. 1990, 28, 3429. [34] C. Decker, Polym. Prepr. (Am. Chem. Soc, Div. Polym. Chem.) 2001, 42(2), 762. [35] C. Decker, K. Moussa, Eur. Polym. J. 1990, 26, 393. [36] C. Decker, B. Elzaouk, D. Decker, J. Macromol. Sci. 1996, A33, 173. [37] C. Decker, Macromolecules 1990, 23, 5217. [38] C. Decker, Polym. Photochem. 1983, 3, 131. [39] V. Ivanov, C. Decker, Polym. Int. 2001, 50, 113. [40] M. Jacobi, A. Henne, J. Rad. Curing 1983, 19(4), 16. [41] C. Decker, K. Zahouily, D. Decker, T. Nguyen Thi Viet, Polymer 2001, 42, 7551. [42] W. Schnabel, T. Sumiyoshi, in: New Trends in the Photochemistry of Polymers, N. S. Allen, J. F. Rabek, Eds., Elsevier, London 1985, p. 69. [43] K. Dietliker, M. Kunz, J. P. Wolf, I. Gatlik, D. Neschadin, G. Gescheidt, P. Rzadek, G. Rist, J. Am. Chem. Soc. 1999, 121, 8382. [44] C. Decker, K. Zahouily, A. Valet, Macromol. Mater. Eng. 2001, 286, 5. [45] L. G. Lovell, H. Lu, J. E. Elliot, J. W. Stansburg, C. N. Bowman, Dent. Mater. 2001, 17, 504. [46] L. G. Lovell, K. A. Berchtold, J. E. Elliot, H. Lu, C. N. Bowman, Polym. Adv. Technol. 2001, 12, 335. [47] C. Decker, K. Moussa, Eur. Polym. J. 1991, 27, 881. [48] C. Decker, K. Moussa, J. Appl. Polym. Sci. 1987, 34, 1603. [49] C. Decker, D. Decker, F. Morel, in: Photopolymerization Fundamentals and Applications, A. B. Scranton, C. N. Bowman, R. W. Peiffer, Eds., ACS Symp. Series 673, Am. Chem. Soc., Washington 1997, p. 63. [50] K. S. Anseth, C. M. Wang, C. N. Bowman, Polymer 1994, 35, 3243. [51] K. S. Anseth, C. M. Wang, C. N. Bowman,Macromolecules 1994, 27, 650. [52] J. G. Kloosteboer, G. M. M. Van de Hei, R. G. Gossink, G. C. Dortant, Polym. Commun. 1984, 25, 32. [53] C. Decker, K. Moussa, J. Polym. Sci., Part A: Polym. Chem. 1987, 25, 739. [54] C. Decker, B. Elzaouk, Eur. Polym. J. 1995, 31, 1155. [55] C. Decker, in: Materials for Microlithography, L. F. Thompson, C. G. Willson, J. M. J. Fréchet, Eds., ACS Symp. Ser. 266, Am. Chem. Soc., Washington 1984, p. 207. [56] H. Kaczmarek, C. Decker, J. Appl. Polym. Sci. 1994, 54, 2147. [57] C. Decker, A. Jenkins, Macromolecules 1985, 18, 1241. [58] E. Beck, Proc. Rad. Technol. Eur. 2001, p. 643. [59] C. Decker, S. Biry, Progr. Org. Coat. 1996, 29, 81. [60] J. V. Crivello, R. Narayan, S. S. Sternstein, J. Appl. Polym. Sci. 1997, 64, 2073. [61] [61a] J. V. Crivello, J. Lee, in: ‘‘Radiation Curing of Polymer Materials’’, C. E. Hoyle, J. F. Kinstle, Eds., ACS Symp. Series 417, American Chemical Society, 1092 C. Decker Washington 1990, p. 398; [61b] J. V. Crivello, J. Lee, in: ‘‘Photopolymerization Fundamentals and Applications’’, A. C. Scranton, C. N. Bowman, R. W. Peiffer, Eds., ACS Symp. Series 673, American Chemical Society, Washington 1997, p. 82. [62] C. Decker, K. Moussa, J. Coat. Technol. 1987, 59(751), 97. [63] J. V. Crivello, N. Narayan, Macromolecules 1996, 29, 433. [64] C. Decker, H. Le Xuan, T. Nguyen Thi Viet, Proc. Rad. Tech. Asia 1995, 397. [65] J. V. Crivello, K. Dietliker, in: Photoinitiators for Free Radicals, Cationic and Anionic Polymerization, G. Bradley, Ed., Wiley, New York 1998, p. 53. [66] U. Müller, Trends Photochem. Photobiol. 1999, 5, 117. [67] J. V. Crivello, S. Kong, Macromolecules 2000, 33, 825. [68] Y. Yagci, S. Yildirim, A. Onen, Macromol. Chem. Phys. 2001, 202, 527. [69] A. Carroy, S. Ilg, T. Bolle, Farbe Lack, 2001, 107(8), 90. [70] A. Onen, Y. Yagci,Macromol. Chem. Phys. 2001, 202, 1950. [71] Z. Gomurashvili, J. V. Crivello, Macromolecules 2002, 35, 2962. [72] C. Decker, T. Nguyen Thi Viet, H. Pham Thi, Polym. Int. 2001, 50, 986. [73] J. V. Crivello, R. Narayan, S. S. Sternstein, J. Appl. Polym. Sci. 1997, 64, 2073. [74] S. F. Thames, H. Yu, Surf. Coat. Technol. 1999, 115, 208. [75] C. Decker, C. Bianchi, D. Decker, F. Morel, Prog. Org. Coat. 2001, 42, 253. [76] Y. M. Kim, L. K. Kostanski, J. F. Mac Gregor, Polymer, in press. [77] S. K. Rajamaran, W. A. Mowers, J. V. Crivello, J. Polym. Sci., Part A: Polym. Chem. 1999, 37, 4007. [78] C. Decker, T. Nguyen Thi Viet, D. Decker, E. Weber-Koehl, Polymer 2001, 42, 5531. [79] D. D. Gerst, Radiat. Curing 1982, 82, 27. [80] K. Doren, P. Freitag, D. Stoye, ‘‘Waterborne Coatings. The Environmentally Friendly Alternative’’, Hanser, Berlin 1994, p. 204. [81] S. Paul, Surface Coatings, Science and Technology, Wiley, Chichester 1996, p. 673. [82] [82a] S. Peeters, J. M. Loutz, Proc. Rad. Tech. Asia 1999, 13; [82b] S. Peeters, J. M. Loutz, Proc. Rad. Tech. Asia 1999, 153. [83] S. Kojima, T. Moriga, Y. Watanabe, J. Coat. Technol. 1993, 65(818), 25. [84] F. Masson, C. Decker, T. Jaworek, R. Schwalm, Prog. Org. Coat. 2000, 39, 15. [85] F. Masson, C. Decker, T. Jaworek, R. Schwalm, Proc. Rad. Tech. Eur. 2001, 723. [86] D. R. Burfield, K. L. Kim, K. S. Law, J. Appl. Polym. Sci. 1984, 29, 1661. [87] C. Decker, H. Le Xuan, T. Nguyen Thi Viet, J. Polym. Sci., Part A: Polym. Chem. 1995, 33, 2759. [88] C. Decker, T. Nguyen Thi Viet, H. Le Xuan, Polym. Mater. Sci. Eng. 1995, 72, 415. [89] H. Le Xuan, C. Decker, J. Polym. Sci., Part A: Polym.Chem. 1993, 31, 769. [90] C. Decker, T. Nguyen Thi Viet, H. Le Xuan, Eur. Polym. J. 1996, 32, 559. [91] C. Decker, H. Le Xuan, T. Nguyen Thi Viet, J. Polym. Sci., Part A: Polym. Chem. 1996, 34, 1771. [92] L. H. Sperling, V. Mishra, in: Polymeric Materials Encyclopedia, J. C. Salomone, Ed., CRC Press, Boca Raton, FL 1996, Vol. 5, p. 3292. [93] [93a] C. Decker, T. Nguyen Thi Viet, Macromol. Chem. Phys. 1999, 200, 358; [93b] C. Decker, T. Nguyen Thi Viet, Macromol. Chem. Phys. 1999, 200, 1965. [94] C. Decker, T. Nguyen Thi Viet, Polymer 2000, 41, 3905. [95] C. Decker, T. Nguyen Thi Viet, J. Appl. Polym. Sci. 2001, 82, 2204. [96] K. Zahouily, G. BlacheÌ€re, C. Decker, Proc. Radiol. Tech. Eur. 1997, 206. [97] C. Decker, K. Moussa, T. Bendaikha, J. Polym. Sci., Part A: Polym. Chem. 1991, 29, 739. [98] C. Decker, S. Biry, K. Zahouily, Polym. Degrad. Stabil. 1995, 49, 111. [99] C. Decker, S. Biry, Prog. Org. Coat. 1996, 29, 81. [100] C. Decker, K. Zahouily, J. Polym. Sci., Part A: Polym. Chem. 1998, 36, 2571. [101] C. Decker, K. Zahouily, A. Valet, J. Coat. Technol. 2002, 74, 87. [102] C. Decker, K. Moussa, T. Bendaikha, J. Polym. Sci., Part A: Polym. Chem. 1991, 29, 739. [103] C. Decker, K. Zahouily, Polym. Mater. Sci. Eng. 1993, 68, 70. [104] C. Decker, Polym. Photochem. 1983, 3, 131. [105] C. Decker, B. Elzaouk, J. Appl. Polym. Sci. 1997, 65, 833. [106] C. Decker, Opt. Mem. Neural Networks 2001, 10, 125. [107] C. Decker ‘‘Chemistry and Technology of UV and EB Formulation for Coatings, Inks and Paints’’, P. K. T. Oldring, Ed., SITA Technol., London 1994, Vol. 5, p. 145. [108] G. Y. Chen, Printed Circuit Board Technol., Pap. Conf. 1986, 1, 41. [109] G. S. Kumar, D. C. Neckers, Macromolecules 1991, 24, 4322. [110] S. Calixto, Appl. Opt. 1987, 26, 3904. [111] J. G. Woods, in: Radiation Curing Science and Technology, S. P. Pappas, Ed., Plenum Press, New York 1992, p. 333. [112] C. Decker, K. Moussa, J. Appl. Polym. Sci. 1995, 55, 359. [113] C. Decker, T. Bendaikha, J. Appl. Polym. Sci. 1998, 70, 2269. [114] L. A. Linden, in: Radiation Curing in Polymer Science and Technology, Vol. 4, J. P. Fouassier, J. P. Rabek, Eds., Elsevier Applied Science, London 1993, p. 387. [115] M. Alvarey, N. Davidenko, R. Garcia, A. Alonso, R. Rodriguez, R. M. Guerra, R. Sastre, Polym. Int. 1999, 48, 699. [116] C. Decker, Chimia 1993, 47, 378. [117] C. Vu, A. Eranian, C. Faurent, P. Noireaux, J. Devaux,Proc. Rad. Tech. Eur. 1999, 523. [118] M. Misra, A. Guest, M. McTilley, Surf. Coat. Int. 1998, 81, 594. [119] I. R. Bellobono, L. Righetto, in: ‘‘Radiation Curing in Polymer Science and Technology’’, J. P. Fouassier, J. F. Rabek, Eds., Chapman and Hall, London 1993, Vol. 4, p. 151. [120] L. S. Coons, B. Rangarajan, D. Godshall, A. B. Scranton, in: Innovative Processing and Characterization of Composite Materials, R. T. Gibson, T. W. Chon, P. K. Raju, Eds., ASME, New York 1995, Vol. 20, p. 227. [121] M. Alexandre, P. Dubois, Mater. Sci. Eng. 2000, 28, 1. [122] K. Zahouily, S. Benfarhi, T. Bendaikha, J. Baron, C. Decker, Proc. Rad. Tech. Eur. 2001, 583. [123] C. Decker, K. Zahouily, L. Keller, S. Benfarhi, T. Bendaikha, J. Baron, J. Mater. Sci. 2002, 37, 1. Kinetic Study and New Applications of UV Radiation Curing 1093 High-Speed Photocrosslinking of Thermoplastic Styrene- Butadiene Elastomers C. DECKER, T. NGUYEN THI VIET Laboratoire de Photochimie Générale (UMR-CNRS No. 7525), Ecole Nationale Supérieure de Chimie de Mulhouse, Université de Haute-Alsace, 3, rue Alfred Werner, 68200 Mulhouse, France Received 2 June 1999; accepted 5 November 1999 ABSTRACT: Photoinitiated thiol/ene polymerization was used to crosslink a triblock styrene/butadiene/styrene (SBS) polymer of low vinyl content (8%). The crosslinking process was followed by infrared spectroscopy (loss of unsaturation), insolubilization, swelling, and hardness measurements. The photogenerated thiyl radicals react with both the vinyl and the 2-butene double bonds of the copolymer. Concentrations of less than 1 wt % in the trifunctional thiol crosslinker and in the acylphosphine oxide photoinitiator proved to be sufficient to create, within 0.5 s, a permanent chemical network in the elastomeric phase. This UV-curing technology was successfully applied to crosslink rapidly commercial SBS-Kratont thermoplastic elastomers. It proved also effective in the case of the much less reactive triblock styrene/isoprene/styrene (SIS) polymer which contains no vinyl double bonds. The thiol/ene polymerization was shown to be a much more efficient process to crosslink SBS and SIS thermoplastic elastomers than was the copolymerization of the rubber double bonds with a diacrylate monomer. © 2000 John Wiley & Sons, Inc. J Appl Polym Sci 77: 1902-1912, 2000 Key words: photochemistry; crosslinking; UV curing; thermoplastic elastomers; po- lybutadiene; thiol INTRODUCTION Thermoplastic interpenetrating polymer net- works (IPNs) combine the properties of thermo- plastic materials and of elastomers. They have found various industrial applications, in particu- lar, for the manufacturing of molded articles, pro- tective coatings, and pressure-sensitive adhe- sives. Thermoplastic elastomers are usually made of block copolymers containing glassy domains and elastomeric domains which tend to segregate and form a two-phase morphology.1 A typical ex- ample is the polystyrene-block-polybutadiene- block-polystyrene (SBS)2 commercialized by Shell under the trademark Kratont. Typical SBS triblock thermoplastic elastomers exhibit high ex- tensibility, elastic recovery, and resilience; they are resistant to water and alcohol, but remain soluble in ketones, esters, and hydrocarbons. Because the network is made of physical crosslinks rather than chemical crosslinks, the glassy polystyrene (PS) domains lose their co- hesion upon heating above the Tg (around 80°C) and the polymer starts to flow. Such IPNs are therefore ill-suited for producing hot-melt adhe- sives or flexographic printing plates which must exhibit a high resistance to elevated tempera- tures (.150°C) and to aromatic solvents as well. The physical network can be strengthened by creating a chemical network through perma- nent covalent bonds joining together the chains of the elastomeric phase to give an insoluble material. Correspondence to: C. Decker. Contract grant sponsor: Shell Research, Louvain la Neuve, Belgium. Journal of Applied Polymer Science, Vol. 77, 1902-1912 (2000) © 2000 John Wiley & Sons, Inc. 1902 It was recently shown that such a crosslinking process is easily induced in SBS thermoplastic elastomers by UV irradiation in the presence of a radical-type photoinitiator.3,4 The shear adhesion failure temperature (SAFT) was found to increase from 80 to over 160°C upon UV curing, while the tack remained essentially unchanged.5,6 The net- work was considered to be formed by polymeriza- tion of the pendent vinyl double bonds, which are known to be more reactive than are the 2-butene double bonds of the polybutadiene chain. The crosslinking process can be markedly accelerated by the addition of multifunctional acrylate or thiol monomers that copolymerize with the poly- butadiene unsaturations.4,7 The use of light to induce the chain reaction has several advantages, such as high cure speed, low-energy consumption, ambient temperature operations, dry processing without VOC emission, and selective curing in the illuminated areas. By performing the UV expo- sure through a mask, one can generate, after sol- vent development, the high-relief images which are needed for the manufacture of flexographic printing plates. In a previous work, we showed that photoiniti- ated thiol-ene polymerization is an effective way to crosslink thermoplastic SBS rubbers, total in- solubilization being achieved within less than 1 s.8 As thiyl radicals were found to be 16 times more reactive toward vinyl than were 2-butene double bonds, the first experiments were con- ducted on a specially designed SBS sample hav- ing a high vinyl content (59%). Crosslinking in the elastomeric phase was shown to still proceed when the polybutadiene vinyl content was de- creased to 8%, as long as the thiol concentration was kept above 5 wt %. To reduce the bad smell associated with the use of mercaptans, an at- tempt was made to further lower the concentra- tion of the thiol crosslinker. Another objective of this study was to reduce the photoinitiator con- centration below 1 wt %, the lowest value used so far, in order to increase the penetration of UV radiation into the sample and be able to cure thick samples. Most of the experiments were car- ried out with a typical SBS sample, which, like most commercial SBS rubbers, contains a low amount of pendent vinyl groups. The crosslinking process was quantitatively studied by following the disappearance of the vinyl groups, as well as the insolubilization and the hardening of the SBS sample upon UV irradiation. EXPERIMENTAL Materials The SBS from Shell used in this study was spe- cially designed by Shell (Louvain-la Neuve, Bel- gium). It contained 8% pendent vinyl groups, the 2-butenylene units (-CH2-CHACH-CH2) lo- cated on the polybutadiene backbone, represent- ing 92% of the unsaturation content. Because of its relatively low styrene content (;20%), the SBS behaves similarly to a cured conventional vulca- nized rubber. Some experiments were carried out with commercial SBS samples (Kratont D-1102 and Kratont D-1186 from Shell) that differ mainly by the molecular weight and the structure of the polybutadiene chain, which was linear or branched, respectively. The UV-curing perfor- mance of a polystyrene-block-polyisoprene-block- polystyrene (SIS) was also evaluated on a com- mercial sample (Kratont D-1107 from Shell). A trifunctional thiol was selected as the crosslinking agent, trimethylolpropane mercapto- propionate (TRIS), from Evans Chemetics (Lex- ington, MA). It was added to a toluene solution of SBS at a concentration between 0.2 and 3% by weight of SBS. Among the various radical-type photoinitiators tested, acylphosphine oxides proved to be the most efficient ones to pho- tocrosslink the SBS samples. All the experiments were carried out using 2,4, 6-trimethylbenzoyl diphenylphosphine oxide (Lucirin TPO from BASF, Ludwigshafen, Germany) as the photoini- tiator, at a typical concentration of 1 wt %. The chemical formulas of the three components used in this study are given in Figure 1. Figure 1 Chemical formulas of the SBS rubber, thiol, and photoinitiator used. HIGH-SPEED PHOTOCROSSLINKING OF SBS ELASTOMERS 1903 Irradiation In a typical UV-curing experiment, 20-mm thick films were cast from a toluene solution containing the SBS rubber, the thiol monomer, and the pho- toinitiator onto either a KBr crystal for infrared analysis or a glass plate for insolubilization and hardness measurements. Samples were exposed at ambient temperature to the radiation of a 80 W/cm medium-pressure mercury lamp (IST) in the presence of air, at a passing speed of 60 m/min, which corresponds to an exposure dura- tion of 0.1 s per pass. The maximum light inten- sity at the sample position was measured by ra- diometry (IL-390 light bug) to be 600 mW cm22 in the UV range. Analysis The kinetics of the light-induced crosslinking of the rubber film was studied quantitatively by FTIR spectroscopy, by following the decrease upon UV exposure of the absorption band charac- teristic of the vinyl double bond at 910 cm21. The degree of conversion was calculated from the ratio of the corresponding IR absorbance before and after UV exposure (A0 and At). It should be men- tioned that even very small variations of the IR band intensity can be monitored accurately, be- cause the analysis is performed on the same sam- ple exposed to UV light for various durations, at exactly the same spot. The gel fraction and the degree of swelling (SR) of the irradiated polymer were determined by soaking the sample in toluene for 1 day at room temperature. The insoluble polymer was recov- ered by filtration and dried at 70°C to a constant weight. The hardness of the coating was evalu- ated before and after irradiation by monitoring the damping time of the oscillations of a pendu- lum (Persoz hardness). The hardness was shown to be strongly dependent on the glass transition temperature,9 with Persoz values ranging typi- cally from 30 s for soft elastomeric materials to 300 s for hard and glassy polymers. RESULTS Photoinitiated thiol-ene polymerization has been thoroughly investigated over the past 20 years, the main findings having recently been reviewed by Jacobine in a comprehensive survey.10 The polymerization proceeds by a step-growth addi- tion mechanism which is propagated by a chain- transfer reaction involving the thiyl radicals (RSz): A polymer network will be formed only if a diene or a polyene reacts with a multifunctional thiol, such as TRIS: By contrast to most radical-induced crosslink- ing processes, the thiol/ene polymerization is not very sensitive to oxygen inhibition, because the peroxyl radicals (PO2 z ) formed by O2 scavenging can also react with the thiol and contribute to the propagation of the chain reaction: All the curing experiments described in this article were carried out on thin films in contact with air. 1904 DECKER AND VIET There are only a few reports in the literature on the photocrosslinking of the thiol-SBS sys- tem,7,8,11 with essentially no information on the reaction kinetics. We examined the influence of the thiol and photoinitiator concentrations on the UV curing of a low-vinyl SBS sample and on some commercial SBS rubbers as well. Influence of the Thiol Concentration Vinyl Consumption The SBS samples containing 1 wt % of a photo- initiator (Lucirin TPO) and various concentra- tions of thiol (0.2 wt % # [TRIS] # 3 wt %) were exposed to UV radiation for different times up to 3 s. As the thiol consumption could not be fol- lowed by IR spectroscopy at the low concentration used, the polymerization reaction was monitored through the decrease of the vinyl IR band at 910 cm21. Although this decrease is quite small, it can still be quantified with great accuracy because each sample undergoing photocrosslinking was analyzed by the IR beam at exactly the same spot. Figure 2 shows some typical kinetic curves of the vinyl group consumption upon UV exposure and the effect of the thiol content on the curing reaction which follows a two-step kinetics. The thiol-vinyl copoly- merization proceeds rapidly during the first 0.3 s of irradiation, until the thiol monomer has been con- sumed. At a 5 wt % concentration of TRIS, the SH group was found to have essentially disappeared after a 0.3-s UV exposure. The slow decrease of the vinyl double bonds observed upon further irradia- tion parallels that in the thiol-free SBS and was therefore attributed to the homopolymerization of these groups. Although the decrease of the 2-butene double bond cannot be followed by IR spectroscopy be- cause of cis-trans isomerization (bands at 3060 and 965 cm21, respectively), it is still possible to evaluate the contribution of the thiol-butene co- polymerization to the crosslinking process from the difference between the loss of the thiol group D [SH] and that of the vinyl group D [V]: Loss of vinyl by copolymerization: D [V]copo 5 D [V]SBS1TRIS 2 D [V]SBS. Loss of thiol group by copolymerization: D [SH] 5 D [V]copo 1 D [B]copo. Loss of butene group [B] by copolymerization: D [B]copo 5 D [SH] 2 D [V] SB 1TRIS 1 D [V]SBS. In Figure 3, the amounts of vinyl and butene groups which have copolymerized with the thiol after a 0.3-s exposure are plotted as a function of the SH concentration. The two quantities [V]copo and [B]copo were found to be quite similar and to increase linearly with the thiol content, as ex- pected. The fact that a single curve was obtained implies that the rate constants kv and kb of the two copolymerization processes are in the reciprocal ra- tio of the vinyl and 2-butene concentrations: Figure 2 Influence of the thiol concentration on the vinyl polymerization of SBS upon UV exposure. [Luci- rin TPO] 5 1 wt %. Figure 3 Thiol dependence of the amount of vinyl (F) and butene (Å’) double bonds copolymerized after 0.3-s UV exposure of SBS. [Lucirin TPO] 5 1 wt %. HIGH-SPEED PHOTOCROSSLINKING OF SBS ELASTOMERS 1905 kv@V# 5 kb@B# kv kb 5 @B# @V# 5 11.7 The higher reactivity of the vinyl double bond toward thiols is in good agreement with previous structure-reactivity studies on the addition of thiols on olefins.12 It was attributed to an increase in electron density of the olefin which reacts with a relatively electrophilic thiyl radical.13 These data also show that the copolymeriza- tion of vinyl groups with a trithiol is favored over the homopolymerization (D [V]copo/D [V]homo 5 5.3 at [SH]0 5 0.075 mol kg 21), even though the vinyl concentration is much higher than is the thiol concentration ([vinyl]0/[SH]0 5 16.8 at 1 wt % TRIS). This means that the propagating alkyl radicals are much more reactive toward the thiol group than toward the vinyl double bond: Insolubilization The crosslinks created in the elastomeric phase by both copolymerization and homopolymeriza- tion reactions led to insolubilization and solvent resistance of the UV-irradiated SBS-thiol sys- tem. Figure 4 shows some typical plots of the insoluble fraction as a function of the exposure time for different SBS formulations containing 1 wt % Lucirin TPO and increasing amounts of TRIS (from 0.2 to 3 wt %). For TRIS concentra- tions above 1 wt %, essentially all of the polymer was found to become insoluble after a 0.3-s irra- diation. At lower TRIS contents, the SBS polymer remained partly soluble, the gel fraction increas- ing slowly upon further exposure due to homopo- lymerization of the vinyl groups, the small amounts of thiol having already been consumed. From the quantity of reactive groups consumed after a 0.3-s exposure (2.2 3 1022 , N , 23 3 1022 mol kg21), one can calculate the concen- tration of branch points in the elastomeric phase and the number of crosslinks (X) formed per SBS chain at the various thiol concentrations used: Xtotal 5 ~@vinyl#homo 1 @vinyl#copo 1 @butene#copo! MSBX 1000 the vinyl and butene concentrations being ex- pressed in mol kg21. This number was found to increase linearly with the thiol content up to a value of 34 crosslinks per chain at [TRIS] 5 3 wt %, as shown in Figure 5. Total insolubilization requires the formation of about 15 crosslinks per SBS chain. This is clearly apparent when the gel fraction, measured after a 0.3-s exposure, is plot- ted as a function of the number of crosslinks formed per SBS chain (Fig. 6). Figure 4 Influence of the thiol concentration on the insolubilization of SBS upon UV exposure. [Lucirin TPO] 5 1 wt %. 1906 DECKER AND VIET For the SBS sample that contains no thiol, crosslinking results only from the homopolymer- ization of the few pendent vinyl double bonds, as 2-butene double bonds are much less reactive.13 After 0.3 s, the vinyl consumption was measured to be 7 3 1023 mol kg21, which corresponds to only one crosslink per SBS molecule. This ex- plains why the polymer remains soluble at that stage (Fig. 4), insolubilization requiring the buildup of a least three bridges per chain. The amount of solvent retained by the swollen polymer is directly related to the network crosslink density or to the number-average mo- lecular weight of the network chain.14 Because the interaction parameter of the Flory swelling equation was unknown for the SBS/toluene cou- ple studied, it was only possible to make a quali- tative evaluation of the effect of the UV dose and the thiol content on the tightness of the polymer network formed. Figure 7 shows how the swelling ratio (SR 5 swollen solvent weight/dry polymer weight) varies with the exposure time for the various TRIS/SBS formulations. At very low TRIS concentrations (0.2 wt %), the SR value decreases from 35 to 15 with increasing exposure time, which indicates that a very loose polymer net- work has been formed. As the thiol concentration was increased to 3 wt %, a tighter network was formed, the swelling ratio decreasing to a mini- mum value of 5. Figure 6 shows the swelling ratio dependence on the number of crosslinks per SBS chain. Hardening The crosslinking caused by the thiol-ene polymer- ization during the first 0.3 s of UV exposure did not affect significantly the elastomeric character Figure 6 Dependence of the gel fraction and swelling ratio on the number of crosslinks per SBS chain after a 0.3-s exposure. [Lucirin TPO] 5 1 wt %. Figure 7 Influence of the thiol concentrationon the swelling of a photocrosslinked SBS. [Lucirin TPO] 5 1 wt %. Figure 5 Thiol dependence of the amount of vinyl and butene crosslinks formed per chain after 0.3-s UV exposure of SBS. [Lucirin TPO] 5 1 wt %. HIGH-SPEED PHOTOCROSSLINKING OF SBS ELASTOMERS 1907 of the polymer, which remained soft and flexible as long as the TRIS concentration did not exceed 1 wt %. At higher concentrations, the Persoz hardness was found to increase from 40 to 75 s, as shown in Figure 8. As the UV irradiation was pursued, the SBS hardness increased further, es- pecially for [TRIS] . 1 wt %. This unwanted hardening may reduce the adhesive properties (tackiness) of the UV-cured polymer and should be strictly controlled. The influence of the thiol on the pho- tocrosslinking of the SBS elastomer is illustrated in Figure 9, which shows the variation with the TRIS concentration of the gel fraction, the swell- ing ratio, and the Persoz hardness of a sample UV-exposed for 0.3 s in the presence of Lucirin TPO (1 wt %). A TRIS concentration of 1 wt % and a UV dose of 0.3 s (which generates about 12 crosslinks per chain) appear as the best compro- mise for achieving a fast curing and still ensuring the high adhesion requested for pressure-sensi- tive adhesives. Influence of the Photoinitiator Concentration All the UV-curing experiments reported so far were performed with the same concentration of the photoinitiator, namely, 1 wt % Lucirin TPO. To increase the penetration of UV radiation for curing of thick samples and to reduce the formu- lation cost as well, the photoinitiator concentra- tion was decreased to 0.1 wt %. This leads to a slowing of the polymerization reaction, as ex- pected from the decreased initiation rate. Figure 10 shows the decay profiles of the vinyl double bond upon UV exposure of the SBS/TRIS (1 wt %) system at various concentrations in Lucirin TPO. Figure 9 Influence of TRIS concentration on the pho- tocrosslinking of SBS. [Lucirin TPO] 5 1 wt %. Figure 8 Influence of the thiol concentration on the hardness of a photocrosslinked SBS. [Lucirin TPO] 5 1 wt %. Figure 10 Influence of the photoinitiator concentra- tion on the vinyl polymerization of SBS upon UV expo- sure. [TRIS] 5 1 wt %. 1908 DECKER AND VIET As expected, the vinyl conversion was found to be reduced when the photoinitiator concentration was decreased, but it was still sufficient to cause insolubilization of SBS, as long as [Lucirin TPO] $ 0.2 wt %. Figure 11 shows the corresponding insolubilization profiles for TPO concentrations between 0 and 1 wt %. One may notice that at a TPO content of 0.2 wt % the thiol-ene polymerization proceeds effi- ciently enough to yield a nearly completely insol- uble material within 0.3 s. Very similar polymer- ization and insolubilization profiles were obtained by operating under the following experimental conditions: [TRIS] 5 1 wt % and [TPO] 5 0.2 wt %, or [TRIS] 5 0.5 wt % and [TPO] 5 1 wt %. The former formulation has the advantage of provid- ing a more uniform deep-through cure, especially for thick samples, while being also more cost ef- fective. It can be seen in Figures 10 and 11 that curing occurs even in the absence of any added photoini- tiator, although at a much slower pace. This is probably due to the presence in SBS of some im- purities which act as initating chromophores, but also to the high reactivity of the thiyl radicals. Indeed, no crosslinking could be detected upon UV-irradiation of neat SBS under the selected experimental conditions. The chemical network formed in the elasto- meric phase of the SBS rubber has a loose struc- ture when low concentrations of the photoinitia- tor were used. After a 0.3-s UV exposure, the value of the SBS swelling ratio was found to in- crease from 11 to 17 when [TPO] was decreased from 1 to 0.1 wt %, as shown in Figure 12. The polymer hardening upon UV exposure is less pro- nounced at low photoinitiator concentrations, as expected, but it still occurs, even in the photoini- tiator-free sample (Fig. 13). At [TPO] 5 0.2 wt %, the Persoz value increased from an initial value of 35 to 60 s after a 1-s exposure, compared to 100 s at [TPO] 5 1 wt %. Operating at such low photo- initiator concentrations, and at irradiation times below 1 s, is therefore recommended for adhesive applications, where the polymer should remain soft and flexible after crosslinking. Figure 13 Influence of the photoinitiator concentra- tion on the hardness of a photocrosslinked SBS. [TRIS] 5 1 wt %. Figure 11 Influence of the photoinitiator concentra- tion on the insolubilization of SBS upon UV exposure. [TRIS] 5 1 wt %. Figure 12 Influence of the photoinitiator concentra- tion on the swelling of a photocrosslinked SBS. [TRIS] 5 1 wt %. HIGH-SPEED PHOTOCROSSLINKING OF SBS ELASTOMERS 1909 Figure 14 summarizes these results by show- ing the variation with the photoinitiator concen- tration of the vinyl double-bond consumption, gel fraction, swelling ratio, and Persoz hardness, in an SBS/TRIS (1 wt %) sample UV-irradiated for 0.3 s. A Lucirin TPO concentration of 0.2 wt % appears to be enough to obtain a crosslinked elas- tomer showing the required performance with re- spect to the photoreactivity, the viscoelastic prop- erties, and the solvent and heat resistance. By operating at such low photoinitiator concentra- tions, a few millimeter-thick SBS samples were readily crosslinked by UV-irradiation. This is due to both an increased penetration of UV light and to a fast photobleaching of Lucirin TPO which promotes a frontal polymerization.15 Photocrosslinking of Kraton Thermoplastic Elastomers A similar study was performed on two commercial thermoplastic styrene-butadiene elastomers, Kratont D-1102 and Kratont D-1186 from Shell. The styrene/butadiene ratio (30/70) and the poly- butadiene vinyl content (;10%) were similar to that of the SBS sample previously studied. They differ mainly by the molecular weight of the po- lybutadiene block and its structure which was linear or branched, respectively. Both samples proved to be very reactive when exposed to UV radiation in the presence of TRIS (1 wt %) and Lucirin TPO (1 wt %). Insolubiliza- tion was achieved within less than 1 s, with for- mation of a soft (Persoz hardness 5 60 s) and loosely crosslinked (SR 5 10) polymer. In the absence of thiol, the polymerization of the vinyl double bond proceeds less efficiently, so that in- solubilization hardly occurs upon UV exposure, as shown in Figure 15 for Kratont D-1102. In the presence of 1 wt % TRIS, the branched SBS (Kratont D-1186) was found to become more readily insoluble than was the linear SBS, so that the photoinitiator concentration could be de- creased to 0.2 wt % and still achieve insolubiliza- tion within 0.3 s (Fig. 16). As expected, the branched structure of the elastomeric phase causes an increase of the crosslink density of the UV-cured rubber, as shown by the swelling data: an SR value of 8 for Kratont D-1186 and of 13 for Kratont D-1102. This trend was also found when neat SBS was UV-irradiated in the presence of a photoinitiator. After a 3-s exposure, the gel frac- tion was measured to be 70% for Kratont D-1186 (Fig. 17), compared to only 20% for Kratont D-1102 (Fig. 15). To assess the efficiency of the thiol/ene system to crosslink SBS rubbers, in comparison to the Figure 15 Photocrosslinking of Kraton® D-1102. In- fluence of the thiol. [Lucirin TPO] 5 1 wt %. Figure 14 Influence of the photoinitiator concentra- tion on the photocrosslinking of SBS. [TRIS] 5 1 wt %. UV exposure: 0.3 s. 1910 DECKER AND VIET usual acrylate-based system, we UV-irradiated a sample of Kratont D-1186 containing hexanediol diacrylate (HDDA) as a crosslinking agent (20 wt %), under exactly the same conditions. It can be seen in Figure 17 that, in spite of the much higher crosslinker content, insolubilization takes place at a slower pace than for the sample containing 1 wt % TRIS. A gel fraction of 90% was reached after 3 s with the diacrylate and after only 0.3 s for the SBS/TRIS system. The crosslink density of the UV-cured polymer is also much lower in the SBS/acrylate system, as shown by the swelling profiles (Fig. 18). After a 1-s exposure, the follow- ing SR values were measured: 70 for neat SBS, 22 for SBS/HDDA, and 8 for SBS/TRIS. The fact that the acrylate-based rubber is less crosslinked than is the thiol-based rubber was attributed to the competitive homopolymerization of the acrylate double bonds, which does not generate any addi- tional crosslinks between the SBS chains. Another type of thermoplastic rubber (SIS) consists of a styrenic triblock copolymer where the elastomeric midphase is made of a polyiso- prene, such as Kratont D-1107, which is com- monly used in flexographic printing plates and adhesives applications. Its styrene/rubber ratio is 14/86, the linear polyisoprene chains containing no vinyl double bonds. SIS is therefore much less reactive than is SBS and remains completely sol- uble after a 5-s UV exposure in the presence of Lucirin TPO (1 wt %). The addition of a diacrylate monomer (20 wt % of HDDA) leads to partial insolubilization (Fig. 19), the crosslinks resulting from the copolymerization of the amylene and acrylate double bonds.16,17 Crosslinking was found to proceed much more efficiently in a SIS/ TRIS (2 wt %) system, UV-irradiated in the pres- ence of 1 wt % Lucirin TPO. Total insolubilization of SIS was achieved within 3 s (Fig. 19), by reac- tion of the thiyl radicals with the amylene double bond: Figure 16 Photocrosslinking of Kraton® D-1186 in the presence of (F) 0.2 wt % or (Å’) or 1 wt % Lucirin TPO. [TRIS] 5 1 wt %. Figure 17 Influence of thiol and acrylate monomers on the photocrosslinking of Kraton® D-1186. [TRIS] 5 1 wt %; [HDDA] 5 20 wt %; [Lucirin TPO] 5 1 wt %. Figure 18 Influence of thiol and acrylate monomers on the swelling ratio of photocrosslinked Kraton® D-1186. [TRIS] 5 1 wt %; [HDDA] 5 20 wt %; [Lucirin TPO] 5 1 wt %. HIGH-SPEED PHOTOCROSSLINKING OF SBS ELASTOMERS 1911 The best performance was obtained by combin- ing the two crosslinking agents: 2 wt % of TRIS and 20 wt % of HDDA. It should be emphasized that this SIS-based formulation proved actually to be more reactive than the SBS-based formula- tion containing 20 wt % HDDA (Fig. 17), with formation of a tighter polymer network (SR val- ues of 12 and 16, respectively). CONCLUSIONS Thermoplastic styrene/butadiene elastomers con- taining a low amount of the vinyl group (8%) can be readily crosslinked at ambient temperature by UV irradiation in the presence of very low amounts (#1 wt %) of a trifunctional thiol crosslinker. Both the vinyl and butene double bonds of the polybutadiene chain participate in the thiol/ene polymerization which proceeds by step-growth addition, the thiyl radicals acting as chain-transfer agents. The covalent bonds gener- ated within the elastomeric phase led to a marked increase of the solvent and heat resistance of the UV-cured polymer. Insolubilization takes place within 0.3 s upon intense illumination, with for- mation of a soft and flexible material, well suited for adhesive and flexographic applications. The photocrosslinking technology offers a num- ber of advantages, such as on-line processing, high- speed curing, solvent-free formulations, low-energy consumption, low cost of chemicals, and selective cure in the illuminated areas. It is easily applicable to commercial styrene/butadiene thermoplastic elastomers, and even to the less reactive styrene/ isoprene rubbers, by adding a diacrylate comono- mer. The main applications of these photocurable rubbers are expected to be found in industrial sec- tors where cure speed and spatial control are a major concern, such as for the manufacturing of hot-melt adhesives and sealants, safety glasses, photoresists, and flexible printing plates. The authors wish to thank Shell Research (Louvain la Neuve, Belgium) for a research grant. REFERENCES 1. Sperling, L. H.; Mishra, V. In Polymeric Materials Encyclopedia; Salamone, J. C., Ed.; CRC: Boca Ra- ton, FL, 1996; Vol. 5, p 3292. 2. Hosine, G. H. In Polymeric Materils Encyclopedia; Salamone, J. C., Ed.; CRC: Boca Raton, FL, 1996; Vol. 10, p 8002. 3. Huber, H. F. In Radiation Curing in Polymer Sci- ence and Technology; Fouassier, J. P.; Rabek, J. F., Eds.; Elsevier: London, 1993; p 51. 4. Decker, C.; Nguyen Thi Viet, T. Macromol Chem Phys 1999, 200, 358. 5. Dupont, M.; De Keyser, N. In Proceedings of the Tech Europe, Conference Maestricht 1995, p174. 6. Mayenez, C.; Muyldermans, X. EP Patent 0 696 761, 1996. 7. Rice, C. S.; Sasaki, Y.; Plamthottan, S. P. U.S. Patent 5 166 226, 1992. 8. Decker, C.; Nguyen Thi Viet, T. Polymer 2000, 41, 3905. 9. Schwalm, R.; Häussling, L.; Reich, W.; Beck, E.; Enenkel, P.; Menzel, K. Prog Org Coat 1997, 32, 191. 10. Jacobine, A. F. In Radiation Curing in Polymer Science and Technology; Fouassier, J. P.; Rabek, J. F., Eds.; Elsevier: London, 1993; Vol. 3, p 219. 11. Hein, P. R.; Evans, J. A.; Yang, M. W. U.S. Patent 4 237 676, 1980. 12. Walling, C.; Helmereich, W. J. J Am Chem Soc 1959, 81, 1144. 13. D’Souza, V. T.; Nanjundiak, R.; Baeza, J.; Szmant, H. H. J Org Chem 1987, 52, 1720. 14. Flory, P. J. Principles of Polymer Chemistry; Cor- nell University: Ithaca, NY, 1953. 15. Decker, C. Polym Int 1998, 45, 133. 16. Le Xuan, H.; Decker, C. J Polym Sci Polym Chem Ed 1993, 31, 769. 17. Decker, C.; Nguyen Thi Viet, T.; Le Xuan, H. Eur Polym J 1996, 32, 549, 559. Figure 19 Influence of thiol and acrylate monomers on the photocrosslinking of Kraton® KX-601. [TRIS] 5 2 wt %; [HDDA] 5 20 wt %; [Lucirin TPO] 5 1 wt %. 1912 DECKER AND VIET Photocrosslinking of functionalized rubbers IX. Thiol-ene polymerization of styrene-butadiene-block-copolymers C. Decker* , T. Nguyen Thi Viet Laboratoire de Photochimie Ge´n´rale (UMR N87525), Ecole Nationale Supe´ri ure de Chimie de Mulhouse, Universite´ d Haute Alsace, 3, rue Werner, 68200 Mulhouse, France Received 16 June 1999; received in revised form 1 September 1999; accepted 10 September 1999 Abstract A thermoplastic elastomer, polystyrene-block-polybutadiene-block-polystyrene (SBS) with a high vinyl content (59%), was crosslinked within a fraction of a second by UV-irradiation in the presence of a trifunctional thiol and an acylphosphine oxide photoinitiator. The curing process was followed by IR spectroscopy, insolubilization and hardness measurements. Both the thiol and the photoinitiator concentrations were shown to affect the kinetics of the thiol-ene polymerization. Insolubilization was found to occur even at very low concentrations in thiol (#1 wt%) and in photoinitiator (0.1 wt%). The heat resistance of the photosensitive formulation is high enough at 1508C to allow this system to be used for hot-melt adhesive applications.q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Photochemistry; Crosslinking; UV-curing 1. Introduction Three-dimensional polymer networks can be readily generated by photoinitiated polymerization of multi- functional monomers or of polymers bearing reactive func- tions [1]. This technology has been mainly used, so far, to perform a quasi-instantaneous liquid to solid phase change, a process called UV-curing. But it can also serve to transform irreversibly a soluble polymer into an insoluble material showing improved resistance to elevated tempera- tures and chemicals. Among the various functionalized polymers studied so far, promising results have been obtained with rubber-type elastomers bearing epoxy rings [2-4], acrylate functions [5,6] or vinyl [7,8] double bonds on the main chain. Crosslinking was achieved within seconds upon UV exposure in the presence of a cationic type or radical-type photoinitiator, respectively. The main advantages of using light to initiate the chain reaction, besides the rapidity of the curing process, consist in the solvent-free formulations, the low energy consumption, ambient temperature operations and control of the location and time of the process, which occurs specifically in the illuminated areas. In a recent paper [8], we have shown that styrene- butadiene-block-copolymers (SBS) undergo a rapid cross- linking when they are exposed to UV radiation in the presence of a radical photoinitiator. Complete insolubiliza- tion requires the reaction of 17 double bonds per polymer chain for an SBS containing 8% pendent vinyl double bonds. These thermoplastic elastomers, known under the trademark KRATONw, have a two-phase morphology and are commonly used as pressure sensitive adhesives, in combination with a tackfying resin [9,10], and as flexo- graphic printing plates [11]. UV-curing creates covalent bonds within the elastomeric phase, thus reinforcing irrever- sibly the already existing physical network. As a result, the photocured SBS becomes more resistant to high tempera- tures, the values of the Shear Adhesion Failure Temperature (SAFT) rising from 808C to over 1608C [10]. The crosslinking process was markedly accelerated by adding to SBS a diacrylate monomer which copolymerizes with the polybutadiene vinyl and butene double bonds [8], but at the expense of the heat resistance, thus making the photosensitive material ill-suited for hot-melt adhesive applications. Another approach, which proved to be even more effective, consisted in incorporating a trifunctional thiol into SBS to perform crosslinking through a thiol-ene photopolymerization [12,13], according to the following reaction scheme: Photoinitiator1 RSH!hn RSz RSz 1 CH2yCH-R 0 ! RS-CH2- C z H-R0 RS2 CH2-C z H-R0 1 RSH! RS-CH2-CH2-R0 1 RSz Polymer 41 (2000) 3905-3912 0032-3861/00/$ - see front matterq 2000 Elsevier Science Ltd. All rights reserved. PII: S0032-3861(99)00649-7 * Corresponding author. The polymerization proceeds by a step growth addition mechanism which is propagated by a chain transfer reaction involving the thiyl radical (RSz) [14,15]. A thiol concentra- tion of 5 wt% proved to be enough to increase 10-fold the insolubilization rate [16]. The heat resistance of the uncured formulation, although superior to the acrylate-SBS system, was still not sufficient to meet the specifications required for the manufacturing of hot-melt adhesives or flexographic printing plates (stability of the unreacted sample during 20 min upon processing at 1508C) [7]. Moreover, one of the major drawbacks of the polyene-thiol system lies in the bad smell of mercaptans, which is reduced as polymer- ization proceeds, the thiol becoming incorporated into the polymer network. The main objective of the present study was to determine how much the thiol concentration can be reduced, without affecting substantially the reactivity, in order to improve the heat resistance of the photosensitive formulation, and reduce its smell as well. For the same reason, the concentra- tion of the photoinitiator was also decreased, with the expected additional benefit of allowing thick samples to be deep-through cured, because of the increased penetration of UV radiation. The progress of the crosslinking reaction has been followed quantitatively through the insolubiliza- tion and hardening of the UV-exposed SBS-thiol sample, and by monitoring the disappearance of the polybutadiene double bonds by infrared spectroscopy. 2. Experimental 2.1. Materials The polystyrene-block-polybutadiene-block-polystyrene (SBS) from SHELL used in this study contained 59% pendent vinyl groups, the 2-butenylene units (-CH2- CHyCH-CH2) located on the polybutadiene backbone representing 41% of the unsaturation content. Because of its relatively low styrene content (,15%) the SBS behaves similarly to a conventionally vulcanized rubber. A tri- functional thiol was selected as crosslinking agent, trimethylolpropane mercaptopropionate (TRIS) from Evans Chemetics. It was added to a toluene solution of SBS at concentrations between 0.2 and 5% by weight of SBS. TRIS CH3 CH2 C CH2 O C CH2 H2 SH O CH2 O C CH2 CH2 SH O CH2 CH2 CH2 SHCO O C Among the various radical-type photoinitiators tested, acylphosphine oxides proved to be the most efficient ones to photocrosslink SBS samples. All the experiments have been carried out by using 2, 4, 6-trimethylbenzoyl-diphenyl- phosphine oxide (Lucirin TPO from BASF) as photo- initiator, at a typical concentration of 1 wt%. C P O O Lucirin TPO 2.2. Irradiation In a typical UV-curing experiment, 20mm thick films were cast from a toluene solution containing the SBS rubber, the thiol monomer and the photoinitiator onto either a KBr crystal for infrared analysis or a glass plate for insolubilization and hardness measurements. If a solvent- free process is required for industrial applications, the photoinitiator and the thiol can also be incorporated into the liquid resin heated at 1508C, the usual processing temperature. Samples were exposed at ambient temperature in the presence of air to the radiation of a 80 W/cm medium pressure mercury lamp (IST), at a passing speed of 60 m/ min which corresponds to an exposure duration of 0.1 s per pass. The maximum light intensity at the sample position was measured by radiometry (IL-390 light bug) to be 600 mW cm22 in the UV range. 2.3. Analysis The kinetics of the light-induced crosslinking of the rubber film was studied quantitatively by FTIR spectro- scopy, by following the decrease upon UV exposure of the absorption band characteristic of the vinyl double bond at 1827 cm21. The degree of conversion (x) was calcu- lated from the ratio of the corresponding IR absorbance before and after UV exposure (A0 and At) by using the following equation: x %† ˆ ‰1 2 At=A0†Š × 100 This value was not corrected for shrinkage which was found to be insignificant, based on the invariance of the C-H peak at 2960 cm21. It should be mentioned that even very small variations of the IR band intensity can be monitored accu- rately, because the analysis is performed on the same sample exposed to UV light for various durations, exactly on the same spot. The gel fraction and the degree of swelling of the irradiated polymer were determined by soaking the sample in toluene for one day at room temperature. The insoluble polymer was recovered by filtration and dried at 708C to a constant weight. The hardness of the coating was evaluated before and after irradiation by monitoring the damping of the oscillations of a pendulum (Persoz hardness). The C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-39123906 hardness was shown to be strongly dependent on the glass transition temperature [17], with Persoz values ranging typi- cally from 30 s for soft elastomeric materials to 300 s for hard and glassy polymers. The thermal stability of the various formulations was tested by heating the uncured sample for up to 1 h at 1508C, the processing temperature for hot-melt adhesive applications. 3. Results In a previous study on the photocrosslinking of SBS rubber [16], we have shown that the thiol-ene polymeriza- tion proceeds much more efficiently with the pendent vinyl double bonds arising from 1-2 polymerization of butadiene than with the backbone 2-butene double bonds of the poly- butadiene chain (by a factor 16). UV-curing experiments were therefore conducted on a specially designed SBS sample where as many as 59% of the polybutadiene unsaturations were vinyl double bonds. The formation of the polymer network in the elastomeric phase upon UV irradiation was followed through the vinyl group disappearance and the polymer insolubilization. 3.1. Loss of the vinyl group upon photocrosslinking of SBS The SBS samples containing 1 wt% Lucirin TPO and various concentrations of TRIS (between 0.2 and 5 wt%) were exposed to UV radiation for different times, up to 3 s. The crosslinking reaction was monitored through the decrease of the vinyl IR band at 1826 cm21. Fig. 1 shows the influence of the thiol concentration on the decay profile of the vinyl group upon UV irradiation. The initial sharp drop which occurs within the first 0.3 s was attributed to the copolymerization with the thiol which is rapidly consumed. At a TRIS concentration of 5 wt%, as much as 90% of the thiol was found, by IR spectroscopy (band at 2566 cm21), to have reacted after this short exposure, as shown in Fig. 2. As expected, the amount of vinyl groups polymerized [V] t after a given exposure (0.3 s) increases with the initial thiol content [SH]0 (Fig. 3). Vinyl groups are also under- going a radical induced homopolymerization reaction which can be quantified by substracting [SH]t from [V] t. Fig. 3 shows how the ratio copolymerization/homopolymeriza- tion, calculated from the ratio [SH]t/([V] t 2 [SH]t), increases with the thiol concentration. The fact that the vinyl-thiol copolymerization is as important as the vinyl homopolymerization, even though the SH concentration (0.2 mol kg21) is much lower than that of the vinyl groups (9.3 mol kg21), clearly shows that the propagating alkyl radicals (Pz) are much more reactive toward the thiol than toward the vinyl double bond. C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-3912 3907 Fig. 1. Influence of the thiol concentration on the vinyl group consumption upon UV exposure of the SBS/TRIS system. [LucirinTPO]ˆ 1 wt%. CH CH2 CH CH2 + RS RSH CH CH2 CH2 CH2 S R CH CH2 CH CH CH2 kp ks CH CH2 CH CH2 S R (P ) Copolymerization Homopolymerization CH2 CH CH2 S R The 2-butene double bonds of the polybutadiene chain do not contribute significantly to the crosslinking process as they were shown to be 16 times less reactive toward the thiyl radicals than the vinyl double bonds [16], and as they are in lower concentration. If allowance is made for them, the value of the copo/homo ratio in Fig. 3 would be roughly 10% lower. 3.2. Insolubilization Both reactions of the vinyl groups, homopolymerization and copolymerization with the thiol, generate crosslinks between the polybutadiene chains, with formation of a tri- dimensional network which is schematically represented in Fig. 4. As a result, the UV-exposed polymer becomes insoluble in organic solvents, like chloroform or toluene. Fig. 5 shows the insolubilization profiles obtained at various concentrations of the thiol crosslinker, between 0.5 and 4 wt%. At a TRIS concentration of 1 wt%, a single pass under the lamp at a speed of 60 m/min proved to be suffi- cient to get a 90% insoluble material. This value did not change much upon further exposure, even when the TRIS content was increased to 4 wt%. This limiting value suggests that the SBS sample contains about 5% non polymeric materials. In the absence of TRIS, a 30 times as long UV exposure (3 s) is needed to reach the same gel fraction, the vinyl consumption being evaluated to be only 2% of its original value, i.e. 0.18 mol kg21. This result suggests that the poly- merization of the pendent vinyl double bonds is not a very effective crosslinking process in this SBS. The most prob- able reason is that, because of the high vinyl content (59% of the polybutadiene units), polymerization proceeds prefer- entially between neighboring vinyl groups to form six- membered rings along the polymer chain which remains soluble. However, because the polybutadiene chain contains 41% butene-2 units, the cyclization process will propagate along the polymer chain over only a few adjacent vinyl C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-39123908 Fig. 3. Influence of the thiol concentration [SH]0 on the amount of vinyl and thiol groups polymerized after a 0.3 s UV exposure of the SBS/TRIS system. R S S S Fig. 4. Schematic representation of the photocrosslinked SBS/TRIS polymer network formed in the elastomeric phase. Fig. 5. Insolubilization profiles of UV-exposed SBS/TRIS films for differ- ent thiol concentrations. [LucirinTPO]ˆ 1 wt%. Fig. 2. Thiol conversion and vinyl loss in photocrosslinking of the SBS/ TRIS system. [Thiol]ˆ 5 wt%; [LucirinTOP]ˆ 1 wt%. units. Larger loops can still be formed by reaction of more distant vinyl double bonds located on the same polymer chain. Such an intramolecular process was already found to occur during the cationic polymerization of epoxidized polyisoprene at high epoxy contents (.50%) [3,4]. Cross- links will be formed and the polymer will become insoluble only if vinyl groups located on two different polybutadiene chains undergo polymerization. As a result, a very loose polymer network is generated, as shown by the high value of its swelling ratio: SRˆ weight of absorbed solvent/ weight of polymerˆ 50 for UV-cured SBS [8]. R CH2 CH CH2 CH R R R INTRA INTER In the presence of the trifunctional thiol, a branch point will be created each time a thiyl radical reacts with a vinyl double bond, which leads to a tightening of the polymer network. Indeed, the swelling ratio of the UV-exposed polymer was found to drop down to a value of 8 upon addition of as little a 1 wt% TRIS to SBS, as shown in Fig. 6. The SR value was cut by half when the TRIS concen- tration was further increased to 4 wt%. It should yet be noticed that, because of the high vinyl content of the SBS used, two of the three thiol groups of the TRIS molecule may typically react with two neighboring vinyl double bonds located on the same polymer chain (Fig. 4). Such an intramolecular process will have no net effect on the crosslink density, or on the polymer insolubilization. When all the thiol crosslinker has reacted, typically after 0.3 s UV exposure, the concentration of branch points will be equal to the initial concentration of the thiol groups, e.g., [SH]0ˆ 0.075 mol kg21 at [TRIS]ˆ 1 wt%. This corre- sponds to 1 attachment formed for every 124 vinyl double bonds, or to 10 attachments per SBS chain. Even if this number is reduced to 7 to take into account the above mentioned intramolecular process, it is still sufficient to get total insolubilization of the rubber. After this short UV exposure, crosslinking is mainly due to the thiol-ene polymerization, the homopolymerization of vinyl groups providing only a minor contribution, as shown by the low value of the gel fraction (8%) found in the neat SBS sample UV-exposed for 0.3 s. 3.3. Hardening While crosslinking is proceeding upon UV exposure, the amount of elastic chains which become bounded increases, and the hardness of the polymer rises consequently, as shown in Fig. 7. It is important to control precisely the crosslink density in order to obtain cohesion, but without hardening too much the polymer, which would be detri- mental for its adhesion. At a low thiol concentration (,1 wt%) and short exposure (#0.5 s), the UV-cured polymer remains still soft (Persoz value of less than 80 s) and keeps its elastomeric character. This is important if such rubbers are to be used as adhesives or as impact resistant materials to maintain the tack and wetting required to achieve a good adhesion. Harder but still flexible elastomers can be produced for flexographic printing plates or coatings applications, simply by increasing the thiol concentration up to 5 wt%, and/or the UV-exposure time up to 2 s. The influence of the thiol concentration on the insolubi- lization, swelling and hardening processes is illustrated in Fig. 8 for a SBS sample UV-exposed for 0.3 s in the presence of 1 wt% Lucirin TPO. It clearly appears that a TRIS concentration of 0.5 wt% is already enough to obtain a crosslinked elastomer showing the required characteristics for adhesive applications. The outer polystyrene segments of the block copolymer, which are phase separated, are not expected to participate to C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-3912 3909 Fig. 6. Influence of the thiol concentration and the exposure time on the swelling of a photocrosslinked SBS/TRIS polymer. [LucirinTPO]ˆ 1 wt%. Fig. 7. Variation of the polymer hardness upon UV-exposure, for various thiol contents of the SBS/TRIS system. [LucirinTPO]ˆ 1 wt%. the crosslinking process, except by a possible chain transfer reaction if some propagating radicals succeed in abstracting a labile hydrogen on the PS chain at the interphase. Increasing the PS chain length should not affect insolubili- zation, which is ensured through the polybutadiene cross- links, but it will increase the degree of swelling by lowering the crosslink density. 3.4. Influence of the photoinitiator concentration The photoinitiator (PI) plays a keyrole by controlling both the rate of initiation and the penetration of the incident light. An increase of the PI concentration will accelerate the cross- linking reaction, but it will also steepen the cure depth gradient in the UV exposed sample and lead to insufficient polymerization at the sample/substrate interface. This radia- tion inner filter effect is a critical factor, specially for thick samples, because it is often responsible of the poor adhesion of UV-cured coatings or adhesives. It is therefore recom- mended to work at low PI concentrations, but it will be at the expense of the polymerization speed, and of the cure extent as well. Only with highly reactive systems will it be possible to achieve a fast and complete deep-through cure. In this respect, the Lucirin TPO-TRIS-SBS system shows outstanding performance because crosslinking appears to proceed efficiently even at very low concentration in photoinitiator. Fig. 9 shows the insolubilization profiles obtained upon UV exposure of a SBS-TRIS (1 wt%) sample containing various amounts of Lucirin TPO, between 0.1 and 0.8 wt%. It is quite remarkable that the crosslinking reaction still proceeds very rapidly at these very low PI concentrations, insolubilization being achieved in less than 1 s for a Lucirin TPO content of 0.1 wt%. The polymer network formed is almost as tight as that obtained by operating at higher photo- initiator concentrations, as shown by the swelling curves of Fig. 10. After 1 s exposure, SR values of 12 and 8 were measured at PI concentrations of 0.1 and 0.8 wt%, respectively. An interesting feature was observed when the SBS-TRIS system was exposed to UV radiation without any added photoinitiator. After 1 s, as much as 85% of the polymer has become insoluble in toluene with formation of a rela- tively tight network (SRˆ 15), as shown in Figs. 9 and 10. This result suggests that the SBS sample contains some C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-39123910 Fig. 8. Influence of the thiol concentration on the photocrosslinking of the SBS/TRIS system. UV exposure: 0.3 s; [LucirinTPO]ˆ 1 wt%. Fig. 9. Influence of the photoinitiator concentration on the insolubilization of the SBS/TRIS system upon UV exposure. [TRIS]ˆ 1 wt%. Fig. 10. Influence of the photoinitiator concentration and the exposure time on the swelling of a photocrosslinked SBS/TRIS polymer. [TRIS]ˆ 1 wt%. Fig. 11. Variation of the polymer hardness upon UV exposure, for various photoinitiator concentrations in the SBS/TRIS system. [TRIS]ˆ 1 wt%. chromophores which absorb the UV radiation of l . 250 nm. The free radicals generated upon photolysis would then produce the propagating thiyl radicals by abstracting an hydrogen atom from the thiol molecule. As expected, the hardening of the polymer upon UV exposure becomes less pronounced as the photoinitiator concentration is decreased (Fig. 11). After a 1 s irradiation, the Persoz hardness value was found to increase from 40 to 55 s at [TPO]ˆ 0.1 wt%, and up to 85 s for [TPO]ˆ 0.8 wt%. The UV-cured elastomer remains soft and flexible, as required for adhesive applications. The influence of the photoinitiator concentration on the properties of the polymer material obtained after a 0.3 s UV exposure of the SBS-TRIS (1 wt%) system is shown in Fig. 12. It clearly appears that a TPO concentration of 0.1 wt% is already sufficient to get the expected performance. This formulation presents a number of advantages: • high reactivity upon UV-exposure; • no residual thiol and photoinitiator in the UV-cured rubber; • low cost of the tiny amounts of additives used; • low-modulus polymer well suited for adhesives applications; • deep through cure because of the low PI concentration; • curing of thick samples. For example, a 2 mm thick sample of SBS was rendered insoluble within less than 1 s, with formation of a relatively tight polymer network. Fig. 13 shows the insolubilization and the swelling profiles obtained with such a sample upon UV exposure in the presence of air, at ambient temperature. The % gel increase is slower than in thin films, as might be expected from the inner filter effect which reduces the amount of radiation received by the deep-lying layers. More surprising is the fact that the swelling ratio of the thick UV-cured sample was found to be lower than that of the thin film (7 and 11, respectively), thus indicating that a tighter polymer network has been performed. This could be due to a less pronounced inhibitory effect of atmospheric oxygen in the thick sample. An additional reason could be the larger rise in temperature for the thick sample caused by the exothermal polymerization, as already shown in UV- curable resins [17]. The resulting increase in molecular mobility would favour intermolecular reactions and lead to a higher crosslink density. 3.5. Thermal stability For some applications, like hot-melt adhesives and flexo- graphic printing plates, the highly reactive SBS/thiol system is to be processed at elevated temperatures (usually 1508C). It is therefore important to assess the heat resistance of the formulation before UV-curing. In a previous study on a SBS formulation containing 5 wt% TRIS and 1 wt% Lucirin TPO, we found that about one-third of the sample has become insoluble in toluene after a 10 min heating at 1508C [16]. This poor thermal stability makes this system unfit for hot-melt adhesive applications. We have repeated this study with a formulation contain- ing only 0.5 wt% TRIS and 1 wt% Lucirin TPO. Under C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-3912 3911 Fig. 12. Influence of the photoinitiator concentration on the photocrosslink- ing of the SBS/TRIS system. UV exposure: 0.3 s; [TRIS]ˆ 1 wt%. Fig. 13. Photocrosslinking of a 2 mm thick SBS/TRIS sample. [LucirinTPO]ˆ 0.1 wt%; [TRIS]ˆ 1 wt%. Fig. 14. Insolubilization of a SBS/TRIS upon heating at 1508C. [LucirinTPO]ˆ 1 wt%; [TRIS]ˆ 0.5 wt%. ---: SBS without thiol. those conditions, the vinyl content remained unchanged and no gelation was found to occur during the first 20 min of heating, very much like for neat SBS (Fig. 14). This formu- lation appears thus to meet the required specifications for heat resistance. The formation of insoluble polymer, which was observed upon further heating, could probably be delayed by the addition of an antioxidant, which should not interfere with the UV-curing process. Interestingly, the thermooxidation of SBS, which can be quantitatively followed by monitoring the formation of carbonyl and hydroxyl groups by IR spectroscopy, was shown to be depressed in the presence of the TPO and TRIS additives. These oxidation products are produced in significant amounts only after a 30 min heating at 1508C, as shown in Fig. 15. 4. Conclusion The photoinduced thiol-ene polymerization has been successfully applied to crosslink at ambient temperature a styrene-butadiene rubber with a high vinyl content (59%). A very low amount of thiol crosslinker (#1 wt%) proved to be already enough to achieve total insolubilization within less than 1 s upon intense illumination in the presence of a acylphosphine oxide photoinitiator. Crosslinking occurs predominantly by reaction of the thiyl radicals with the pendent vinyl double bonds and, to some extent, by homo- polymerization of the vinyl groups. Hardening is hardly taking place upon UV-curing, so that the thermoplastic elastomer retains its viscoelastic properties, while becoming more resistant to heat and chemicals. The photoinitiator concentration can be lowered down to 0.1 wt%, thus allowing relatively thick samples to be rapidly crosslinked because of the increased penetration of UV radiation. This UV-curable formulation exhibits a sufficiently high resistance to heating to allow its use in hot-melt adhesive and flexographic printing plate applica- tions. Further studies on the photocrosslinking of the thiol/ SBS combination are in progress, aiming at lowering the SBS vinyl content down to 8%, to be able to cure rapidly commercial styrene-butadiene rubbers. Acknowledgements The authors wish to thank SHELL RESEARCH (Louvain-la-Neuve-Belgium) for a research grant. References [1] Decker C. Progr Polym Sci 1996;21:593. [2] Crivello JV, Yang B. J Macromol Sci Chem 1994;A31:517. [3] Decker C, Le Xuan H, Nguyen Thi Viet T. J Polym Sci, Polym Chem Ed 1995;33:2759. [4] Decker C, Le Xuan H, Nguyen Thi Viet T. J Polym Sci, Polym Chem Ed 1996;34:177. [5] Le Xuan H, Decker C. J Polym Sci, Polym Chem Ed 1993;31:769. [6] Decker C, Nguyen Thi Viet T, Le Xuan H. Europ Polym J 1996;32:549-59. [7] Huber HF. In: Fouassier JP, Rabek JF, editors. Radiation curing in polymer science and technology, 4. London: Elsevier Applied Science, 1993. p. 51. [8] Decker C, Nguyen Thi Viet T. Macromol Chem Phys 1999;200:358. [9] Dupont M, De Keyser N. WO Patent 9502, 640, 1995. [10] Dupont M, De Keyser N. Proceedings of RadTech Europe Conference, Maestricht, 1995;174. [11] Mayenez C, Muyldermans X. US Patent 0696 761, 1996. [12] Hein PR, Evans JA, Yang MW. US Patent 1980;4:234,676. [13] Rice CS, Sanaki Y, Plamthottan SP. US Patent, 1992;5:166,226. [14] Morgan CR, Magnolta F, Ketley AD. J Polym Sci, Polym Chem Ed 1977;15:627. [15] Jacobine AF. In: Fouassier JP, Rabek JF, editors. Radiation curing in polymer science and technology, 3. London: Elsevier Applied Science, 1993. p. 219. [16] Decker C, Nguyen Thi Viet T. Macromol Chem Phys 1999;200:1965. [17] Decker C, Decker D, Morel F. In: Scranton AB, Bowman CN, Peiffer RW, editors. ACS Symp.Series, 673. Washington, DC: ACS, 1997. p. 63. C. Decker, T. Nguyen Thi Viet / Polymer 41 (2000) 3905-39123912 Fig. 15. Thermal oxidation of the SBS/TRIS system upon heating at 1508C. [LucirinTPO]ˆ 1 wt%; [TRIS]ˆ 0.5 wt%. ---: SBS without thiol. Macromol. Chem. Phys.200,1965-1974 (1999) 1965 Photocrosslinking of functionalized rubbers, 8a) The thiol-polybutadiene system Christian Decker*, Trieu´ Nguyen Thi Viet Département de Photochimie (UMR - CNRS N87525), Ecole Nationale Supérieure de Chimie de Mulhouse - Université de Haute Alsace 3, rue A. Werner - 68200 Mulhouse, France (Received: October 6, 1998; revised: January 13, 1999) SUMMARY: The light-induced crosslinking of polystyrene-block-polybutadiene-block-polystyrene (SBS) was carried out in the presence of a trifunctional thiol and a phosphine oxide photoinitiator. The curing pro- cess was followed by infrared spectroscopy, insolubilization and hardness measurements, and was shown to proceed extensively within less than 1 s under intense illumination in air at ambient temperature. The relative reactivity of the thiyl radicals toward the polybutadiene vinyl and butene double bonds was determined, as well as the competition between homo- and copolymerization for the propagating alkyl radical. An increase of the SBS vinyl content from 8 to 59% has little effect on the crosslinking process, because it enhances mainly intramolecular reactions. The thiol/ene polymerization is much more effective to crosslink thermopla- stic SBS elastomers than the copolymerization of the polybutadiene double bonds with a diacrylate mono- mer. Introduction The crosslinking of vinyl-functionalized polymers can be readily achieved at ambient temperature by UV-irradia- tion in the presence of a free radical-type photoinitia- tor1-4). Most of these studies were performed on styrene- butadiene-styrene (SBS) block copolymers to make such thermoplastic elastomers more resistant to temperature and organic solvents by creating covalent bonds between the polybutadiene chains. In a previous paper of this ser- ies5), we have shown that a styrene-butadiene rubber hav- ing only 8% of its unsaturations as pendent vinyl double bonds can be made totally insoluble within a few seconds upon UV exposure in the presence of air. The addition of a diacrylate monomer, which copolymerizes with the vinyl and butene double bonds, was found to accelerate markedly the crosslinking process, but at the expense of the heat resistance of the plasticized rubber. This highly reactive photosensitive material is therefore ill-suited for hot-melt adhesive applications or flexographic printing plates requiring processing at 1508C. In the present study, another approach has been used to achieve a rapid photocrosslinking of SBS rubber. It is based on the photoinduced addition of a thiol (RSH) onto an olefinic double bond, and it requires incorporation of a multifunctional thiol into the polyene elastomer. A great number of kinetic and mechanistic studies have been devoted to thiol-ene photopolymerization, a topic which has been recently reviewed by Jacobine in a comprehen- sive monography6). The chain reaction proceeds by a step growth addition mechanism which is propagated by a chain transfer reaction involving thiyl radicals (RS9)7, 8). The overall process can be formally represented as fol- lows: Crosslinking requires the association of a diene or polyene and of a multifunctional thiol. Because peroxy radicals formed by the O2 scavenging of alkyl radicals are also capable of abstracting hydrogen atoms from the thiol, UV-curable thiol-ene systems are much less sensi- tive to air inhibition9) than conventional radical-induced polymerization10): Macromol. Chem. Phys.200, No. 8 i WILEY-VCH Verlag GmbH, D-69451 Weinheim 1999 1022-1352/99/0808-1965$17.50+.50/0 a) Part 7: cf. ref.5) 1966 C. Decker, T. Nguyen Thi Viet There are only a few reports in the literature on the photocrosslinking of thiol-SBS system11, 12), with little information on the reaction kinetics. Like in our previous study5), the main objective of this work was to determine how fast and how extensively styrene-butadiene rubbers could be crosslinked by UV-irradiation at ambient tem- perature in the presence of a thiol monomer. Special attention has been given to the influence on the cure kinetics of the three key components: the photoinitiator, the thiol monomer and the SBS vinyl double bonds. The progress of the crosslinking reaction has been followed through the insolubilization and hardening of the UV-irra- diated sample, and by monitoring the thiol and vinyl groups disappearance by IR spectroscopy. Experimental part Materials Two types of polystyrene-block-polybutadiene-block-poly- styrene (SBS), both from SHELL, have been used in this study. They differ only by their content of pendent vinyl groups: 8% of the total double bond content for the low- vinyl polymer (LV-SBS), and 59% for the high-vinyl poly- mer (HV-SBS). The 2-butenylene units (1CH21CH2 CH1CH21) located on the polybutadiene backbone repre- sent 92% and 41% of the unsaturation content, respectively. In some experiments, a paraffinic oil (Nujol from Aldrich) was added to the formulation to increase the molecular mobility in the polymer film. A trifunctional thiol was selected as crosslinking agent, trimethylolpropane tris- (3-mercaptopropionate) (TRIS) from Evans Chemetics. It was added to a toluene solution of SBS at a concentration between 3 and 20 wt.-% of SBS. Different types of photoinitiators were introduced in the polymer, at a typical concentration of 2 wt.-% SBS, namely benzophenone (BZP from Aldrich), isopropylthioxanthone (Quantacure ITX from BioSynthetics), 4,49-diphenoxybenzo- phenone (DPB from SNPE) and 2,4,6-trimethylbenzoyl(di- phenyl)phosphine oxide (Lucirin TPO from BASF). Scheme 1 shows the chemical formulas of the various com- pounds used in this study. In a typical UV-curing experiment, 20lm thick films were cast from a toluene solution containing the SBS rubber, the thiol monomer and the photoinitiator onto either a KBr crys- tal for infrared analysis or a glass plate for insolubilization and hardness measurements. Samples were exposed to the radiation of a 80 W/cm medium pressure mercury lamp, in the presence of air, at a passing speed of 60 m/min, which corresponds to an exposure duration of 0.1 s per pass. The maximum light intensity at the sample position was meas- ured by radiometry (IL-390 light bug) to be 600 mWN cm-2 in the UV range. Analysis The kinetics of the light-induced crosslinking of the rubber film was studied quantitatively by FTIR spectroscopy, by following the decrease upon UV exposure of the absorption bands characteristic of the vinyl double bond at 910.8, 1637.8 and 1827 cm-1, and of the but-2-ene double bond at 964.7 cm-1. The disappearance of the thiol monomer upon UV-exposure was followed through the IR absorbance of the SH group at 2569 cm-1. The degree of conversion (x) was calculated from the ratio of the corresponding IR absorbance before and after UVexposure (A0 andAt) by using the follow- ing equation: x (%) = [1 - (At /A0)] N 100 This value was not corrected for shrinkage, as it was found to account for less than 2%, based on the variation of the C1H peak at 2960 cm-1. The loss of the photoinitiator in the UV-irradiated sample was followed quantitatively by UV- spectroscopy. The gel fraction and the degree of swelling of the irra- diated polymer were determined by soaking the sample in toluene for one day at room temperature. The insoluble poly- mer was recovered by filtration and dried at 708C to a con- stant weight. The hardness of the coating was evaluated before and after irradiation by monitoring the damping of the oscillations of a pendulum (Persoz hardness). The hardness was shown to be strongly dependent on the glass transition temperature13), with Persoz values ranging typically from 30 s for soft elastomeric materials to 300 s for hard and glassy polymers. The thermal stability of the various formu- Scheme 1: Chemical formulas of compounds used Photocrosslinking of functionalized rubbers, 8 1967 lations was tested by heating the uncured sample for up to 1 h at 1508C, the processing temperature for hot-melt adhe- sive applications. Results The thiol-ene polymerization is known to proceed more efficiently with olefins in which the double bond is located at the end of the monomer molecule (R1CH 2CH2) rather than in its backbone (R1CH2CH1R)6). Consequently, a faster crosslinking was expected to occur with SBS rubbers containing a high content of pendent vinyl double bonds arising from the 1-2 polymerization of butadiene. Our first UV-curing experiments were therefore conducted with a specially designed SBS sam- ple where as many as 59% of the polybutadiene unsatura- tions were vinyl double bonds (HV-SBS). Influence of the photoinitiator on the photocrosslinking of the thiol-SBS system The photoinitiator (PI) used in UV-curing applications can be classified into two major categories, depending on the way the free radicals are generated: - by photocleavage of the PI excited states - through hydrogen abstraction by the PI excited states As thiols exhibit a strong hydrogen donor character, the second type of photoinitiator is often used in the thiol-ene photopolymerization. The benzoyl radical produced by the photocleavage reaction is also capable to abstract a hydrogen atom from the thiol to generate the propagating thiyl radical6): Among the different types of photoinitiators examined, the following four were found to give the best perfor- mance: benzophenone (BZP), isopropylthioxanthone (ITX), diphenoxybenzophenone (DPB) and 2,4,6-tri- methylbenzoyl(diphenyl)phosphine oxide (Lucirin, TPO). Fig. 1 shows how the degree of conversion of the thiol monomer varies with the exposure time for a HV-SBS rubber containing 2 wt.-% photoinitiator and 20 wt.-% TRIS, which corresponds to a double bond/SH molar ratio of 8.3. The phosphine oxide proved to be the most efficient initiator and was therefore used in our further studies. One may still notice that TPO is 10 times more efficient than benzophenone in the absence of thiol (Fig. 1 of ref.5)), but only 1.6 times in the presence of thiol (Tab. 1). It is quite remarkable that the polymerization of the thiol proceeds so rapidly in the solid state, reaching 50% conversion within 0.5 s exposure. The slowing down observed upon further irradiation was attributed to two factors: (i) a fast consumption of the photoinitiator and of the thiol and (ii) molecular mobility restrictions resulting from the build up of the tridimensional polymer network. A similar ranking of the photoinitiators was found by following the disappearance of the vinyl double bond of SBS (Fig. 1). Here again, the reaction slows down after 0.5 s and the vinyl content is leveling off at around 60% of its original value after 3 s. At that final stage of the polymerization, the absolute amount of vinyl groups which have disappeared is about twice that of the SH groups consumed. This result suggests that a significant amount of vinyl double bonds undergoes homopolymeri- zation, besides their copolymerization with the trifunc- tional thiol. Both processes will lead to crosslinking and therefore insolubilization of the polymers, if the vinyl double bonds are located on different SBS chains. Fig. 2 shows the insolubilization profiles of the HV-SBS/TRIS system exposed to UV-light in the presence of 2 wt.-% photoinitiator. With the most efficient initiator (Lucirin TPO), 80% insolubilization was achieved within 0.1 s. At that stage of the reaction, 5% of the vinyl double bonds have polymerized, i.e., 70 groups per SBS chain. As inso- lubilization requires the build up of only a few crosslinks (at least 3 per chain), we are left to conclude, like in our previous study on the UV-curing of the same SBS sam- Fig. 1. Influence of the photoinitiator (2 wt.-%) on the photo- crosslinking of the HV-SBS/Tris system. [Tris] = 20 wt.-%. Light intensity: 600 mWN cm-2 1968 C. Decker, T. Nguyen Thi Viet ple5), that the polymerization involves mainly neighbor- ing vinyl groups located on the same polybutadiene chain. Such an intramolecular process seems feasible in consideration of the large number of pendent vinyl groups in HV-SBS (6 for every 10 monomer units). Lucirin TPO proved to be also the best photoinitiator for reaching rapidly a high crosslink density, as shown by the swelling profiles of Fig. 2. A swelling ratio of 5 was measured for the sample UV-irradiated during 0.1 s in the presence of this phosphine oxide, compared to values ranging from 7 to 9 with the other photoinitiators. These values are less than half those measured in UV-cured SBS5), thus show- ing that a tighter polymer network has been formed upon UV-irradiation of the thiol-SBS system. A performance analysis of the photoinitiators studied in UV-curing of the TRIS/HV-SBS combination (20/80 weight ratio) is reported in Tab. 1. The two sets of results (rate of poly- merization and insolubilization data) correlate perfectly well for the 4 compounds. Influence of the SBS vinyl content The photoinduced thiol-ene polymerization of two SBS samples having very different vinyl group contents (59 and 8%) has been studied in order to evaluate the relative reactivity of vinyl and butene double bonds, and to quan- tify the competition between the homopolymerization of vinyl double bonds and the transfer copolymerization with the trifunctional thiol. The overall crosslinking pro- cess can be represented by the following reaction scheme. Thiyl radicals are also produced by hydrogen abstrac- tion on the thiol by any of the secondary free radicals formed by the addition process, as shown in the introduc- tion reaction scheme. While vinyl groups undergo both homopolymerization and copolymerization with the thiol, the butene double bonds are only reacting with the thiyl radicals. Indeed, homopolymerization of the butene group was not observed upon UV exposure of SBS in the absence of thiol5). Crosslinking results from both the copolymerization of the butadiene double bonds with the trifunctional thiol and the homopolymerization of the pendent vinyl groups. The structure of the polymer network obtained is repre- sented schematically in Fig. 3. Photocrosslinking of HV-SBS rubber The SBS sample containing 59% pendent vinyl double bonds has been exposed to UV-radiation in the presence of the trifunctional thiol ([TRIS] = 20 wt.-%) and Lucirin TPO (2 wt.-%). Infrared spectra were recorded after var- Tab. 1. Influence of the photoinitiator on the photocrosslinking kinetics of the thiol/HV-SBS system. [Tris] = 20 wt.-%. Light inten- sity: 600 mWN cm-2 Photoinitiator (2 wt.-%) Rs molNkgÿ1 Nsÿ1 a† Rv molNkgÿ1 Nsÿ1 b† Gel in %c) Swelling ratioc) Benzophenone (BZP) 1.5 2.8 64 9 Isopropylthioxanthone (ITX) 1.8 3.0 69 8 Diphenoxybenzophenone (DPB) 2.0 3.3 78 7 Lucirin (TPO) 2.4 3.6 83 5 a) Initial rate of thiol consumtion. b) Initial rate of vinyl consumption. c) After 0.1 s UVexposure. Fig. 2. Influence of the photoinitiator (2 wt.-%) on the photo- crosslinking of the HV-SBS/Tris system. [Tris] = 20 wt.-%. Light intensity: 600 mWN cm-2 Photocrosslinking of functionalized rubbers, 8 1969 ious exposure times (up to 2 s) to monitor the disappear- ance of the three functional groups: thiol, vinyl and butene. The butene double bond was found to be much less reactive than the vinyl double bond in the ene-thiol polymerization. Its initial rate of copolymerization with the thiol Rb was measured to be 0.1 molN kg-1 N s-1, com- pared to 2.3 molN kg-1 N s-1 for the vinyl copolymerization rate (Rv)copo(Tab. 2). The relative reactivity of the two types of double bonds toward the trifunctional thiol can be determined from the ratio of the rate constantskv andkb of the addition reac- tion of the thiyl radical to the vinyl and butene double bonds, respectively. (Rv)copoˆ kv [RS9] [vinyl] (1) Rb ˆ kb [RS9] [butene] (2) kv kb ˆ Rv†copo Rb N ‰buteneŠ ‰vinylŠ ˆ 16 3† The higher reactivity of the vinyl double bond toward thiols is in good agreement with previous structure-reac- tivity studies of the addition of thiols on olefins14). For instance,a-methylstyrene was found to be 25 times as reactive toward thiol glycolate asa,b-dimethylstyrene15). The enhanced reactivity of the vinyl group can be attribu- ted to an increase in electron density of the olefin which is reacting with a relatively electrophilic thiyl radical16). An additional factor in HV-SBS is the poor accessibility of the backbone butene double bonds by thiyl radicals. From the data shown in Fig. 1, one can evaluate the competition which exists for the vinyl radical between homopolymerization and copolymerization with the thiol. The rate equation for the thiol consumption can be written as: Rs = ks [P9] [RSH] (4) where P9 are the alkyl radicals produced by the addition reaction of the thiyl radical to the double bond (consump- tion of the thiol by reaction with the photoinitiator was neglected). Vinyl groups are consumed by reaction with both P9 radicals and thiyl radicals, which leads to the rate equation: Rv = kp [P9] [vinyl] + kv [RS9] [vinyl] (5) As the thiyl free radical is assumed to react only with the SBS double bonds to generate a copolymer unit, one can write: ks [P9] [RSH] = [RS9] (kv [vinyl] + kb [butene]) (6) By neglectingkb [butene] which is small in regard ofkv [vinyl] (a 1/23 ratio), the rate equation of the vinyl con- sumption becomes: Rv = kp [P9] [vinyl] + ks[P9] [RSH] (7) The competition between homopolymerization and copolymerization can then be quantified from the ratio Rv/Rs: Rv Rs ˆ kp‰vinylŠ ‡ ks‰RSHŠ ks‰RSHŠ 8† Fig. 3. Chemical structure of the photocured SBS/Tris polymer network Tab. 2. Performance analysis of the photocrosslinking of styr- ene-butadiene rubbers LV-SBS and HV-SBS in the presence of a trifunctional thiol (20 wt.-%). Photoinitiator: [Lucirin TPO] = 2 wt.-% LV-SBS HV-SBS Vinyl content in % 8 59 Ene/thiol molar ratio 8.4 8.4 Initial loss rate in molN kg-1 N s-1: Thiol Rs 1.8 2.4 Vinyl Rv 0.83 3.6 (Rv)homo 0.06 1.3 (Rv)copo 0.77 2.3 Butene Rb 0.9 0.1 Relative reactivity: Vinyl group copo/homopolymerization:ks/kp - 10 Copolymerization vinyl/butene:kv/kb 10 16 Insolubilization rate in %N s-1 80 90 Swelling degree (after 0.1 s) 6 5 1970 C. Decker, T. Nguyen Thi Viet which leads to kp ks ˆ Rv Rs ÿ 1   ‰RSHŠ ‰vinylŠ   9† In the photocrosslinking of HV-SBS, the following values were obtained:Rv = 3.6 molN kg-1 N s-1, Rs = 2.4 mol N kg-1 N s-1 and [RSH]/[vinyl] = 0.2. The ratioks/kp was calculated from Eq. (9) to be as high as 10, which means that the propagating alkyl radicals P9 are 10 times as reactive toward the thiol group as they are toward the vinyl double bond, in agreement with previous observa- tions6). It should be mentioned that the value of the vinyl homopolymerization rate in the thiol-SBS system, calcu- lated from the equationRhomo = Rv-Rs, is quite similar to the rate of polymerization of vinyl groups in neat SBS photocrosslinked under the same conditions5), 1.2 and 1.0 mol N kg-1 N s-1, respectively. The network formation is not only making the rubber insoluble in organic solvents, but it also increases the Shear Adhesion Failure Temperature (SAFT) from 80 to 1608C, and the hardness of the UV-exposed polymer as well, as shown in Fig. 4. Because hardening occurs mainly upon prolonged UV irradiation, for up to 2 s, it is still possible to obtain a soft and insoluble polymer by shortening the expo- sure time down to 0.3 s. Another way to get a crosslinked polymer showing a strong elastomeric character for adhe- sive or flexographic applications is by introducing a plasti- cizer in the thiol-SBS formulations, such as Nujol, a paraffi- nic oil. At a weight concentration of 35%, insolubilization of SBS was found to proceed as fast as for the unplasticized sample, but the crosslinked polymer remained soft upon UVexposure for up to 2 s, as shown in Fig. 4. Photocrosslinking of LV-SBS rubber The same study was carried out with a SBS sample con- taining only 8% vinyl group (LV-SBS), a value which is more typical of commercial styrene-butadiene thermo- plastic elastomers. In the presence of 20 wt.-% TRIS, the insolubilization, swelling and hardening profiles (Fig. 5) were found to be very similar to those obtained with the HV-SBS sample. The same behavior was previously observed in the photocrosslinking of neat SBS copoly- mers5). This surprising result suggests that, in HV-SBS, a substantial fraction of the vinyl groups disappear through intramolecular processes involving neighboring pendent double bonds, without any net effect on the insolubiliza- tion and hardening. By contrast, in LV-SBS such intramo- lecular reactions are less likely to occur, given the aver- age distance separating the pendent vinyl double bonds (12 monomer units). Crosslinking will therefore proceed readily through intermolecular reactions between func- tional groups located on different polymer chains. The initial rate of thiol consumption in LV-SBS (Rs = 1.8 molN kg-1 N s-1) was found to be superior to the initial rate of vinyl consumption (Rv = 0.83 molN kg-1 N s-1), most probably because the thiyl radicals react with the backbone butene double bonds, which are present in lar- ger amounts and are more accessible than in HV-SBS. From these data and by using theks/kp ratio determined in the previous section, one can evaluate the relative reactiv- ity of the thiyl radicals toward the vinyl and butene dou- ble bonds of polybutadiene in LV-SBS. Vinyl groups are disappearing by both homopolymeri- zation and copolymerization with the thiol, so that one can write: Rv = (Rv)homo + (Rv)copo (10) The contribution of each process can be evaluated by the equation Rv†copo Rv†homo ˆ ks‰RSHŠ kp‰vinylŠ 11† Fig. 4. Hardening of the HV-SBS/Tris system upon UV expo- sure. [Tris] = 20 wt.-%; [Lucirin TPO] = 2 wt.-% Fig. 5. Photocrosslinking of the LV-SBS/Tris system. [Tris] = 20 wt.-%; [Lucirin TPO] = 2 wt.-% Photocrosslinking of functionalized rubbers, 8 1971 From the value of the ratioks/kp = 10, evaluated pre- viously, and the thiol/vinyl molar ratio (1.5), one can determine the rate of copolymerization of the vinyl group: (Rv)copo= 0.933Rv = 0.77 molN kg-1 N s-1 As the thiol is consumed by reaction with both the vinyl and butene double bonds, one can write: Rs = (Rv)copo+ (Rb)copo (12) Butene double bonds do not undergo significant homo- polymerization upon UV exposure of LV-SBS in the pre- sence of Lucirin TPO, so that (Rb)copo X Rb. In the pre- sence of 20 wt.-% TRIS, they disappear at an initial rate of 0.9 molN kg-1 N s-1, a value in good agreement with that calculated from Eq. (12): Rb = 1.8-0.77 = 1.03 molN kg-1 N s-1 The relative reactivity of thiyl radicals toward vinyl and butene double bonds in LV-SBS was calculated from the ratio of the two addition rate constants: kv kb ˆ Rv†copo Rb 6 ‰buteneŠ ‰vinylŠ 13† ˆ 0:77 0:9 611:7 ˆ 10 This value confirms that thiyl radicals are much more reactive toward the pendent vinyl double bonds than toward the backbone butene double bonds. The larger value found in HV-SBS (kv/kb = 17) may result from a decrease ofkb caused by the more difficult access by thiyl radicals of the butene double bonds squeezed between the many pendent vinyl groups. In Tab. 2, we have summarized our kinetic data on the photocrosslinking of the two types of SBS rubber studied, in the presence of 20 wt.-% TRIS and 2 wt.-% Lucirin TPO. Influence of the thiol concentration All the photocrosslinking experiments reported so far have been carried out with SBS samples containing the trifunctional thiol at a weight concentration of 20%, which corresponds to a double bond/SH molar ratio of 8.4. As insolubilization of polymers requires the creation of only a few bridges between the polymer chains, we have reduced the content of TRIS down to 10 wt.-% and 5 wt.-%, i.e., ene/thiol ratios of 19 and 40, respectively. Fig. 6 shows the vinyl loss and thiol conversion profiles for the two HV-SBS samples exposed to UV radiation for up to 2 s in the presence of 1 wt.-% Lucirin TPO. The crosslinking reaction appears to proceed as efficiently as at the higher thiol concentration, thus leading to a rapid insolubilization of the SBS rubber upon UV irradiation (Fig. 7). A 0.1 s exposure proved to be sufficient to get a 90% insoluble polymer showing nearly the same value of the swelling ratio (6) as the polymer photocured with 20 wt.-% TRIS (swelling ratio of 5). Lowering the thiol concentration down to 5 wt.-% has therefore no detrimental effect on the cure kinetics and on the crosslinking extent of the SBS rubber. Besides reduc- ing the cost of the formulation, it will also lessen some of the disadvantages associated with the use of sulfur com- pounds, in particular the bad smell. In this respect, it should be emphasized that the photocured polymer shows Fig. 6. Influence of the thiol content on the photocrosslinking of the HV-SBS/Tris system. [Lucirin TPO] = 1 wt.-%. Thiol con- version (0), vinyl loss (h) Fig. 7. Influence of the thiol content on the photocrosslinking of the HV-SBS/Tris system. [Lucirin TPO] = 1 wt.-%. Gel frac- ion (0, h); swelling ratio (9, H) 1972 C. Decker, T. Nguyen Thi Viet no unpleasant odor anymore, because essentially all the thiol groups have reacted and are chemically bonded to the polymer network. The few unreacted SH groups are still linked to the polymer because of the trifunctional character of the mercaptan used. Very similar results have been obtained with the LV- SBS sample UV-irradiated in the presence of 3 wt.-% TRIS and 1 wt.-% Lucirin TPO. Here again crosslinking of this thermoplastic elastomer was found to proceed readily, to yield a tight tridimensional polymer network. Fig. 8 shows the polymerization, insolubilization, harden- ing and swelling profiles of the UV-exposed sample. Insolubilization was achieved within 0.1 s at a vinyl con- sumption of 5%, i.e., 0.05 mol/kg SBS, which corre- sponds to 8 bridges per polybutadiene chain. Nearly the same amount of crosslinks was produced by reaction of the thiol with the butene double bonds, which leads to an average molecular weight between crosslinks of 7000. The drastic effect of the trifunctional thiol on the photocrosslinking of SBS is illustrated in Fig. 9, which shows the variation of the sol fraction and the hardness with exposure time, for both systems. Insolubilization required 20 times less energy than in the thiol-free SBS, with formation of a much harder crosslinked polymer. Moreover, the sharp drop of the swelling ratio of UV- cured SBS, from 25 to 4 in the presence of TRIS, con- firms that a much more efficient crosslinking process is taking place in the thiol/polybutadiene system. In the presence of 5 wt.-% TRIS, insolubilization was found to occur not only much faster than in the neat rub- ber, but even faster than in an SBS sample containing 20 wt.-% of a very reactive diacrylate monomer5), as shown in Fig. 10. This result was unexpected, based on the monomer functionality:f = 3 for the thiol andf = 4 for the diacrylate. Indeed, a linear polymer will be formed by using either a monoacrylate (f = 2) or a difunctional thiol (f = 2) associated to a diene, because the latter chain reac- tion proceeds by a step growth addition. If the thiol com- pound contains only 1 SH group, there will be no poly- merization and no crosslinking, but simply an addition of the thiol to the SBS double bond, as shown by the reac- tion scheme given in the introduction section. Consequently, the copolymerization of a diacrylate (or a telechelic acrylate) with the SBS double bonds will gen- erate 4 branch points, instead of only 3 in the case of the trifunctional thiol. The lower crosslink density actually found in the SBS/acrylate system (swelling ratio of 12) was attributed to the competitive homopolymerization of Fig. 8. Photocrosslinking of LV-SBS in the presence of 3 wt.-% of Tris. [Lucirin TPO] = 1 wt.-%. Vinyl loss (0); gel frac- tion (f); swelling ratio (f) and hardness (h) Fig. 9. Influence of a trifunctional thiol (3 wt.-%) on the photocrosslinking of LV-SBS. [Lucirin TPO] = 1 wt.-%. Soluble fraction (0, 9) ; hardness (h, H) Fig. 10. Photocrosslinking of HV-SBS, neat (f), in the pre- sence of a diacrylate (20 wt.-%) (h), or of a trifunctional thiol (5 wt.-%) (0). [Lucirin TPO] = 1 wt.-% Photocrosslinking of functionalized rubbers, 8 1973 the acrylate double bonds, which does not generate any additional crosslinks between SBS chains. Heat resistance of UV-curable SBS If these photocurable thermoplastic elastomers are to be used as hot-melt adhesives, they must exhibit a fair heat resistance at the processing temperature, usually 1508C. A 20 lm thick film of HV-SBS containing 5 wt.-%, TRIS and 1 wt.-% Lucirin TPO was heated at 1508C for up to 30 min. The thiol content was found to drop rapidly dur- ing this thermal treatment, 50% loss after 5 min, as shown in Fig. 11. This loss is mainly due to volatilization of the liquid TRIS out of the thin film, rather than to a chemical reaction, as indicated by the following three observations: - the thiol disappears as rapidly in an SBS film contain- ing no photoinitiator, - the thiol loss is much reduced in thick SBS samples, - gelation of SBS does not occur during the first 5 min of heating when half of the TRIS is lost. After 10 min heating in the dark, one third of the sam- ple has become insoluble in toluene (Fig. 12), which makes this very reactive system unfit for hot-melt adhe- sive applications. In the absence of photoinitiator, gela- tion started to proceed 5 min later, but insolubilization did follow a similar profile as in the presence of Lucirin TPO, as shown in Fig. 12. In a previous study, we found neat SBS to be stable for over 20 min at 1508C5). Reduc- ing the thiol content and adding an antioxidant might therefore help in improving the heat resistance of the photocurable thiol/SBS formulation. Conclusion The light-induced thiol/ene polymerization is a very effective method to crosslink rapidly styrene-butadiene block copolymers at ambient temperature. Under intense illumination, the curing reaction proceeds within a frac- tion of a second by a step growth addition mechanism to yield an insoluble polymer network. Both the butene and the vinyl double bonds of the polybutadiene chain are attacked by the thiyl radicals, which are at least 10 times more reactive toward the pendent vinyl groups. It is how- ever not necessary to increase the vinyl content of SBS above 8% because it will mainly favor intramolecular homo- and copolymerization reactions between neighbor- ing vinyl double bonds and will not accelerate the net- work formation. This result is important from an econom- ical point of view because it makes this UV-technology directly applicable to commercial SBS rubbers which have generally a low vinyl content. Another interesting feature is that the thiol content can be reduced down to 5 wt.-%, without detrimental effect on the cure speed. The main applications of this photocurable rubber are expected to be found in industrial sectors where spatial control and cure speed are a major concern, such as for the manufacturing of protective coatings, adhesives, photoresists and printing plates. Further studies on the photocrosslinking of the thiol/SBS combination are in progress, aiming at lowering both the thiol and the photo- initiator content, to make this system even more econom- ically attractive, more resistant to heat, and to extend its field of applications to thick samples and composite materials. Acknowledgement: The authors wish to thankSHELL- Research(Belgium) for a research grant. Fig. 11. Loss of thiol group upon heating at 1508C of the HV- SBS/Tris system. [Tris] = 5 wt.-%; [Lucirin TPO] = 1 wt.-% (0) or 0% (h) Fig. 12. Insolubilization of HV-SBS upon heating at 1508C. [Tris] = 5 wt.-%; [Lucirin TPO] = 1 wt.-% (0) or 0% (h); neat SBS (f) 1974 C. Decker, T. Nguyen Thi Viet 1) H. F. Huber, inRadiation Curing in Polymer Science and Technology, Vol. 4, J. P. Fouassier, J. F. Rabek, Eds., Elsevier Applied Science, London 1993, p. 51 2) S. M. Ellenstein, S. A. Lee, T. K. Palit,ibit. p. 73 3) M. Dupont, N. De Keyser,Proc. RadTechEurope Conf. Maestricht, 1995, p. 174 4) C. Decker, T. Nguyen Thi Viet,Polym. Mater. Sci. Eng. 74, 327 (1996) 5) C. Decker, T. Nguyen Thi Viet,Macromol. Chem. Phys. 200, 358 (1999) 6) A. F. Jacobine,in Radition Curing in Polymer Science and Technology, Vol. 3, J. P. Fouassier, J. F. Rabek, Eds., Elsevier Applied Science, London 1993, p. 219 7) R. W. Lenz,Organic Chemistry of Synthetic High Polymers, Interscience, New York 1967, p. 196 8) C. R. Morgan, F. Magnolta, A. D. Ketley,J. Polym. Sci., Polym. Chem. Ed. 15, 627 (1977) 9) C. E. Hoyle, R. D. Hensel, M. B. Grubb,Polym. Photochem. 4, 69 (1994) 10) C. Decker, A. Jenkins,Macromolecules18, 1241 (1985) 11) US Patent 2,834,768 (1979), Grace Co., invs.: P. R. Hein, J. A. Evans, M. W. Yang;Chem. Abstr. 91, 115, 383 (1979) 12) US Patent 5,166,226 (1992), Avery Dennision Corp., invs.: C. S. Rice, Y. Sasaki, S. P. Plamthottan 13) R. Schwalm, L. Häussling, W. Reich, E. Beck, P. Enenkel, K. Menzel,Progr. Org. Coat. 32, 191 (1997) 14) C. Walling, W. J. Helmreich,J. Am. Chem. Soc. 81, 1144 (1959) 15) J. I. G. Cardozan, I. H. J. Sadler,J. Chem. Soc., B. 1191 (1996) 16) V. T. D’Souza, R. Nanjundiak, J. Baeza, H. H. Szmant,J. Org. Chem.52, 1720 (1987) Photocrosslinking of functionalized rubbers, 7a Styrene-butadiene block copolymers Christian Decker*, Trieu Nguyen Thi Viet Laboratoire de Photochimie Ge´n´rale (URA N8431-CNRS), Ecole Nationale Supérieure de Chimie de Mulhouse - Université de Haute Alsace, 3, rue Werner - 68200 Mulhouse; France (Received: May 23, 1998; revised: July 13, 1998) SUMMARY: The photocrosslinking of polystyrene-block-polybutadiene-block-polystyrene (SBS) was stu- died by means of infrared spectroscopy, by monitoring the disappearance of the pendent vinyl double bonds, which was shown to proceed within a few seconds upon UV exposure in the presence of an acylphosphine oxide photoinitiator. Complete insolubilization requires the reaction of 17 double bonds per polymer chain for an SBS sample containing 8% vinyl groups. An increase of the vinyl content has little effect on the cross- linking process, because it enhances mainly intramolecular reactions. A paraffinic oil proved to be an effec- tive plasticizer for the photocrosslinked elastomer. The addition of a telechelic acrylate oligomer causes a substantial increase of both the reaction rate and the final degree of conversion of the SBS double bonds. The light-induced copolymerization of the vinyl and butene double bonds with the acrylate double bonds leads to the formation of a hard and flexible polymer material within a fraction of a second. Introduction Light-induced polymerization is one of the most effective methods to generate tridimensional polymer networks, because of the high initiation rates reached under intense illumination1-4). In most UV-curing applications, a sol- vent-free liquid resin is converted quasi-instantly into a highly crosslinked polymer, selectively in the exposed areas, to produce protective coatings, quick-setting adhe- sives or high-resolution relief images. The same photo- chemical process has been successfully used to crosslink solid polymers bearing polymerizable functional groups on their backbone chain5), e.g. cinnamates6), chalcones7), epoxides8, 9) or acrylates10). A distinct advantage of photo- initiation is to afford a precise temporal control of the chemical process. The crosslinking reaction will start immediately, as soon as light is shed on the sample, and it can be stopped at any time by switching off the UV lamp. Moreover, the reaction rate can be varied in a large range, simply by changing the intensity of the UV-beam. In the previous articles of this series, we have shown that epoxy11, 12) or acrylate13) functionalized polyisoprene can be “photovulcanized” within seconds at ambient tem- perature by UV light in the presence of a cationic or radi- cal-type photoinitiator, respectively. Under the same irra- diation conditions, polyisoprene (natural rubber) did not undergo any crosslinking, because of the low reactivity of the amylene double bond. This is not the case for polybu- tadiene which contains the more reactive butene-2 and vinyl double bonds formed by 1-4 and 1-2 polymeriza- tion of butadiene, respectively. There are, however, only a few reports in the literature on the photocrosslinking of polybutadiene-based rubbers14-18). Most of these studies have been performed on styrene-butadiene-styrene (SBS) tri-block copolymers19), known under the tradename KRATON, which are commonly used in pressure-sensi- tive and hot-melt adhesive applications20). These copoly- mers, which have a two-phase morphology, combine the properties of elastomers and thermoplastic materials, the tyrenic domains acting as physical crosslinks below the Tg. Photocuring proved to be an effective method to make these thermoplastic elastomers more resistant to solvent and temperature, while improving greatly the adhesive strength14). By creating covalent bonds within the elastomeric phase, UV-curing is reinforcing irreversi- bly the already existing physical network. The main objective of the present study was to deter- mine how fast and how extensively styrene-butadiene rubbers can be crosslinked by UV irradiation at ambient temperature. We have examined in particular the influ- ence of the photoinitiator and of the polybutadiene vinyl content on the kinetics of the crosslinking process and on some of the properties of the UV-cured polymer network. An attempt has been made to overcome the rate limita- tions inherent to solid state reactions by incorporating in the starting material a plasticizer or an acrylate monomer. The photooxidation process which is known to occur Macromol. Chem. Phys.200, No. 2 i WILEY-VCH Verlag GmbH, D-69451 Weinheim 1999 1022-1352/99/0202-0358$17.50+.50/0 a Part 6: cf. ref.12) 358 Macromol. Chem. Phys.200,358-367 (1999) Photocrosslinking of functionalized rubbers, 7 359 upon UV irradiation of this type of rubber in the presence of air21) was minimized by the rapid consumption of oxy- gen during the short exposure. Experimental part Materials Two types of polystyrene-block-polybutadiene-block-poly- styrene (SBS), both from SHELL, have been used in this study. They differ only by their content of pendent vinyl groups: 8% of the total double bond content for the low- vinyl polymer (LV-SBS), and 59% for the high-vinyl poly- mer (HV-SBS). The butene-2 double bonds located on the polybutadiene backbone represent 92% and 41% of the unsa- turation content, respectively. In some experiments, a paraf- finic oil (Nujol from Aldrich) was added to the formulation to increase the molecular mobility in the polymer film. Such a plasticizing effect was also achieved by using a diacrylate monomer (Ebecryl 150 from UCB) which, in addition, was expected to act as a crosslinking agent by copolymerizing with the polybutadiene double bonds. Different types of photoinitiators were introduced in the polymer, at a typical concentration of 3 wt.-% SBS, namely benzophenone (BZP from Aldrich), 2,6-bis(azidobenzyli- dene)-4-methylcyclohexanone (bisazide from Aldrich), 2,2- dimethoxy-2-phenylacetophenone (Irgacure 651 from Ciba), 1-benzoyl-1-hydroxycyclohexane (Irgacure 184 from Ciba) and 2,4,6-trimethylbenzoyldiphenylphosphine oxide (Lucirin TPO from BASF). 20lm thick films were cast from toluene solution of SBS and the photoinitiator onto either a KBr crystal for infrared analysis or a glass plate for insolubiliza- tion and hardness measurements. Samples were exposed to the radiation of a 80 W/cm medium pressure mercury lamp, in the presence of air, at a passing speed of 50 m/min. The maximum light intensity at the sample position was mea- sured by radiometry (IL-390 light bug) to be 600 mW cm-2. Analysis The kinetics of the light-induced crosslinking of the rubber film was studied quantitatively by FTIR spectroscopy, by following the decrease upon UV exposure of the absorption bands characteristic of the vinyl double bond at 910.8, 1637.8 and 1827 cm-1, and of the butene-2 double bond at 964.7 cm-1. In formulations containing the acrylate mono- mer, the disappearance of this double bond upon UV-expo- sure was followed at 812 cm-1 (CH22CH twisting). The degree of conversion (x) was calculated from the ratio of the corresponding IR absorbance before and after UV exposure (A0 andAt) by using the following equation: x(%) = [1 - (At /A0)] N 100 This value was not corrected for shrinkage, as it was found to account for less than 5%, based on the variation of the C1H peak at 2960 cm-1. The loss of the photoinitiator in the UV-irradiated sample was followed quantitatively by UV spectroscopy. The gel fraction and the degree of swelling of the irradiated polymer were determined by soaking the sam- ple in toluene for one day at room temperature. The insoluble polymer was recovered by filtration and dried at 708C to a constant weight. The hardness of the coating was evaluated before and after irradiation by monitoring the damping of the oscillations of a pendulum (Persoz hardness). The hardness was shown to be strongly dependent on the glass transition temperature22), with Persoz values ranging typically from 30 s for soft elastomeric materials to 300 s for hard and glassy polymers. The physical significance of pendulum hardness has been reviewed by Sato23) who considered it as a reliable measure of the viscoelastic properties of a polymer coating. The adhesion of the UV-cured coating on glass or other substrates was evaluated by using the standard cross- hatch adhesive tape test. The thermal stability of the various formulations was tested by heating the uncured sample for up to 1 h at 1508C, the processing temperature for hot-melt adhesive applications. Results The photoinitiator (PI) is a key factor for achieving a fast and extensive crosslinking by means of UV light. Indeed, both SBS samples were found to remain soluble when they were exposed to UV radiation in the absence of a photoinitiator, for doses up to 5 JN cm-2. It was therefore important to select the most efficient photoinitiator for the given SBS rubber, and to determine its optimum con- centration. The role of the pendent vinyl double bond was also investigated to know whether it is necessary to increase its content to form readily a 3D polymer net- work. Finally, the effect of an acrylate crosslinker on the reaction kinetics was examined, in an attempt to enhance the reactivity of both the vinyl and the butene double bonds through copolymerization. Influence of the radical photoinitiator Most of the radical-type photoinitiators used in UV-cur- able systems consist of aromatic ketones which undergo a carbon-carbon bond cleavage when exposed to UV radia- tion: Various factors govern the initiation efficiency of these compounds: their intrinsic absorbance of the incident light, the lifetime of the excited states, the rate of photo- lysis, the efficiency of the cleavage reaction and the reac- tivity of the radical fragments. Insolubilization profiles were determined for the two SBS samples UV-irradiated in the presence of different types of radical photoinitiators. Fig. 1 shows some char- acteristic curves obtained by plotting the gel fractionver- 360 C. Decker, T. Nguyen Thi Viet susthe exposure time for five photoinitiators in HV-SBS, at a weight concentration of 3%. They can be classed in the following order of increasing efficiency: Bisazidea benzophenonea Irgacure 651 a Irgacure 184a Lucirin TPO The bisazide, commonly used as photoinitiator to crosslink cyclized polyisoprene, appears to be the least efficient of the five initiators tested. The best perfor- mance was reached by using the phosphine oxide initia- tor, nearly all the polymer becoming insoluble after a 1 s exposure. The same classification was obtained when the crosslinking process was followed through the loss of the vinyl double bond or the increase of the rubber hardness. Consequently, this photoinitiator, Lucirin TPO, was selected for all of our further studies. Kinetics of the crosslinking reaction IR analysis shows that only minor chemical modifications are taking place in SBS upon UV exposure, mainly a small decrease of the three peaks of the vinyl double bond at 910.8, 1637.8 and 1827 cm-1. After 1 s, only 4% of the pendent vinyl groups of HV-SBS were found to have reacted, whereas the concentration of the backbone double bonds absorbing at 964.7 cm-1 remained essen- tially unchanged. It can be seen in Fig. 2 that the SBS vinyl content drops rapidly at the very beginning of the UV irradiation and levels off after 3 s to a constant value (95% of the original content). This apparent loss of reac- tivity of the remaining vinyl double bonds can be attribu- ted to the rapid photolysis of the photoinitiator which has essentially disappeared after a 2-second exposure (see below). Mobility restrictions of the reactive sites in the network under formation are also likely to contribute to the premature stop of the polymerization. The photoinitiated polymerization of vinyl groups located on different polybutadiene chains leads to the for- mation of a tridimensional polymer network and accounts for the observed insolubilization of the irradiated rubber (Fig. 3). The polymer network presents, however, a fairly loose structure, as indicated by its high degree of swelling (Psolvent/Ppolymer= 20). A plot of the gel fractionversusthe degree of conversion (Fig. 4) shows that insolubilization is achieved once 5% of the vinyl double bonds have reacted, which corresponds to 62 pendent groups per polybutadiene chain. Theoretically, only three bridges need to be formed per polymer chain to obtain a cross- linked material of infinite molecular weight. The fact that insolubilization of SBS requires the reaction of a much larger number of vinyl double bonds strongly suggests that, in addition to the intermolecular reaction, an intra- molecular polymerization process is taking place between vinyl double bonds located on the same polybutadiene chain. It should lead to the formation of cyclic struc- tures24) along the polybutadiene chain, which would Fig. 1. Influence of the photoinitiator (3 wt.-%) on the insolu- bilization profile of a 20lm thick HV-SBS film exposed to UV radiation (I = 600 mWN cm-2) Fig. 2. Photocrosslinking of a SBS rubber with a high vinyl content (59%) in the presence of Lucirin TPO (3 wt.-%) Fig. 3. Insolubilization profile of HV-SBS exposed to UV radiation in the presence of Lucirin TPO (3 wt.-%) Photocrosslinking of functionalized rubbers, 7 361 therefore remain soluble. This is feasible because of the high vinyl content of the HV-SBS sample (59% of the total number of double bonds). However, because the polybutadiene chain contains alsocis and trans 2-butene units, the cyclisation process will propagate along the polymer chain over only a few adjacent vinyl units. The two inter- and intramolecular polymerization processes are represented formally in Scheme 1. As crosslinking proceeds upon UV exposure, the amount of bounded elastic chains becomes higher and the hardness increases, as shown in Fig. 2. The UV-cured polymer remains still soft (Persoz value of 80 s), which is important for adhesive applications to maintain the tack and wetting required to achieve a good adhesion. The outer polystyrene segments of the block copoly- mer, which are phase separated, are not expected to parti- cipate in the crosslinking process, except by a possible chain transfer reaction if some propagating radicals suc- ceed in abstracting a labile hydrogen on the PS chain at the interphase. Increasing the PS chain length should not affect insolubilization, which is ensured through the poly- butadiene crosslinks, but it will increase the degree of swelling by lowering the crosslink density. It should be noted that a fraction of the vinyl groups is likely to disappear by a photooxidation process21, 25, 26). While no significant changes in the hydroxyl and ether region of the IR spectrum could be detected after the short UV exposure, there was indeed a small but distinct increase of a carbonyl band at 1694 cm-1 during the first 0.5 s of irradiation. Its absorbance was found to level off at a constant value of 0.015 upon further exposure, prob- ably because most of the oxygen dissolved in the film has then been consumed by reaction with the initiator radi- cals. The concentration of the carbonyl group was calcu- lated to be on the order of 2610-2 mol N kg-1, which cor- responds to less than 10% of the total amount of the vinyl groups lost. The photoinitiated polymerization appears thus as the main process responsible for the vinyl double bond consumption during the UV irradiation in the pre- sence of air. Under O2-free conditions, the vinyl groups polymerize only slightly faster than in air (Fig. 2), which shows that O2 inhibition is not very important in these solid films exposed to intense radiation. Influence of the photoinitiator concentration In light-induced reactions, the photoinitiator concentra- tion [PI] controls the rate of initiation (r i) which obeys the following equation: r i ˆ Ui I0 1ÿ eÿel‰PIŠ† whereUi is the initiation quantum yield,I0 the incident light intensity, e the molar extinction coefficient of PI, and l the thickness of the sample. Because of the limited penetration of the incident radiation in the UV-absorbing medium, crosslinking within the SBS sample will follow a surface to depth gradient which is directly dependent on [PI]. Consequently, an increase in the PI concentration will both accelerate the crosslinking reaction and steepen the cure depth gradient in the irradiated sample. Depend- ing on the considered application, the best compromise must be found between cure speed and cure depth, the two extremes being either a uniform but slow deep- through cure for low absorbing samples (low [PI]), or a rapid but differential through-cure for highly absorbing ample (high [PI]). Fig. 5 shows the insolubilization profiles of a 20lm thick HV-SBS film containing 1 to 5 wt.-% Lucirin TPO exposed to UV radiation for up to 3 s. The rate increase w th the PI concentration is particularly important for values up to 2 wt.-%. At higher [PI], the accelerating effect becomes less pronounced, because of the reduced penetration of the incident light in the deep-lying layers (inner filter effect). An increase of the PI concentration leads also to the formation of a tighter polymer network, because of the increased number of initiating radicals. The swelling ratio Fig. 4. Dependence of the gel fraction and hardness of a photo- crosslinked HV-SBS on the degree of vinyl conversion Scheme 1: 362 C. Decker, T. Nguyen Thi Viet of UV-cured HV-SBS was found to drop from 55 to 25 when [TPO] was increased from 1 to 5 wt.-%. Fig. 6 shows the variation of the gel fraction and of the swelling ratio with the photoinitiator concentration, for a 20lm thick sample exposed for 1 s to UV radiation. It should be mentioned that, when the TPO concentration was increased above 6 wt.-%, the overall cure extent started to decrease because of incomplete crosslinking of the bot- tom layer. The [PI] value where maximum efficiency is reached depends directly on the sample thickness, e.g. at a 10 times as low PI concentration for a 10 times as thick film. A similar behavior was observed by following the hardness increase upon UV exposure (Fig. 7), and the vinyl loss as well. The variation of these two parameters with the TPO concentration is shown in Fig. 8. Here again, maximum efficiency was reached for a photoinitia- tor concentration of about 5 wt.-%. For a 20lm thick film, a TPO concentration of 2 wt.-% appears to give the best overall performance, with respect to deep-through cure, insolubilization, network density and moderate hardness increase. It should be emphasized that the depth of cure gradient due to the PI light absorbance is progressively smoothing off upon UV irradiation, because the photoinitiator is con- verted in non-absorbing photoproducts. Fig. 9 shows the exponential decay of Lucirin TPO, monitored by UV spectroscopy at 380 nm, upon UV exposure in a SBS matrix at different concentrations. Such a first-order kinetic law was expected for a direct photolysis process, where the reaction rate depends primarily on the prob- ability of a chromophore molecule to be hit by a photon and is therefore proportional to the chromophore concen- tration, throughout the reaction. From the slope of the straight line obtained in the semi-log plot,ln [PI]t /[PI]0 = -k N t, the decay rate constantk was calculated. Its value Fig. 5. Influence of the photoinitiator concentration (Lucirin TPO) on the insolubilization profiles of HV-SBS upon UV expo- sure Fig. 6. Dependence of the gel fraction and swelling ratio of the photocrosslinked HV-SBS on the photoinitiator concentration. UVexposure: 1 s Fig. 7. Influence of the photoinitiator concentration (Lucirin TPO) on the hardening of HV-SBS upon UVexposure Fig. 8. Dependence of the vinyl loss and the hardness of photo- crosslinked HV-SBS on the photoinitiator concentration. UV exposure: 1 s Photocrosslinking of functionalized rubbers, 7 363 was found to drop from 4.3 s-1 to 2.1 s-1 when [PI]0 was increased from 1 to 5%, thus showing the importance of the radiation inner filter effect27). One of the main consequences of this photobleaching process is to make the top layer of the irradiated sample more transparent to UV light, which can therefore pene- trate deeper into the sample, thus inducing a frontal poly- merization28). Up to a few millimeter thick specimens have been crosslinked within less than 1 min by this UV tech- nology. Influence of the vinyl content of SBS The photocrosslinking experiments reported so far have all been carried out on a SBS sample with a high content (59%) of pendent vinyl double bonds, which are known to be more reactive than the in-chain butene double bonds. As SBS rubbers usually contain a lower amount of pendent vinyl groups, we have repeated some of these experiments with a SBS sample where only 8% of the unsaturations are vinyl double bonds (LV-SBS). Surpris- ingly, the crosslinking of this rubber was found to pro- ceed nearly as fast upon UV exposure as that containing 59% vinyl group. Fig. 10 shows the decrease of the vinyl double bond in a 20lm thick LV-SBS film containing 3 wt.-% Lucirin TPO exposed for up to 3 s to UV-radiation. Although the vinyl double bonds seem to disappear faster in LV-SBS than in HV-SBS (Fig. 10 and 2), this is not the case. This mislead- ing result is due to the chosen representation (relative loss of vinyl group versusexposure time). The initial rate of polymerization of the vinyl group, expressed in molN kg- 1 N s-1, is still 3 times greater in the sample containing a large amount of vinyl double bonds (1 molN kg-1 N s-1 ver- sus0.3 molN kg-1 N s-1). After 0.5 s and a 10% vinyl loss, the reaction is slowing down, most probably because of the fast consumption of the photoinitiator. At the same time, only a slight decrease (2%) of the butene double bond at 967 cm-1 was found to occur upon UV irradiation, very much like in HL-SBS. Crosslinking was character- ized by both an increase of the polymer hardness (Fig. 10) and a nearly complete insolubilization of the irradiated sample (Fig. 11). Although the initial rate of gelation was lower in LV-SBS than in HV-SBS, the crosslink density of the polymer network was similar for both samples, as shown by the values of the swelling ratio: 27 and 25 after a 2 s exposure for LV-SBS and HV-SBS, respectively. All these results, which are summarized in Tab. 1, indi- cate that it is not necessary to increase the vinyl content to achieve a fast photocrosslinking of SBS rubber. In LV- SBS, gelation occurs when about 10% of the vinyl groups have reacted, which corresponds to 17 double bonds per polybutadiene chain. An increase of the vinyl content of SBS will mainly favor the intramolecular polymerization between neighboring pendent double bonds, without much effect on the crosslinking process. Fig. 9. Disappearance of the photoinitiator (Lucirin TPO) upon UV irradiation of HV-SBS Fig. 10. Loss of the polybutadiene double bonds and hardness increase in a UV-irradiated SBS rubber with a low vinyl content (8%). Photoinitiator: [Lucirin TPO] = 3 wt.-% Fig. 11. Insolubilization profile of a LV-SBS rubber exposed to UV-radiation in the presence of Lucirin TPO (3 wt.-%) 364 C. Decker, T. Nguyen Thi Viet Influence of plasticizers Like any reaction carried out in the solid state, the photo- crosslinking of SBS is hampered by the restricted mobi- lity of the reactive sites (initiator and polymer radicals, as well as vinyl double bonds). The addition of a plasticizer like Nujol, a parrafinic oil, was thus expected to acceler- ate the polymerization process. Such an effect was actu- ally observed only for Nujol concentrations above 20%, probably because of the elastomeric character of the SBS matrix which provides already enough molecular mobi- lity in the neat polymer. For a 50/50 mixture by weight of LV-SBS and Nujol, the vinyl double bonds were found to disappear almost twice faster than in the neat SBS sample (Fig. 12). Insolubilization was accelerated in the same manner (Fig. 13), but the photocrosslinked polymer remained very soft because of the presence of the plastici- zer occluded in the polymer network (Fig. 12). It should be mentioned that the gel fraction of the UV- cured polymer, which reaches close to 100% in 2 s, was calculated on the basis of the SBS content, i.e. 50% of the irradiated sample had become insoluble in toluene. Interestingly, the degree of swelling of the polymer net- work was cut by half when LV-SBS was UV-cured in the presence of Nujol (Fig. 13), thus showing that the cross- link density has increased. This result is in good agree- ment with the observed doubling of the vinyl polymeriza- tion rate and suggests that the plasticizer is favoring inter- molecular polymerization by increasing the molecular mobility of the SBS chains. Similar results were obtained when Nujol was added to the SBS sample containing 59% pendent vinyl groups. The effect of the plasticizer concentration on the vinyl conversion, insolubilization and hardness is illustrated in Fig. 14 for a HV-SBS sample exposed to UV radiation during 1 s. For both SBS rubbers, a Nujol concentration of 30 wt.-% appears to be an optimum value to achieve an efficient crosslinking and obtain a polymer showing a strong elastomeric character and excellent adhesion on various substrates (glass, metals, plastics). It should be noticed that the plasticized SBS films remain perfectly clear, up to a Nujol concentration of 70%, thus showing the good compatibility of these two compounds. Tab. 1. Performance analysis of SBS photocrosslinking Thermoplastic elastomer LV-SBS HV-SBS Initial vinyl content in % 8 59 Rate of vinyl loss % sÿ1 mol N kgÿ1 N sÿ1  25 10 0.3 1.0 Rate of gelation (% s-1) 150 240 Rate of hardness increase 60 80 Swelling ratio after 2 s 27 25 Fig. 12. Photocrosslinking of a blend of LV-SBS rubber and a paraffinic oil ([Nujol] = 50 wt.-%) Fig. 13. Insolubilization profile of a 50/50 LV-SBS/Nujol blend exposed to UV irradiation in the presence of Lucirin TPO (3 wt.-%). - - - -: neat LV-SBS Fig. 14. Influence of the plasticizer content (Nujol) on the characteristics of a 1 s UV-irradiated HV-SBS rubber Photocrosslinking of functionalized rubbers, 7 365 Influence of an acrylate crosslinker In an attempt to further increase the crosslinking effi- ciency, an acrylate monomer, was introduced in the SBS rubber, together with the photoinitiator (Lucirin TPO). At 20% by weight of a diacrylate bisphenol A telechelic oli- gomer (Ebecryl 150), both the pendent vinyl and the backbone butene double bond were found to react rapidly upon UV exposure, as shown by Fig. 15. A 0.1 s exposure proved to be sufficient to induce the polymerization of 14% of the vinyl groups and 6% of the butene-2 double bonds, compared to 1% and 0% in neat HV-SBS, respec- tively. At the same time, 70% of the acrylate groups had polymerized, as shown by the sharp drop of the character- istic absorption peak at 812 cm-1. This result strongly suggests that an effective copolymerization is taking place between the acrylate double bonds and the vinyl double bonds. A strong correlation was found to exist between the rate of polymerization of the polybutadiene unsaturation and that of the acrylate double bonds, during the whole reaction time. The same behavior was observed with the LV-SBS sample. Fig. 16 shows schematically the structure of thenet-polybutadiene-i diacrylate formed upon UV irradiation of the SBS-diacrylate mixture. As expected from the kinetic profiles of Fig. 15, insolubiliza- tion of the SBS rubber is rapidly taking place with forma- tion of a relatively tight polymer network (Fig. 17). After a 1 s UV exposure, the swelling ratio of HV-SBS was measured to be 12, compared to 32 for the neat rubber. The additional crosslinks generated by the diacrylate monomer are causing a substantial increase of the poly- mer hardness (Persoz values of 200 s). Because of the elastomeric character of the polybutadiene chains, the UV-cured film still maintains a high flexibility and passes successfully the severe zero-T-bend test. These highly reactive UV-curable systems are therefore more suited for protective coating than for adhesive applications, spe- cially on flexible substrates. Thermal stability of the UV-curable formulation When these UV-curable rubbers are used as hot-melt adhesives or as flexographic printing plates, the formula- tion is processed at temperatures up to 1508C19, 20). It was therefore important to evaluate the thermal stability of the SBS rubber containing the photoinitiator and the addi- tive (Nujol or diacrylate). The photoinitiator selected (Lucirin TPO) was found to be relatively stable at that temperature, its concentration in the sample dropping by only 20% after a 30 min heating at 1508C in the presence of air (Fig. 18). The UV-curing performances were there- fore not very much affected by this thermal treatment. Fig. 15. Polymerization of the acrylate and polybutadiene dou- ble bonds, upon UV exposure of a 20/80 blend of diacrylate monomer (Ebecryl 150) and HV-SBS rubber. [Lucirin TPO] = 3 wt.-% Fig. 16. Structure of polymer network formed by photopoly- merization of acrylate and polybutadiene double bonds Fig. 17. Insolubilization profile of HV-SBS/Ebecryl 150 blend (80/20 by weight) exposed to UV radiation in the presence of Lucirin TPO (3 wt.-%). - - - -: neat HV-SBS 366 C. Decker, T. Nguyen Thi Viet The concentration of vinyl double bonds remained essen- tially unchanged during the first 20 min of heating, and started to decrease upon further treatment, as shown in Fig. 18 for the LV-SBS sample. The butene double bonds proved to be somewhat more resistant to this thermal treatment, as expected from their lower reactivity. The loss of double bonds is accompanied by polymerization and oxidation reactions which both occur after a certain induction period. The SBS rubber started to become partly insoluble after a 20 min heating (Fig. 19), the gel fraction reaching 50% after 40 min, with formation of a relatively tight polymer network (swelling ratio of 10). Different types of functional groups (alcohol, ketone, ether), resulting from thermooxidation were detected by infrared spectroscopy for heating periods exceeding 20 min, as shown in Fig. 20. Similar results have been obtained with the SBS sample containing 59% vinyl dou- ble bonds (HV-SBS) which proved to be nearly as resis- tant to heat as LV-SBS. If necessary, the thermal stability of the UV-curable rubber can be increased by the addition of a phenolic antioxidant, like Irganox 1010 from Ciba15). Interestingly, the addition of paraffinic oil (Nujol) was found to improve the heat resistance of the SBS rubber (Fig. 19). This is not the case of the very reactive diacryl- ate monomer (Ebecryl 150) which promotes the thermal curing, as shown in Fig. 21. After a 20 min heating at 1508C in air, 30% of both the butene and vinyl double bonds had polymerized, and as many as 70% of the acryl- ate double bonds, with formation of a totally insoluble polymer. The decreased stability of the butene double bond in the presence of an acrylate monomer is an argu- ment in favor of its copolymerization. The heat resistance of SBS rubber containing a phos- phine oxide initiator is strong enough to allow this UV- curable system to be used in hot-melt adhesive applica- tions. This is not true for the much more reactive acryl- Fig. 18. Heat resistance at 1508C of a LV-SBS rubber contain- ing 3 wt.-% Lucirin TPO Fig. 19. Insolubilization of LV-SBS upon heating at 1508C in the presence of Lucirin TPO (3 wt.-%) Fig. 20. Oxidation products formed upon heating at 1508C of a LV-SBS rubber containing 3 wt.-% Lucirin TPO Fig. 21. Loss of acrylate and polybutadiene double bonds upon heating at 1508C of a LV-SBS rubber containing 3 wt.-% Lucirin TPO Photocrosslinking of functionalized rubbers, 7 367 ate-loaded SBS formulation which is best suited for the production of fast-drying protective coatings showing excellent adhesion and impact resistance. Conclusion Polystyrene-block-polybutadiene-block-polystyrene can be readily crosslinked at ambient temperature by UV irra- diation in the presence of a phosphine oxide photoinitia- tor. The radical-induced polymerization of the pendent vinyl groups occurs within seconds to yield an insoluble elastomeric material. The reaction kinetics can be finely controlled through the photoinitiator concentration which governs also the cure depth profile. Increasing the vinyl double bond content of the polybutadiene chain from 8 to 59% has little effect on both the cure speed and the cross- link density of the polymer network, as it mainly favors intramolecular reactions between neighboring pendent double bonds. The crosslinking reaction can be markedly accelerated by the addition of a telechelic acrylate oligo- mer which promotes an effective copolymerization with the polybutadiene double bonds, but at the expense of the formulation heat resistance. The photocrosslinking technology offers a number of advantages29), such as cure speed, low energy consump- tion, ambient temperature operations, no emission of sol- vent and spatio-temporal control of the initiation step. Because of its process facility and efficiency, it has been successfully applied to crosslink a commercial-type ther- moplastic elastomer and modify, selectively in the illumi- nated areas, its physico-chemical characteristics. Such UV-curable rubbers are particularly well suited to produ- cing fast-drying coatings, hot-melt adhesives and flexible printing plates. Acknowledgement: The authors wish to thankSHELL RESEARCH(Louvain la Neuve - Belgium) for a research grant. 1) C. E. Hoyle, J. F. Kinstle, Eds.,Radiation Curing of Poly- meric Materials, ACS Symp. Ser. Vol 417 American Chemi- cal Society, Washington JC (1990) 2) P. K. T. Oldring, Ed.,Chemistry and Technology of UV and EB Formulation for Coatings, Inks and Paints, Vol. 1-5, Wiley/SITA, London 1991 3) C. Decker,Progr. Polym. Sci. 21, 593 (1996) 4) A. B. Scranton, C. N. Bowman, R. W. Peiffer, Eds., Photopo- lymerisation. Fundamentals and Applications, ACS Symp. Ser. Vol. 673, American Chemical Society, Washington DC 1997 5) G. E. Green, B. P. Stark, S. A. Zahir,J. Macromol. Sci., Rev. Macromol. Chem.C21, 187 (1982) 6) J. Paczkowski,Polymeric Materials Encyclopedia, Vol. 7. J. C. Salomone, Ed., CRC Press, New York 1996, p. 5142 7) A. Reiser, P. L. Egerton,Photogr. Sci. Eng. 23, 144 (1979) 8) J. E. Puskas, G. Kaszas, J. P. Kennedy,J. Macromol. Sci. Chem. A28, 65 (1991) 9) J. V. Crivello, B. Yang,J. Macromol. Sci. Chem. A31, 517 (1994) 10) H. Le Xuan, C. Decker,J. Polym. Sci., Polym. Chem. Ed. 31, 769 (1993) 11) C. Decker, H. Le Xuan, T. Nguyen Thi Viet,J. Polym. Sci., Polym. Chem. Ed. 33, 2759 (1995); C. Decker, H. Le Xuan, T. Nguyen Thi Viet,J. Polym. Sci., Polym. Chem. Ed. 34, 1771 (1996) 12) C. Decker, T. Nguyen Thi Viet, H. Le Xuan,Eur. Polym. J. 32, 1319 (1996) 13) C. Decker, T. Nguyen Thi Viet, H. Le Xuan,Europ. Polym. J. 32, 549 (1996); C. Decker, T. Nguyen Thi Viet, H. Le Xuan, Europ. Polym. J. 559 (1996) 14) M. Dupont, N. De Keyser,Proc. Münchener Klebstoff Con- ference, Munich 1994, p. 120; M. Dupont, N. De Keyser, Proc. RadTech Europe Conf., Maastricht 1995, p. 174 15) Europ. Pat. EP 696761 (1996), C. Mayenez, X. Muylder- mans;Chem. Abstr.124(20), 274561w 16) K. Nityl, Proc. RadTech Europe Conf.,Maastricht, 1995, p. 185 17) C. Decker, T. Nguyen Thi Viet,Polym. Mater. Sci. Eng. 74, 327 (1996) 18) J. Hrivikova, A. Blazkova, L. Lapcik,Chem. Pap. 43, 191 (1989) 19) S. M. Ellenstein, S. A. Lee, T. K. Palit, in:Radiation Curing in Polymer Science and Technology, Vol. 4, J. P. Fouassier and J. F. Rabek, Eds., Elsevier Applied Science, London 1993, p. 73 20) H. F. Huber, inRadiation Curing in Polymer Science and Technology, Vol. 4, J. P. Fouassier and J. F. Rabek, Eds., Else- vier Applied Science, London 1993, p. 51 21) M. A. De Paoli,Eur. Polym. J. 19, 761 (1983) 22) R. Schwalm, L. Haussling, W. Reich, E. Beck, P. Enenkel, K. Menzel,Progr. Org. Coat. 32, 191 (1997) 23) K. Sato,J. Coat. Technol. 56(708), 47 (1984) 24) M. A. Golub,Pure Appl. Chem.30, 105 (1972) 25) A. Ghaffar, A. Scott, G. Scott,Eur. Polym. J. 17, 941 (1981) 26) H. Xingzon, L. Zubo,Polym. Degr. Stab.48, 99 (1995) 27) C. Decker,J. Polym. Sci., Polym. Chem. Ed. 30, 913 (1992) 28) C. Decker,Polymeric Materials Encyclopedia, Vol. 7, J. C. Salomone, Ed., CRC Press, New York 1996, p. 5181 29) C. Decker,Polym. Intern. 45, 133 (1998) 2452 S. 0. JONES AND moved by the reaction of the organic halide with the adjacent atoms of magnesium. If the film is magnesium hydroxide its disappearance on stand- ing might be explained by the reaction with the magnesium bromide from the Grignard equilib- rium to produce basic magnesium bromide which is soluble in ether. Summary (1) Magnesium amalgam electrodes gave zero dark voltage and no light sensitivity with ether solutions of either phenylmagnesium bromide or E. EMMET REID Vol. 60 ethylmagnesium bromide. (2) Magnesium elec- trodes which had been cleaned by reacting with the organic halide gave the same results. (3) Magnesium electrodes which had not been cleaned in this manner showed variable dark voltages and variable photo-voltaic responses. Both of these could be destroyed by adding the organic halide or by allowing the cell to stand for several clays. (4) "Cleaned" magnesium electrodes were made sensitive to light by exposure to oxygen. C O I . U N B I A , MU "?CEIVt:D JUNE 23, 19.78 ._ . . . .- . . . [CONTRIBUTION FROM THE CHEMICAL LABORATORY OF TAB JOHNS HOPKINS UNIVERSITY] The Addition of Sulfur, Hydrogen Sulfide and Mercaptans to Unsaturated Hydrocarbons BY S. 0. JONES' AND E. EMMET REID 'There have been many investigations of the action of sulfur and hydrogen sulfide on unsatu- rated hydrocarbons* and a few on the addition of mercaptans3 but so many points have not been cleared up that further study seemed desirable. Curiously enough the end-products are much the same whether an unsaturated hydrocarbon is treated with sulfur, hydrogen sulfide or a mer- captan. Thus with ethylene, sulfur gives hy- drogen sulfide which reacts with more ethylene to form ethyl mercaptan which then adds to more ethylene to form ethyl sulfide. Our results help to explain the presence of ethyl, i-propyl and other mercaptans and the corresponding sulfides in petroleum distillates but do not account for the presence of methvl mer- captan and methyl sulfide. The Action of Sulfur.-Quite different results are obtained from the reaction with sulfur ac- cording to the conditions and whether free sulfur or a compound which readily liberates sulfur is used. When ethylene was passed over pyrites at 350' about 1% of thiophene was isolated along with hydrogen s a d e and ethyl mercaptan, while when it was bubbled through sulfur at 325' much hydrogen sulfide was formed dong with 3% of ethyl mercaptan and small amounts of carbon di- sulfide and ethyl suEde. We found it advan- (1) Taken from the Ph. D. dissertation of S. 0. Jones, R. J . Rey- nolds Tobacco Co. Fellow, The Johns Hoplrins University, June, 1936. (2) Mdlhe and Renaudie, Compt. rend., 196, 891 (1932); Duffey, Snow and Keyes, Ind. Eng. Chem., 16,91 (1934). (3) Posner, Bcr , 86, 646 (1905)' Nicolet. THIS J O I T R N A L , 67, 1008 (1935) tageous to use ethyl tetrasulfide as a sulfur donor. This is a liquid which mixes well with organic com- pounds and decomposes on heating, giving off what may be assumed to be atomic sulfur. The proportions were calculated on the basis of the tetrasulfide going down to the disulfide. A note- worthy difference is that with it, appreciable yields of the cyclic sulfides were obtained. When ethylene was bubbled slowly through ethyl tetrasulfide kept a t about 150°, the main product isolated was ethyl mercaptan with some ethyl sulfide and in addition some ethylene sulfide. The results of heating several hydrocarbons in a bomb with ethyl tetrasulfide are given in Table I. TABLE J Hydrocarbons with Et&, 10 hrs. at 180' -- Yields of rducts , IIydrocarboit Mercaptan Surfide 3- Ethylene 5 1s 1 Propylene ti 20 15 Heptene- 1 80 Octene- I 19 Cyclohexeue b 8 The mercaptans, except from ethylene, and sul- fides were all secondary. The Addition of Hydrogen Sulfide.-The addition of hydrogen sulfide was effected by heat- ing in the bomb for ten hours at 180'. Sulfur was added as a catalyst for without it there was little if any addition. The results are given in Table 11. The sulfur is found on the secondary or tertiary carbon atom in accordance with Mar- kownikow's rule. Oct., 1938 ADDITION OF SULFUR AND SULFIDES TO UNSATURATED HYDROCARBONS 2453 TABLE I1 Addition of hydrogen sulfide, 10 hrs. at 180' --Yields of products, %- Hydrocarbon Mercaptan Sulfide Ethylene 11 80 Propylene 7 90 &Butylene 23 6 Octene-1 9 35 Cyclohexene 7 5 The addition takes place according to Mar- kownikow's rule. In all cases the mercaptan that is first formed adds to a second molecule of the hydrocarbon to form the sulfide. In the case of the simple unsaturates this second reaction must be rapid as compared to the first addition as little of the mercaptans is left over. &Butyl and cyclo- hexyl mercaptans evidently do not add so readily to the hydrocarbons. The Addition of Mercaptans.-The unex- pected observation was made in the course of our experiments that peroxides influence the mode of addition of mercaptans to unsaturates just as Kharasch4 found that they influence the addition of hydrogen bromide. When we heated ethyl mercaptan with propylene we obtained ethyl i- propyl sulfide, but with octylene the product was ethyl n-octyl sulfide. The propylene had never been exposed to the air while the octylene had been stored for some time in a partly filled bottle and gave a strong test for peroxides. By adding ethyl mercaptan to propylene in the presence of added peroxides, we obtained ethyl n-propyl sul- fide. It has been observed by Posner and Nicolet3 that thiophenol and p-thiocresol add to unsatu- rates contrary to Markownikow's rule. We have verified this in a large number of cases, but by using freshly distilled thiophenol and p-thiocre- sol we were able to reverse the mode of addition. Kharasch4 observed that p-thiocresol was not effective as an anti-oxidant unless it was freshly distilled. It appears that only very small amounts of peroxides such as are usually present in the hydrocarbon or in the p-thiocresol itself are suf- ficient to influence the addition of mercaptans while for hydrogen bromide much larger amounts are required. In the absence of catalysts scarcely any addition takes place even on heating for twenty-four hours at 180'. Sulfur catalyzes the normal addition and peroxides the abnormal. It is difficult to suppress the abnormal addition entirely. Into each of 3 tubes of 2-cc. capacity (4) Kharasch and Hannum. THIS JOURNAL, 66,712 (1934). were placed 0.25 g. of p-thiocresol and 0.4 cc. of tridecene. The p-thiocresol had been distilled recently in vacuo and the tridecene had just been distilled over sodium. To the first tube was added a trace of ascaridole, to the second a trace of sulfur and nothing to the third. After the usual heating the product in the first tube was solid and on recrystallization gave 0.3 g. which melted at 39.9', the melting point of p-thiocresyl n-tridecyl sulfide made from n-tridecyl bromide in the usual way. From the liquid products in the other two tubes it was possible to isolate 0.01 and 0.08 g., respectively, of the sulfide melt- ing at the same point. In a number of experiments the products were identified by their boiling points and by oxidation to known solid sulfones but for positive proof of structure unsaturated hydrocarbons6 and mer- captans were chosen that would give known crys- talline sulfides which previously had been pre- pared from the potassium salt of the mercaptans and the normal alkyl bromides.6 The tubes were of 2-cc. capacity and the hydrocarbon was in slight excess. The heating was for ten hours and 180'. The results given in Table I11 show con- clusively that the abnormal addition took place with these hydrocarbons which had been exposed to air for some weeks. In another series tridecylene was added to dimercaptan~.~ The results are given .in Table IV. In one experiment lauryl mercaptan was added to allyl lauryl sulfide CI2Hz6SCH2CH=CH2 and the product was C B H ~ ~ ( C H ~ ) ~ S C ~ Z H ~ ~ , m. p. 47O, proved by melting point and mixed melting point to be identical with that from lauryl bro- mide and trimethylene mercaptan. Experimental Ten moles of ethylene bubbled at the rate of 50 cc. per minute through a 25-cm. layer of sulfur in a vertical tube kept at 325' gave much hydrogen sulfide. The 25 cc. of condensate gave 15 g. of ethyl mercaptan, b. p. 34-37', Hg(SEt)s, m. p. 76', 2 g. of carbon disulfide and 3 g. of ethyl sulfide, b. p. 91-92', Et&HgCle, m. p. 76.5". Five moles passed at the rate of 20 cc. per minute through a 20- cm. column of ethyl tetrasulfide at 140-150" gave 30 g. of condensate, largely ethyl sulfide but containing 10 g. of ethyl mercaptan Hg(SEt)*, m. p. 76', and 1.5 g. of ethylene b. p. 54-57', S calcd. 53.36, found 53.09 and 53.12. (5) Prepared by Kozacik and Reid, ibid. , 60, 2436 (1938). (6) The alkyl bromides were those described by Meyer and Reid, (7) Hall and Reid, unpublished results. ( 8 ) DelCpine. Bull. SOC. chim., 788, 03 (1923). ibid. , 66, 1574 (1933). Vol. 60 2454 s. 0. JONES AND E. EMMET REID TABLE 111 MELTING POINTS AND ANALYSES OF SULFIDES OBTAINED BY ADDITION COMPARED WTE THOSE OF SYNTHETIC SULFIDES Un-aturata CHa( CHz)aCH=CHz CH3( CHi)sCH=CHz CHs( CHZ)ECH=CHZ CH3( CHz)aCH=CH? CHa( CHz)ioCH=CHn CHa( CHz)ioCH=CHz CHs(CHz)ioCH=CHz CH3( CHz)ioCH=CHs CH3( CHz)izCH=CHz CH~(CH~)I~CH=CHI CHa(CH1)13CH=CHz CHB( CHi)izCH=C€Iz CH,(CHz)i4CH=CHi CHa( CHn)irCH=CHs CHa(CHz)i4CH=CH? CH3(CHz)irCH=CHz C€Is( CHi)isCH=CHs CH3(CHz)isCH=CHs CH3( CHz)lsCH=CHx CH3(CHz)isCH=CHz BY synthetic addition Sulfide M . p , ' C . hl. p., OC 33.8 20.8 46.8 37.2 43 .8 40.2 54.6 39.2