Glycidyl Ether Reactions with Is, Phenols, Carboxylic Acids, id Anhydrides LEON SHECHTER
AND JOHN WYNSTRA
Bakelite Co., A Division of Union Carbide and Carbon Corp., Bloomfield, N. 1. Epoxy resin curing processes are complicated in their chemistry in that competitive reactions are possible. Model compound reactions and analytical procedures were set up to estimate the relative significance of the different possible routes in the reactions of a glycidyl ether with alcohols, phenols, carboxylic acids, and acid anhydrides. Noncatalyzed, these reactions were sluggish enough to require temperatures of 200' C. or higher to proceed at a reasonable rate; all the possible competing reactions were found to take place. With the addition of a base catalyst these reactions were considerably accelerated, and with acidic reactants they became highly selective. A generalized reaction mechanism suggested by these results i s proposed, and certain other phenomena are shown to be consistent with it.
LTHOUGH the technology of the epoxy resins is well deA veloped, the complexity of the potential curing reactions prompted a detailed study of the chemistry of glycidyl ethers. Becauee the epoxy resin casting systems of commerce reach the gel stage very quickly, model compounds were chosen to represent the epoxy resin and the other reactant(s). Generally, these compounds were monofunctional so that the reaction products remained soluble, permitting the use of conventional analytical methods to determine the extent of the reaction. The model epoxy compound used in most d this work was phenyl glycidyl ether which had been purified to contain less than 0.01% ' hydrolyzable chlorine. Commercial phenyl glycidyl ether was pursed to this specification by slow distillation at about 1-mm. pressure through a 36-inch column packed with glaas helices and collection of the portion distilling within a 0.5' C. range, In some few instances a commercial batch of the diglycidyl ether of 2,2-di-p-hydroxyphenyl propane (Bakelite ERL-2774) was used as the model epoxide. The curing of epoxy resins may be divided into two classescuring with hardeners and curing by catalysts. Hardeners are defined aa polyfunctional compounds which are used with an epoxy resin in a stoichiometric or nearstoichiometric-ratio. Most of the model compound ROH reactions described here and in the accompanying article (IO) are illustrative of epoxy resin/ hardener curing systems. Catalysts, on the other hand, are compounds that cause the epoxy resin to self-polymerize. They may be monofunctional and are always used in much lower amounts than the stoichiometric. The model compound reactions described were carried out under conditions similar to those encountered commercially-for instance, in the absence of solvents and at practical temperatures. The experimental attack on any given epoxy resin/hardener reaction was that of anticipating the possible competing processes and devising methods of analysis (for functional groups) to estimate their relative significance and rate, In the sections that follow, the model compound reactions of a glycidyl ether with alcohols, phenols, carboxylic acids, and acid anhydrides are described. I n most instances these reactants were studied in the absence of catalysts and also in the presence of base and acid catalysts. The analytical data are interpreted in terms of the products formed, and a mechanism by which these results might
have been obtained is suggested for some. Finally, the individual baae-catalyzed mechanisms proposed are correlated into one generalized mechanism, and certain features with which this mechanism is consistent are pointed out. The individual experiments are not described in detail because of space limitations. Instead, the general plan of attack and the analytical methods used for each type of model compound reaction are given, and the data obtained are presented graphically. Alcohol-Glycidyl Ether Reaction The alcohol-glycidyl ether reaction is complicated in that a mixture of isomeric products may be formed and in that new hydroxyl groups, also capable of reacting with epoxide, are formed as the reaction proceeds: 0
+ CH2-CH-
86
0 ''
etc.
--
d
\--+ HocH2cHhR
CH2CH-
2 isomers
\/ 0
(1)
An analytical limitation is that the hydroxyl concentration remains constant so that only the over-all reaction may be followed (by epoxide analysis) Analysis for epoxide was by the pyridinium chloride method (6). Xoncatalyzed, the alcohol-glycidyl ether reaction LYas found to be rather sluggish; a temperature of 200' C. or higher was required to realize a conveniently rapid rate. The choice of a model alcohol was, therefore, rather limited. Commercial grades of the dihydroxy compounds bisphenol-acetone, diethylene glycol, dipropylene glycol, and pinacol were chosen to serve as model phenolic, primary, secondary, and tertiary alcohols.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 48, No. 1
THERMOSETTING RESINS ~~
~
~
They were reacted with the stoichiometric proportion of ERL2774, and a solution viscosity of the polymeric products Waf3 taken aa a measure of the extent of the reaction. Figure 1 indicates a reactivity order of phenol > primary > secondary > tertiary alcohol. Epoxide analysis data supported the same reactivity order. Because of the isomer effects already mentioned, however, no attempt was made to convert these data into quantitative rate differences.
O
I
L
1
2
!---HOURS
Figure 1.
I
I AT 2040 DEG. c.
I
6
I
8
I 12
IO
I 14
Tertiary amines and a quaternary ammonium hydroxide were tested in amounts equivalent to 0.20% potassium hydroxide and found to be somewhat more powerful catalysts (Figure 2). The potassium hydroxide and the quaternary hydroxide data satisfied first-order kinetics whereas the tertiary amine data fitted second-order kinetics. From this it was surmised that the two types of catalysts were functioning in a somewhat different manner. For the tertiary amine, the similarity in effectiveness of benzyldimethylamine and dimethylaniline suggested that the base strength of the amine was not an important factor in determining its catalytic power. Some information on the products of the alcohol-glycidyl ether reaction was obtained in two experiments with phenyl glycidyl ether and isopropyl alcohol (in a 1 :1molar ratio) using potassium hydroxide and benzyldimethylamine as catalysts. Each product, after a nearly complete consumption of epoxide, was stripped to recover any unreacted alcohol. In each caae, about 80% of the charged alcohol was recovered unreacted, indicating that the reaction was largely a self-polymerlzation of epoxide. Additional clews on the mechanism of the reaction were deduced from the relative rates of reaction found when the glycidyl ether was added to different amounts of alcohol. With tertiary amine as a catalyst the reaction rate (Figure 3) was sharply dependent,
0 9 k
Reaction of Bakelite ERL-2774 with diols
Although acid-catalyzed and noncatalyzed alcohol-epoxide reactions are known to lead to a mixture of isomeric products ( l a ) , the base-catalyzed reaction has been reported (S)to give a product that is almost exclusively a primary ether and a secondary alcohol. If, then, a secondary alcohol is reacted with a glycidyl ether under conditions of base catalysis (to yield a new secondary alcohol), the almost exclusive reaction will be that of glycidyl ether with secondary alcohol. The reaction would still be complicated by the competition of the new hydroxyls with the starting alcohol for epoxide. Phenyl glycidyl ether and dipropylene glycol were chosen as the model system. With 0.20% (by weight of reactants) of potassium hydroxide as catalyst, a temperature of 100" C. gave a convenient rate.
-e-+-
Equivalent amounts of epoxide and hydroxyl held without catalyst
a .
4
I
16
AT 250 DEG.c.----
e
020% 048% Q - 0 43% 4-0.60%
--_-_
--
POTASSIUM HYDROXIM BENZYLDIMETHYLAMINE DIMETHYLANILINE BENZYLTRIMETHYL AMMONIUM HYDROXIDE
---cOC83% SODIUM METAL (TO FORM ALKOXIDE ION)
I 1
\\
HOURS AT 100 DEG C
Figure 2. Base catalysts for reaction of equivalent amounts of phenyl glycidyl ether and dipropylene glycol
------o - - - - - - - a - - -
-
09
-.
0 48% BENZYLDIMETHYLAMINE CATALYST
-200
---0-0 L o .
405
P
.-.--. -. -o_.
EQUIV OHPER EPOXIDE
---
-*._
x
08
25 EQUIV OH PER EPOXIDE
------0--000
P
0 20% POTASSIUM HYDROXIDE CATALYST
EQUIVOH PER EPOXIDE I 00 EQUIV OH PER EPOXIDE
8
-1.00
- - -0-0
--- -- - +-.O
EQUlV OH PER EPOXIDE
25 EQUIV. OH PER EPOXIDE 00 EQUIV. OH PER EPOXIDE
--- - - - _ _ _ --_
Figure 3. Effect of alcohol concentration on alcoholglycidyl ether reaction-amine catalyst
Figure 4.
Effect of alcohol concentration on alcoholglycidyl ether reaction-KOH catalyst
Varied ratios of dipropylene glycol and phenyl glycidyl ether reacted under standardized conditions
Ianuary 1956
INDUSTRIAL AND ENGIN.EERING CHEMISTRY
87
on hydroxyl concentration; in the absence of alcohol this catalyst did not polymerize phenyl glycidyl ether. With a fixed alkali catalyst (Figure 4) the reaction rate was much less dependent on hydroxyl concentration, and polymerization was nearly as rapid in the absence as in the presence of alcohol. The polymerization character and the kinetics of the basecatalyzed alcohol-glycidyl ether reaction are consistent with the following mechanism: First, aome concentration of alkoxide ion is generated from the catalyst and either the epoxide or the alcohol. This anion then reacts with epoxide to form a new allroxide ion which can continue to add epoxide:
1
G9 k -
ADDED TO EPOXICE PLUS EQUIV OF OH -. +--.-0.10 EQUIV. OF OH _ _ 0_ - -_ - - .- 0.00 EQUIV OF OH 0.10%SnCI,
-1,OO
0
0,3L
I
I
0 2‘
etc.
ROCHZCH-
I
(2)
Figure 6.
OCH2CH-
I
Stannic chloride-catalyzed reactions
glycidyl ether
00
the addition of phenyl glycidyl ether and the finding (also shown in Figure 2) of a rate of reaction very similar to that with an equivalent amount of potassium hydroxide as catalyst. The pseudo first-order kinetics of the process can be explained on the basis of the constancy of alkoxide ion concentration.
Ratc = k [Roe]
[
yC H :,]-
= 1;’ [CCOHOL-EPOXIDE
--
REACTION -4
-03=
- - - -a ---_ _
z v)
-.
l o PHENOL-EPOXIDE . REACTION -
n
- 4 4 5
n
\
-05 w
----o--