Thermal decomposition of diisopropyl peroxydicarbonate. CIDNP and product studies

European Polymer Journal Vol. 16. pp. 1127 to 1134

0OI4-3057/80:1201-1127502.00:0

¢) Pergamon Press Ltd 1980. Printed in Great Britain

THERMAL DECOMPOSITION OF DIISOPROPYL PEROXYDICARBONATE CIDNP AND PRODUCT STUDIES E. F. J. DUVNSTEE*,M. L. ESS~+ and R, SCHELLEKEN$% DSM. Central Laboratory, Geleen, Netherlands (Receired 7 May 1980)

Al~lr=ct--The thermal decomposition of diisopropyl peroxydicarbonate (IPPI in two solvents (chlorobenzene and tetrachlorcethylene) was studied by [~H]NMR. In both solvents the major decomposition products, acetone and isopropanol, showed interesting CIDNP effects during the first few minutes of the decomposition at 60 and 75 respectively. For decomposition in the presence of excess di-t-butyl-pcresol, no acetone was formed while the other CIDNP effects were completely suppressed. From these results and further analysis of the by-products, it is concluded that the observed CIDNP effects are closely connected with induced decomposition of 1PP, starting by abstraction of the secondary H-atom of the isopropyl group.

INTRODUCTION

Peroxydicarbonates (R--O--CO=)2 are radical initiators used in many commercial processes such as the large scale production of polymers and the curing of resins [1]. Compared with other classes of peroxides, the number of mechanistic studies on the decomposition of these compounds is relatively small [2]. One of the more recent studies was made by Van Sickle[3] who investigated the thermal decomposition of dicyciohexyl peroxydicarbonate. By measuring the yield of the main products (CO=, cyc]ohexanol and cyclohexanone) in three solvents [benzene (inert), toluene (H-donating) and =.methylstyrene (reactive double bond)] he reached the following mechanistic conclusions. (a) Decarboxylation of the initially formed ROCO~ radicals does not occur and the formation of alcohol takes place via the half-ester of carbonic acid

A recent kinetic study of Emanuel et al. [4] also indicated that in inert solvents, such as benzene. induced decomposition of dicyclohexyl peroxydicar-

bonate takes place. With this background information, it seemed interesting to study peroxydicarbonate decomposition by exploring Chemically Induced Dynamic Nuclear Polarization (CIDNP) effects[5]. The CIDNP phenomenon, i.e. the occurrence of intense emission and enhanced absorption lines in NMR spectra during chemical reactions, has been studied for man>, other peroxides such as diacy]pcroxides and peroxyesters. providing useful and detailed information on decomposition mechanisms [6]. To avoid the danger of overemphasizing the importance of the products which show a polarization effect, we have also c',rried out rather detailed product analyses. EXPERIMENTAL Diisopropyl peroxydicarbonate (IPP) (98",. pure) v,as

R--O--C---O. --~ R O C - - O H --. R---OH + CO=. O

obtained from AKZO Chemic and used without further purification. Tetrachloroethylene (Merck, Uvasol). chloro-

O

{b) Formation of ketone occurs via induced decomposition of the peroxide:

i

I

I

I

I

benzene (Baker Analysed). 2-butanol (Merck. p.a.)and 2.6 di-l-butyI-p-cresol lShell Chemical Co.) were used as such.

J

H- - - C - - O - - - C - - O - - - O - - < ' - - O - - ~ - - H Rqco~, . C - - O - - < 7 - - O - - ( O - C - - O - - C - - H

1

O

:

O

rl O

I

J

-. C-----O

f

O

i + CO2

+ -O---C--O---C--H O

* To w h o m correspondence should be addressed. + Trainees Zuid Limburgse LaboratoriumschooL Sinard. Netherlands. g.P..,.16 12--A

1127

The N M R measurements were carried out on a Varian A 60 or a Varian EM-360 L spectrometer. All chemical shift~ were measured with respect to T M S . C I D N P effectswere

1128

E.F.J. Dt:'r.~STEE.M. L. ESSER and R. SCHELLEKENS

located by scanning only a small part I~< 2 ppm) of the spectrum, immediately after placing the sample tube in the preheated probe of the spectrometer. A fresh sample tube was used for every 2 ppm part of the spectrum. Duration of each scan was abou.t IO--30sec and this scan was repeated I0-15 times till no further CIDNP effect w a s seen.

Concentrations in chlorobenzene were determined by comparing integrals of suitable NMR signals with the signal integral of hexamethyldisiloxane (6 = 0.11) (b.p, 99.4 ) of which a known concentration was added to the [PP solution prior to the decomposition. Chemical shifts in chlorobenzene were as follows: IPP (CH3 doublet I.I I. CH septet 4.81). isopropanol (CH3 doublet I.I1. CH septet 3.89). acetone (CH3 singlet 1.84). Product concentrations in tetrachloroethylene were determined by comparing integrals of suitable product signals with the integral of the doublet and septet of IPP as measured before the start of the decomposition. Chemical shifts in tetrachloroethylene were as follows: IPP (CH3 doublet 1.30. CH septet 4.95). isopropanol (CH3 doublet 1.10. CH septet 3.9). acetone (CH3 singlet 2.06). ( C H 3 ) , C H - O - - C O : X (CH3 doublet 1.32. CH septet 4.951. (CH3).,CIOHIX (CHj singlet t.55). HCI was determined by titration with IC,H,~),N'OHin p',ridine. Pure tetrachloroethylene did not consume any titrant. GC MS analyses were carried out using a Finnigan GC MS DATA system. Temperature of the GC injection part was 200-220 and He was used as carrier gas. For analysis of the products in tetrachloroethylen¢, a 1.5 m packed column with OV-I as stationary phase was used and the temperature was raised during the GC run from 70 up to 240 =. For the analysis of the products in chlorobenzen¢, a 2 m packed column with Carbowax 20 M as stationary phase was used and the temp. was programmed from 70 up to 180. For the mass-spectral identification of the GC

peaks, use was made of both Electron Impact (El) and

Chemical Ionization (CI) (isobutane).

RESULTS

C I D N P effects durinq decomposition of I P P From the peroxydicarbonates which are frequently used practically we chose IPP because this peroxide and also its decomposition products have simple [ I H ] N M R spectra and because it is the only peroxydicarbonate for which C I D N P effects have already been observed E7]. As solventswe selected chlorobenzene because it was expected to be rather inert and tetrachloroethylene because it is a protonless, sufficiently high boiling solvent. After solutions of IPP were placed inside the preheated probe of a 60 M H z ['' H ] N M R spectrometer, a strong emission (E) signal from the methylprotons of acetone and a much weaker absorption-emission (AE) signal from the secondary proton of isopropanol were observed. Only the solution of IPP in tetrachloroethylene showed three additional E signals at lower field, near 6 6 ppm. All of these C I D N P effects lasted no longer than about 6 minutes after the start of the reaction. Some illustrative examples of the observed C I D N P effects are shown in Fig. I while chemical shifts and other relevant data are summarized in Table I. All C I D N P effects disappeared when the decomposition of 1PP was carried out in the presence of di-tbutyl-p-cresol IDBPC). a very efficient scavenger for the initially formed R - - O - - C O ' . , radicals. Addition of 0.07-0.1 M of D B P C was sufficient to

b

2.5

2.0

1.5 ppm

4.5

c

4.0

3.5 ppm

6.5

G.O ppm

Fig. 1. CIDNP effects during thermal decomposition of IPP. a--0.70 M IPP in C6HsCI at 60~, 230 see after start: b--- 0.70 M IPP in CoHsCI at 60 ~, 240 sec after start, c-=0.82 M IPP in C2CI, at 75-', 140 sec after start.

Thermal decomposition of IPP

1129

Table I. CIDNP effects during thermal decomposition of IPP (chemical shifts in ppm with respect to TMS) 0.70 M IPP. 60 solvent C~,HsC1

0.82 M IPP. 75 solvent C2C1,

Strong E effect in singlet at ,6 = 1.84

Very strong E effect in singlet at • = 2.06

Decomposition product CH3 \ C-----O CH,~

A E effect in septet at `6 = 3.89

Weak AE effect in septet at `6 = 3.9

CH

OH

3

\

/

/\ CH3 Very strong E effect in singlet at t~ = 6. I Weak E effect in singlet at ,6 = 6.2 Moderate E effect in singlet at `6 = 6.35 suppress completely the emissions during the dec o m p o s i t i o n O n addition of smaller a m o u n t s of DBPC. the emissions are suppressed only during the first period of the decomposition: after that period weaker emissions can still be observed.

H

C] CI [ [ ---C---C--H I J CI CI

Product uriah'sis In order to have a sound basis for a discussion of the observed C I D N P effects, we carried out rather complete product analyses. The decompositions of

Tabte 2. Products formed after complete decomposition of IPP in solvents chlorobcnzenc and tetrachloroethylene 0.70 M lPP in CbHsCI 1 hr at 60

0.82 M 1PP in C~CI,= I hr at 75

Product (tool/tool of peroxide decomposed) CO: CH3

\/ /\

Not determined

1.1

0.5

0.9

0.7

~0.1

0.3

~0.1

0.1

0

0.8

H

C

CH3

Not determined

OH

CH~

\ /

C~-O

CHz '\. / / C / \\

CH3

H'CH3

O--C---O--X 0

CH~

OH

C / \ CH~ X HCI

* Coupling product with solvent (X) (seeTable 3). t- Determined by titration with (C,=H,d4N " O H - in pyridine~.

E.F.J. DV~NSTI-K. M. L EssI!R and R. S('HliLLliKliNS

1130

Table 3. Identification of byproducts after complete decomposition of 1PP in chlorobenzene or in tetrachloroethylene

0.70M IPP in C,H~CI I hr.at 60

0.82 M IPP in C2C1,, I hr at 75 H

CH3

\/

CH3

{M =

~

180}

/

C

\

CH 3

O'---C--O--X O

CH 3 \ /OH

X = --CCI = CCI2

(M = 232)

X = ---...CCI,~CCI2H

(M = 268)

X = --CCI:~CCI 3

(M = 302)

X = --CC1 = CCI---CCI = CCI2

(M = 326)

(M = 170)

CCIs~CCI: H

(M = 200)

(2 isomers)

CCI3---CCI 3

(M = 234)

CCI., = CCI--CCI = CCI.,

(M = 258)

CHj

X = --CCI = CC12

(M = 188)

X - --CCI = CCI---CCI = CCI~

(M = 282)

CHj CH~

I

HO

1

C

OH

C

(M = 1181

\/

CHj CH~

/\

CHj

OH

C

CH~

X

X =

--CCI = CCI---O-=C---O---CH(CHj)2

(M = 256)

H--C--O

I \c=o

O (M = 102)

/

H:C--O

I P P ~ere carried out in the same N M R tubes and under the same conditions as used for the observation of C I D N P : after the decomposition of the peroxide ~ a s complete, the product concentrations were determined from the integrals of the N M R lines. The results are summarized in Table 2. Table 2 shows that chlorobenzene is indeed a rather inert solvent. The sum of the yields of isopropanol and acetone is practically equal to the theoretical a m o u n t 12 mol tool of peroxide) and the a m o u n t of coupling products with solvent is only very small.

In tetrachloroethylene however the sum of the isopropanol and acetone yields is only 60°,,: substantial a m o u n t s of coupling products with solvent and also of HCI are formed. In order to identify these coupling products and to investigate which other products might have been formed, we subjected the decomposed I P P solutions to GC,'MS analysis. For complete identification of every G C peak, it was necessary to use both Electron Impact {El) and Chemical Ionization (CI) techniques. The results are summarized in Table 3 :

Table 4. Products formed after complete decomposition of IPP in the presence of excess DBPC

Product CH~

\/

0.70 M IPP and 1.5 M 0.82 M IPP and 1.7 M DBPC in C,H.~CI DBPC in C.,CI,~, 3 hr 5 h r a t 60 at 75 (tool tool of peroxide decomposed)

H

C

1.7

1.7

o

0

/\ CH3

OH

CH~

\

/

C=O

CHj .Coupling productls) between (CH~hCHOCO~ and ar} Iox~ radical from DBPC HCI

0.4 _+ 0.1

0

0.4 +_ 1

0

Thermal decomposition of IPP It is seen that in both solvents talthough in C6HsCI much less so than in C2CI~J not only coupling products of (CHabCHOCO ~ radicals with solvent but also coupling products of the isopropanol derived (CHah ( ' - - O H radicals with solvent are formed. Apart from these coupling products with solvent, other interesting by-products are shown in Table 3. Their formation will be discussed later. Because DBPC has such a large effect on the CIDNP effects, we also carried out an analysis of the products after complete decomposition of IPP in the presence of a large excess of DBPC. The result of this analysis, based on NMR peak integral measurements, is found in Table 4.

1131

effect for isopropanol and the role played by the isopropanol radical. At't'lOllt

~

cII li ,'i.',iJOil

One of the possible routes for acetone formation. viz. disproportionation of two C3H.,OCO~ radicals inside the solvent cage. can be excluded because in the presence of DBPC (a very eh~cient trap for free radicals but not for "'caged radicals") no acetone is formed. Two other possible routes, analogous with those'already suggested by Van Sickle [3] for cyclohexanone formation from dicyclohexyl peroxydicarbonate, are: induced decomposition of IPP via a radical chain mechanism

C3H~OCO" + IPP--~ CsH~OCOH + "C(CH3):--O--CO--O---C---O---C~H~

iK

ii

o

ii

o

Call-,OH + CO2

(CH3)2CO+ CO2 + CsH-,O---CO-

Pr

O and termination by recombination of two CsH-,OCO" radicals:

I

O 2(CHs)2CH---OCO. ~ (CHs):CHOCOH + .C(CHsh--OCO-

li

I

o

Ii

!r

o

o

( C H s h C H - - O H + CO, Table 4 shows that in both solvents the presence of excess DBPC prevents the formation of acetone. Yields of isopropanol are equal in both solvents and

much higher than in the absence of DBPC. In tetrachloroethylene the formmion of HC1 is completely prevented by excess DBPC. The determination of

(CHshCO + CO~ Although both routes seem likely for the formation of unpolarized acetone neither can explain the formation of polarized acetone. According to CIDNP theory [5] the observed "net polarization" of acetone must involve the formation of radical pairs in which the two radicals have different ,q values. A fourth route to acetone meets this requirement:

C,HTO--C---O. + "C(CH3):--O--C--O--O--C--OCsH~ O

O

--.

O

C3H~O--C--O" -CJCHs)~--O---C---O--O--~--OC3H~ - . acetone + other products O

O

g = 2.0058,

O

g = 2.0026

"'F pair" (A,q :6 0). the amount of coupling product(s) between (CHahCHOCO] and aryloxyradical was not very accurate because of interference of NMR peaks from excess DBPC and dimeric products from DBPC. Nevertheless the results indicate a fairly satislactory IPP product balance.

If it is assumed that acetone is formed as a combination product of this radical pair. the Kaptein rule for the net effect [8] predicts the observed emission : NE = j+EA#A, = + + - + = -

in the presence of DBPC all of the above three routes to acetone are blocked including the one leading to polarized acetone because of the following fast reaction :

DISCUSSION

In this discussion we .first concentrate on the observed emission of acetone. Subsequently we focus attention on the mechanism of by-product formation in tetrachloroethylene and the t;Iosely connected CIDNP emissions at ,6 = 6. Finally we turn to the AE

= emission

O

1132

E.F.J. DUYNSTEE.M. L. ESSERand R. SCHELLEKENS The large amount of HCI (see Table 2) is formed because the c r radicals also abstr:act H atoms. It seems very likely that two of the emissions observed near 6 = 6 must be attributed to the --CCI,H pro-

The only remaining question is the exact way in which acetone is formed from the above F pair. There are at least three possibilities (underlined protons are polarized):

c_H3

c%

C-~H7"0 "C'O'C-O'C-O'O-C" O"C~ H~ ,~q

~

II

I

II o

II

o cH_3o

/

.

o

CH

I~ pair

"

~C.

C-,,O I

cH~

CH

CH3

• C.M.,O-C-OM+ .C-O-O-O-O-C-O-C..--.- C = 0 , II

I

C3H?O-

II

I

,tc.

I

H + C-O-C"O-O'C-O-C.H.--=.. C = 0 0

I

II

If

CH

0

0

-3

The last possibility looks attractive but it requires that we should also see an emission near ~ 4.7 from the intermediate propenylperoxydicarbonate (PPD).

'~

efc.

l

"

CH

-3

tons in the compounds M = 268 and M = 200. The polarization in these compounds is probably created via formation of an F pair with Ag -'- 0 and combination/disproportionation of this F pair:

CH~

CH2

I

I

R---CCIr--CCI~ •C--O--<]--<)--O--C--O--C3H~ CH 3

O

-- R--CCIz--CCI2H

+ C--49--C--(~--OC3H~

O

CH3

O

O

g = 2.0080, g = 2.0026 (R

The absence of CIDNP in the vinyl region does not disprove this pathway because the steady state concentration of PPD could be very low.

or

/

CH~

CI'I 2 = C

\ O---

poses a problem unless the unsaturated compound produces acetone in a very fast reaction. The third emission near 6 = 6 probably also comes from a compound with a ---CCI2--CCI2H group but its concen-cl.

C3H~--~l---O. + CCI, = CC12-----C3H,O--~q--O-~CI2--CCI~) ~ O

M = 232

~

O

M

- C.,CI,

- 3clFormation of M = 232 and M = 326 involves production of c r radicals by/1-fission. Addition of these CI" radicals to solvent seems a likely route to the three solvent transformation products in the right column of Table 3:

CI)

but the fact that we have never observed proton emission near ~ = 4.7

By-products i, C2CI4 and the emissions at ~ ~ 6

In the discussion on the by-product formation in C.,CI,, we first consider the first 7 products in the right column of Table 3. The three last products. involving the isopropanol radical, will be discussed later. A likely route to the first four of these products seems the following:

= C3HTOCO

= 268

"~

M

302

M

325

tration is so low that it escaped detection by GC:MS. Because we know how efficiently DBPC scavenges the (CH3)2CH---O----CO~ radicals, it is no surprise that in the presence of DBPC all reactions with C.,C14 and

CI. + CCI2 = CCI 2 ~ CCI 3 - CCI~

~ -

.

M = 200 234 M

258

Thermal decomposition of IPP therefore also formation of polarized RCCI:---CCI,H are prevented. The AE effect in isopropanol and the role o.f the ( C H 3)2--C--OH radical

As stated above the most likely route for isopropanol formation is via the half-ester of carbonic acid. However this route is not likely to be responsible for the formation of AE polarized isopropanol. In our opinion the formation of this AE polarized isopropanol is closely connected with abstraction of the secondary H-atoms of isopropanol. The formation of M --- 170 and M - 118 in chlorobenzene and the formation of M = 188, M = 282 and M = 256 in tetrachloroethylene tsee Table 31 clearly shows that in both solvents this H-abstraction from isopropanol occurs. It seems likely that the AE multiplet effect in isopropanol is generated by combina-

H

1133

IPP. The formation of isopropanol radical pairs is nicely confirmed by the presence of pinacol IM = 118). the combination product from this radical pair. Because the four CH3 groups in pinacol give a singlet, the muhiplet effect cannot be seen in pinacol. With the role of ICH~L,C--OH radicals thus firmly established, the following question arises: how do we know that it is not the (CH3)2C--OH radical which is responsible for the net effects observed in acetone. M -- 268 and M = 200'? We therefore carried out a decomposition (0.7 M IPP in C~HsC! at 60") in the presence of 0.30 M butanol-2. It was assumed that under this condition the secondarx H from butanol-2 will be abstracted as is also the case for the secondar.~ H of isopropanol. If the alcohol-derived radical does indeed play a role. emission of the underlined protons in butanone should be found:

\ 0

C3H,OC--O.

CH3

\/

---, C3H,OCOH + O = C

.C

\

\

O

CH3

/ 0

CH:

CH:

\

\

CH3

CH3 (or CaH,rOCO--CCI2--CC12H)

(or C3H,rOCO--CCh--CCll)

H

O

O tion/disproportionation of a F pair formed by two isopropanoi radicals:

T h e - - C H , - - quartet of butanone at b = 2.12. which could be observed separated from the singlet of CH3 O H H O CH3 CH3 O H CH3 acetone at fi =1.84, did not show emission but \ \/ \/ \/ absorption, increasing with time as expected for a + C------O non-polarized - - C H 2 - - group. In our view this exC. .C -. C / / \ /\ periment proves unambiguousl.~ that the CHa CH3 CH3 CH~ I~I. (CH3),C--OH radical does not plax a role in the F pair (Ay = 0). generation of net emission effects observed in acetone and the -:--CCI:--CCI,H containing products. The Kaptein rule for the multiplet effect [8] preFinally we comment on the cyclic carbonate ester dicts the observed AE effect in the septet of the under(M = 1021 mentioned in Table 3. Its formation is lined proton: remarkable because it indicates that induced deME = ~EAiAflo~ q = + + + + + - = = AE.. composition of IPP also occurs by H abstraction from one of the ---CH 3 groups of IPP: A similar AE effect in the CH 3 doublet of isopropa-

~'-----'IPP

.-c< / C~t3

H

, O-C-O~>C-O . II 0 0

CH 3

% C~3

nol is obscured by the fact that in chlorobenzene, in which solvent this AE effect is best observed (Table I i, the CH 3 doublet of isopropanol coincides with that of

/ ",

0

0

/ CH3

The yield of this cyclic carbonate ester is only very low and this type of induced decomposition is by far not as important as the one discussed earlier [9].

1134

E. F. J. DL'Y~STEE.M. L. ESSERand R. SCHELLEKENS

Acknowledgements--The authors thank H. Theeuwen, T. Linnartz. L. Cremers and H. Omloo for the GC/MS analyses. Miss H. Greefkes for carrying out preliminary experiments with IPP and Dr E. Konijnenberg for valuable discussions.

REFERENCES I. C. S. Sheppard and V. R. Kamath. Pol.rm. E,#ml Sci. 19. 597 (1979). 2. R. Hiatt In Or.qanic Peroxides (Edited by D. Swern). Vol. II. p. 799. Wiley-imerscience. New York (1971). 3. D. E. Van Sickle. J. ory. Chem. 34. 3446 (1969). 4. Z. S. Kartasheva. A. B. Gagarina and N. M. Emanuel. Dokl. Chem. (Eng. Trans.) 212. 710 (1973): ibidem 229. 6~5 (1976).

5. R. Kaptein. In Chemically Induced Magnetic Polarisation (Edited by L. T. Muus, P. W. Atkins. K. A. McLauchlan and J. B. Pedersen), p. 1. Reidel, Dordrecht (1977). 6. R. Kaptein and H, Fischer, In Chemically Induced Magnetic Polarisation (Edited by A. R, Lepley and (3. L. Closs), p. 137. t97. Wiley, New York (1973). 7. R. A. Cooper, Thesis. Brown University, Providence, p. 119 (1971). 8. R. Kaptein, Chem. Commun. 732 (1971). 9. Another mechanism for formation of M = 102 was suggested by a referee and involved the

CH 2==~--O--C--O. CH3

0

radical formed by homoiysis of the pronenylperoxydicarbonate (PPD).