Environmentally sound polymers


Material Information

Environmentally sound polymers the use of azlactone-phenol adducts for thermoreversible covalent polymer linkages
Physical Description:
vi, 143 leaves : ill. ; 29 cm.
Engle, Lori Prutzman, 1962-
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1991.
Includes bibliographical references (leaves 135-142).
Statement of Responsibility:
by Lori Prutzman Engle.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001684222
notis - AHZ6206
oclc - 25060407
System ID:

Full Text






To my husband Mark, whose patience and loving

sacrifice has enabled me to realize a dream.


I would like to thank Professor Kenneth B. Wagener for his invaluable

assistance throughout my graduate school career. He allowed me the freedom to fail

and the self confidence to learn from failure; he gave me the freedom to succeed and

rejoiced over my successes. He never let me forget that I was free to make what I chose

out of my education, and most of all, he taught me never to give up.

Many warm thanks also go to all the members of Dr. Wagener's research

group, whose equal willingness to share scientific advice, glassware, and drinks at the

Purple Porpoise have made graduate life so joyous. Special thanks go to undergraduate

M. Heather Woodard, who shared this research for three semesters.

Finally, many thanks go to the people at the 3M Company, who have

sponsored this research for more than two years. Besides granting research

assistantships, the scientists involved with the project were helpful and interactive

through every stage of the work. The author can only hope that this work serves as

encouragement to sponsor future graduate assistantships.


ACKNOWLEDGEMENTS ................................... ..................... .. iii

ABSTRA CT............................................................................... v

CHAPTER 1. INTRODUCTION...................................................... 1

Recyclable Polymer Systems................................... ........... 1
A New Functionality for Thermally Reversible Covalent
Polymer Linkages.............................................................. 4
Thermally Reversible Physical Crosslinking............................... 14
History of Thermally Controlled Covalent Polymer Linkages................ 22
Azlactones in Polymer Chemistry.............................. .......... .. 31

CHAPTER 2. EXPERIMENTAL........................................................ 45

General Inform ation............................................................... 45
Synthesis, Purification, and Characterization of Model
Com pounds...................................................................... 47
Synthesis, Purification, and Characterization of Compounds
for Polymerizations and Crosslinking................................... 52
M odel Studies..................... ......................................... .... 57
Polymerization Reactions...................................... ............. 68
Crosslinking Reactions..................... ................................ 71
Error Analysis........................................ .......................... 74

CHAPTER 3. RESULTS AND DISCUSSION........... ................................ 78

A Model Study of the Forward Reaction................................... 79
A Model Study of the Reverse Reaction..................................... 92
A Model Study of Bisphenols VI with Azlactone I............................. 97
The Synthesis of Linear Polymers of Bisazlactone IV and Bisphenols
V a-d ............................................................................. 106
The Synthesis of Network Polymers of Poly(vinyl azlactone) Y, and
Bisphenols V la-d................................................................. 119

CHAPTER 4. SUMMARY AND CONCLUSIONS.................................. 131

REFER EN C ES............................................................. ............. 135

BIOGRAPHICAL SKETCH ................................................. 143

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Lori Prutzman Engle

May, 1991

Chairman: Dr. Kenneth B. Wagener
Major Department: Chemistry

In this work, azlactones (2-oxazolin-5-ones) and phenols were used for

thermally controlled adduct formation. In small molecule chemistry, the 1:1 adduct

formation of a model azlactone, 2-isopropyl-4,4-dimethyl-2-oxazolin-5-one, with

monofunctional phenols was enhanced when phenols with electron withdrawing

groups at the para position were used. In addition, the thermally triggered reverse

reaction to reform azlactone and phenol was enhanced by the same electron

withdrawing substituents. Lewis acid catalysis, however, reverses the relative

reactivity of phenols in the forward reaction, making phenols with electron

withdrawing groups the least reactive in a series of phenol nucleophiles examined.

When this chemistry was extended to the linear, stepwise addition

polymerization of 2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one) with

bisphenols, the trends of reactivity observed in the small molecule chemistry were

generally consistent over a series of bisphenols. Thus linear polymer was formed when

the bisphenol contained an electron withdrawing group in at the site para to the phenol

hydroxyl moiety. The reverse reaction was also enhanced by this type of substitution.

External acid and base catalysis did not produce straightforward results in these

studies. An additional model study using bisphenols rather than phenols more

satisfactorily predicted the reaction behavior with azlactones.

Thermoreversible network formation was also studied as an extension of this

research. Poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one) and a series of bisphenols were

examined for sufficient crosslinking to yield a gelled network. Again, reaction

enhancement was found to follow expected trends based on the models.

Thermoreversibility within the range of thermal stability of the polymer backbone was

observed for one of the bisphenols. This bisphenol gelled the polymer in DMF

solution; the gel reversed upon heating in refluxing DMF under Argon. Cycling of the

gelled network was observed twenty or more times before failure to undergo gelation.


Recyclable Polymer Systems

Since the discovery of synthetic polymers in the 1930s,1 polymer chemists have

sought tough, long lasting materials for various applications. However, in recent years

there has been a growing awareness of environmental issues, one of which, ironically,

concerns the longevity of synthetic polymers. Synthetic plastics and rubbers are indeed

long lasting, sufficiently so to cause appreciable buildup of wastes during their short

existence. The polymer industry has taken some action with respect to degradable and

recyclable materials, but much remains to be accomplished. Polymeric materials that can

be easily degraded or recycled should find ready application in industry.

A degradable polymer contains functionalities that cause the chains to break down

into small-molecule products by the chemical or natural action of light, heat, or some

biological entity such as bacteria. A recyclable material, however, is of higher utility than

a degradable one. A recyclable polymer traditionally is a linear polymer which can be

reshaped in the melt. On the other hand, a linear polymer that could be broken down

upon some treatment to regenerate monomers, ie., depolymerize, would be of higher

utility because the monomers could then be purified and reused in applications other than

reforming the polymer.

Thus one of the goals of this research is to introduce a linear polymer that

depolymerizes via simple addition of heat, producing the monomers that were originally

used to synthesize the polymer.

A system that is capable of regenerating linear polymer from a crosslinked
material is also of interest in this research. Many polymers in their linear form are soluble

and/or able to melt at temperatures below the point where irreversible degradation occurs.

Thus in the linear form they may be recycled. However, when the linear chains are

covalently linked to each other in a three-dimensional network, the same polymers are

insoluble and do not flow above the glass transition or melting temperature.1 These

materials, called thermosets, generally cannot be recycled into usable materials. A

recyclable network, then, would ideally have the mechanical integrity of a thermoset at

lower temperatures, but would become flowable (thermoplastic) or soluble at higher

temperatures due to "decrosslinking." In order to have the mechanical properties of a

thermoset, the envisioned network should be covalently crosslinked: physical crosslinks

do not have the bonding strength of covalent crosslinks.

The goal of the research is thus defined: a system that contains some
thermoreversible covalent linkage is desirable, wherein at moderate (less than 1000C)

temperatures linkages are formed. At elevated temperatures, the reverse reaction is

favored, causing regeneration of starting materials via the unlinking of covalent bonds.

This chemistry should be useful for the syntheses of linear polymer that can be recycled

to monomer, as well as for a crosslinked network that can be recycled to linear polymer.

These goals are schematically represented in Figure 1.

When considering polymer recycling from a practical standpoint, it would be an

advantage to design a system that would not require the addition of catalysts or other

chemicals in order to trigger either the forward reaction (network formation, or

polymerization) or the recycling reaction (network decrosslinking, or depolymerization).

Rather, the ideal system would be capable of forward and reverse reaction through

regulation of temperature. Polymerization or network formation should be facilitated by

simple 1:1 addition of reagents to form the desired linkages.

a. nAA + nBB --AABB-
A \ in

b. nAAB -A

c- TA A A A ool AA A AS

Figure 1. Scheme for thermoreversible covalent bonding to form a., b. linear polymer
and c. network polymer.

There are few instances in the literature of thermally controlled covalent polymer

systems. Of those reported, even fewer claim success in recycling of linear or network

polymers (ie., more than one polymerization depolymerization cycle). The history of

thermally controlled covalent polymer linkages is covered in depth later in this chapter.

For thermoreversible linear polymerizations, the type of chemistry (chain vs. step

polymerization; condensation vs. addition reactions) best suited for the desired system

should be determined. Although both radical and ionic chain polymerizations are

reversible through the concept of ceiling temperature,1,2 once the reactive end is

terminated, depolymerization is usually not possible. For example, anionic

polymerization often results in a "living" anion at the end of the polymerization; if the

anion is not quenched, then the chain can be depolymerized by raising the temperature

above the ceiling temperature. Once exposed to the atmosphere, however, quenching of

anions occurs; therefore this route to thermal reversibility is impractical. Since stepwise

chemistry does not rely on a single reactive end group to propagate the reaction, but

rather on the reactive site of of each monomer molecule, it is by nature better suited for

the system to be developed.

Condensation polymerization (both step and chain type) by definition involves the

loss of some small molecule to form each linkage; in order to depolymerize, the small

molecule would have to be retained in the system. As stated above, the most convenient

situation is one where only temperature regulation is required to affect reversibility in the

system. This is the simplest system that could be envisioned. Stepwise addition

chemistry is thus the best suited for such a system.

Stepwise addition polymerization is in fact schematically represented in Figure la

and lb, where AA and BB difunctional monomers (or an AB type monomer) react with

1:1 adduct formation to give an (AB)n type polymer. The same idea is extended to the

crosslinking chemistry in Figure Ic. In this case, "A" functionalities should be present

along a polymer backbone such that a difunctional (or polyfunctional) "B" type

crosslinking molecule can react by 1:1 A:B addition to crosslink the chains. With the

addition of heat, the AB adducts so formed should undergo the reverse reaction to

A-functionalized polymer and B-functional crosslinker without the addition of other


The potential utility of such a system warrants study in this area. Covalent

crosslinking offers mechanical advantages to physically and ionically crosslinked

systems. Additionally, the reversible covalent linkage is of potential use as a method of

chain extension whereas non-covalent means of thermoreversible crosslinking cannot be

extended to polymerizations. Thus the research described in this thesis focuses on the

potential utility of a new system for thermoreversible covalent crosslinking and chain


A New Functionality for Thermally Reversible Covalent Polymer Linkages

In the research described herein, 2-oxazolin-5-ones ("azlactones"), shown in
Figure 2, are used to form thermally reversible covalent linkages with phenol functional


R2 H O
Figure 2. The reaction examined in this research. RI-R3 may be alkyl or aryl groups; X
may be O, S, or N; G is any group. In this research, X=0 and G=phenyl.

When disubstituted at the 4-site, azlactones act as electrophiles in a ring opening

reaction with a number of nucleophiles,3 forming 1:1 adducts. The reaction system must
contain an available proton for bonding to the amide nitrogen formed in the reaction.
When the nucleophilic entity HXG is a phenol, ie. X=O and G=phenyl, the adduct
formation reverses at elevated temperatures. The phenol nucleophile is preferentially
ejected, and azlactone is reformed.
Reversibility at temperatures below that at which thermal degradation appears to
occur is due to the large stabilizing effect of the phenyl ring on the oxygen anion as the
nucleophile is ejected. Equilibrium favors adduct formation at lower temperatures, but at
higher temperatures the closed ring is favored. Azlactones are also five membered rings

and as such present a favorable thermodynamic profile for ring closure.4
The reversibility of the azlactone phenol reaction has been examined for the
purpose of synthesizing recyclable covalent linear and network polymers as shown in
Figure 3. These are two examples of how this linkage might be used; several other uses
are also feasible. These examples are representative of the scope of use of the system.

Azlactones have precedent in polymer chemistry both as crosslinking

functionalities for linear polymers and as difunctionalized monomers for step

polymerization chemistry.5 In both of these cases, the key reaction involves the

nucleophilic ring opening of azlactones at the lactone site. By using phenols as

nucleophiles, these reactions are reversible: the phenyl functionality is sufficiently

stabilizing to allow ejection of the nucleophile below thermal degradation temperatures.

In order to orient the proposed research towards practical applications, the linkage

must remain intact over a temperature range wide enough to be useful, but is "triggered"

to reverse at some higher temperature within the range of practical application. At

temperatures lower than the trigger temperature, equilibrium should favor the adduct.

Ideally the system will undergo linkage formation via 1:1 addition at temperatures of less

than 1000C. The reverse reaction should occur at temperatures greater than 1000C, such

that the temperature range of boiling water will not cause significant reversal. However,

the temperature at any time probably should not exceed 3000C, so as to avoid irreversible

thermal degradation of materials.


a .n n n

AT^O R=H, CH3 A 2
R'= alkyl, aryl

b. ___ ____ stepwise
/ addition
R, R'= alkyl, aryl

Figure 3. The goals of the research, a. Thermoreversible crosslinking of poly-(2-vinyl-
4,4-dimethyl-2-oxazolin-5-one) with bisphenol; b. Thermoreversible chain extension of a
bis(2-oxazolin-5-one) with bisphenol.

To "tailor" this temperature range of reactivity, substituents on the phenol ring

were examined in the course of this research. An electron-withdrawing substituent, such

as NO2, placed on a monofunctional phenol ring in an activating position was found to

shift the equilibrium towards the ring closed form. At elevated temperatures the

stabilizing effect of the nitro group on the phenolate anion causes ring closure to be

favorable as well. Therefore, the temperature at which the ring closed azlactone is

favored should be controllable via the manipulation of substituents on the phenol ring. To

test this hypothesis, other substitutents (OMe, H) were examined during the course of the

model study.

It should be noted, as the history of azlactones in polymer chemistry is described,

that the potential use of the proposed reaction scheme could apply to many more

situations than those pinpointed in this research.

Thermally Reversible Covalent Crosslinking and Polymerization

The thermodynamics of the 1:1 adduct formation in this system should involve a

reaction where at low temperatures, AB type adduct formation is favored (see Figure 1),

while at higher temperatures the A- and B-type substituents are favored. The situation is

described by examining the equilibrium constant, K, of such a reaction (see Figure 4).6 If


Equation 1. K = -
In K [A] [B]

d(On K)
Equation 2. d(I R

Figure 4. Equilibrium thermodynamic relationships which would hold for the envisioned system.

the enthalpy of reaction, AHO, is large and negative, then the slope of a plot of In K vs.

1/T is steep and positive; this type of reaction is sensitive to temperature changes. The

equilibrium constant K is large at lower temperatures and small at higher temperatures.

A good example of such a reaction is Diels Alder [4+2] cycloaddition, where

often high yields of adduct formation will occur at lower temperatures, and the reverse

reaction (the "retro Diels Alder" reaction) is favorable at higher temperatures.7 Indeed,

this very reaction has been examined for the purpose of thermoreversible crosslinking

and chain extension with some success. The linear polymerizations using this chemistry,

however, do not meet with as much success as the crosslinking schemes.

To synthesize a thermoreversible polymer, the requirements of the polymerization

chemistry must also be met. Stepwise addition polymerization is in theory best suited to

meet the requirements of the system as outlined above. A limitation of stepwise addition

polymerization is that adduct formation must be at least 99% complete in order to form

appreciable molecular weights. For AA + BB systems, very high purity of monomers is

required as a corollary to this general rule. The equilibrium constant, K, must therefore

be as large as possible for these systems during polymerization.

The theory underlying this premise was first described in 1936,8 and the widely

used relationship derived from this work is referred to as the Carothers equation. The

degree of polymerization is dependent on the purity of the monomers, their stoichiometric

balance in the polymerization solution and the total extent of reaction according to

Equation 3.

Equation 3. Xn =
1 + r 2rp

where Xn (same as Xn-bar, or average Xn) is the number average degree of

polymerization, r is the stoichiometric balance, and p is the extent of reaction of the

monomers expressed as a fraction. Thus, when the stoichiometric balance of monomers

is exactly 1, it can be seen that Equation 3 reduces to Equation 4.

Equation 4. Xn =

When the stepwise reaction of a stoichiometrically balanced reaction of AA and

BB difunctional monomers proceeds to 90%, the number average degree of

polymerization will be about 10; at 95%, Xn is 20; at 99%, Xn is 100. The stoichiometric

balance is crucial to obtaining the highest extent of reaction possible, since impurities

present in the monomers will lower the extent of reaction that can occur. For AB type

monomers, this problem does not exist. However, even for AB monomers the reaction

must essentially be complete before high polymer is realized.

Crosslinking reactions based on the same chemistry as stepwise polymerizations

are not subject to the same stringent requirements. Only a few percent crosslinking results

in a gel. Equations to describe crosslinking are known.1 The fraction of crosslinkable

functionalities on the chain is defined as q, and degree of polymerization of the chain is x.

For a polydisperse species, x = x, (or xw-bar, or average x,), the weight-average degree

of polymerization. Thus the functionality of the chain is defined as the total number of

crosslinkable groups, qxw. The fraction of total available crosslinks actually formed is p,

and crosslinking occurs at the critical value Pc, corresponding to an average of one

crosslink per two polymer chains. This leads to Equation 5.

Equation 5. pcqwx = 1

As p increases from Pc to 1, the weight fraction of non-networked polymer goes

from one to zero.

Model Studies of the Proposed System

Understanding the azlactone phenol reactions described above is a goal of the

research, as used to creating thermally reversible covalent polymer linkages. It was

proposed that, in addition to the polymerization and crosslinking experiments, a

mechanistic study should be performed on the azlactone phenol system, as this had not

been done previously. In order to overcome the difficulties inherent in the quantification

of polymerization and crosslinking systems, the mechanistic study was performed using

monofunctional azlactone and phenol adducts. The study was concerned mainly with the

forward reaction, but encompassed the reverse reaction in that if the system provided

reversibility, then the mechanistic study should be consistent with this finding. The

mechanism of the reverse reaction is assumed to be the reverse of the forward reaction.

Model compounds were chosen to mimic the compounds used for crosslinking

reactions. These compounds are depicted in Figure 5. Model compound

2-Isopropyl-4,4-dimethyl-2-oxazolin-5-one, I, has as its pertinent features substitution at

the 4 site of minimum steric bulk, and isopropyl substitution at the 2 site in order to

mimic a polyethylene backbone such as would be the case for radically polymerized

2-vinyl-4,4-dimethyl-2-oxazolin-5-one. The model also closely resembles a bisazlactone

in which the connecting group between azlactone functionalities is an alkyl group, thus

predictions can be made about step addition chemistry of bisphenols with alkyl



I Ia X=N02 IIla X=N02
lc X=OMe IIIc X=OMe
Figure 5. Model compounds for the study of the reaction mechanism of I and II.
Compounds IIIa-c are the expected reaction products of I and IIa-c, respectively.

The nucleophiles for the model study show the effects of various substituents on

both the forward and reverse reactions. Compound IIa contains a nitro group at the site

para to the phenol hydroxyl group, and thus it has an electron withdrawing effect on the

nucleophilic oxygen primarily by resonance. IIa is therefore a weaker nucleophile than

phenol itself. IIc contains an electron donating methoxy group in direct resonance with

the phenol hydroxyl group and therefore causes the oxygen nucleophile to be more

nucleophilic than phenol. The reactivity of phenol itself can then be compared to these

two substituted phenol analogs in the nucleophilic ring opening of I.

Compounds ITa-c are simply the 1:1 adducts of I and ITa-c. The synthesis and

purification of these adducts enabled the reversibility of the reaction to be studied by the

same models that were used to study the forward reaction. Other species III were

introduced during study of the reverse reaction as a means of modelling the reverse

reaction for the specific bisphenols used in the polymerizations and crosslinking.

Mechanistic studies were carried out using equilibration and kinetic data from the

forward reaction. Quantification of phenomena was possible in the model study since

there is only one possible product of the nucleophilic ring opening reaction. Due to the

difficulties in studying reaction mechanisms at high temperatures, actual mechanistic

work was not attempted for the reverse reaction; rather, as stated above, the assumption

was made that the reverse reaction is equal but opposite to the forward reaction.

The uncatalyzed system was studied principally, but at times catalysts were

examined for mechanistic purposes. Lewis and Bronsted acids were examined as well as

Lewis bases. However, major emphasis was placed on the uncatalyzed system since the

simplest polymerization system would be where catalysts or other additives are not

included in the synthesis / recycling scheme. In addition, it became clear during the

course of the work that catalysts cause degradation at high temperatures, so it is best in

this case to have a catalyst free system.

In the model study, reactions were examined in highly concentrated solutions in

order to more closely mimic the actual situation often encountered in polymer chemistry.

Added solvents raise the costs of synthesis and recycling; more importantly, higher

concentrations of reagents give higher rates and ultimate yields of product formation.

This general knowledge is especially important to step polymerization chemistry, where

very high yields are required to produce polymers of appreciable degree of

polymerization. Also, the reverse reaction involves temperatures well above the boiling

point for many common solvents. If the reverse reaction could be accomplished in the

melt, the system would be greatly simplified. For this reason, the model studies of the

forward were examined for the case of both dilute and concentrated solutions, and the

adducts IIl were examined for the reverse reaction both neat and in dilute solution.

Polymerization and Crosslinking Experiments

The model studies demonstrated information about the mechanism of reaction of

the azlactone phenol system, and this information was used to predict which reactions

might be successful for thermally reversible polymerization and crosslinking schemes as

shown in Figure 3. For example, functionalized bisphenols were chosen for step

polymerization based on which phenol para substituents imparted a range of reversibility

that is convenient for practical purposes as stated above. The same bisphenols were used

to crosslink an azlactone functionalized polymer. Information regarding expected

reaction rates, effectiveness of catalysts, etc. obtained from the model studies are of great

utility in the design of an optimized polymerization / crosslinking system.

Poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one), that is, poly(2-vinyl-4,4-dimethyl

azlactone), Y, was the polymer used for crosslinking reactions with bisphenols. It was

the structure of the repeat unit that led to the design of model azlactone I. Bisazlactone

2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one), IV, with a C-4 chain separating

the azlactone moieties, was used for stepwise polymerization with bisphenols. The C-4

chain is also well represented by the model azlactone and is flexible; solubility problems

in the system were lessened by using this bisazlactone over the stiffer bisazlactone

monomers such as para-phenylene bisazlactones. It is therefore the first choice for the

polymerization study. The azlactone structures, shown in Figure 6, are available through

known synthetic routes.9,10

Since the effective structures of the bisphenols must, by the nature of the

substitution, differ from those used in the model study, a final model study was

performed on the actual bisphenols to be used in the polymerizations and crosslinking

experiments. By adding 2 equivalents of I to each of the bisphenols used, the relative

reactivity of the bisphenols were examined. Only three possibilities exist in this case:

(CH2)4 0 n

Figure 6. Azlactones IV, 2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one) and V,

unreacted bisphenol, and bisphenol monosubstituted and disubstituted by the reaction

with I are possible products. A better prediction was made about the exact temperature of

reversal of the adduct forming reaction, since this model is closer to the actual

polymerization and crosslinking schemes. Since the larger effect by far was expected to

be due to the functionalities on the phenol ring, rather than the bisazlactones, further

model studies were deemed unnecessary.

At the conclusion of this research, a system of step polymerization and

crosslinking is demonstrated, wherein the linkages formed are thermally reversible in the

range 1000C-3000C. At lower temperatures, the polymeric or crosslinked form is favored.

The polymerization and crosslinking schemes were worked out to meet the requirements

for the ideal system as outlined in Figure 1, with some exceptions. Most notably,

degradation of monomers in the reverse polymerization reaction (depolymerization)

caused the system to frustrate attempts at cycling the polymer without repurification of


Thermally Reversible Physical Crosslinking

The thermal recycling of polymer systems is known. There are several instances

in the literature of physical linkages that are thermally reversible; for example,

crosslinking by hydrogen bonding and crystallization,11-42 ionic attraction,4348 and

glassy phase formation18,49-53 are all known. These physical crosslinks, made up by the

interactions of chain segments, form at lower temperatures and disassemble at elevated

temperatures; thus these polymers behave as thermoplastics in their macroscopic

properties. Additionally, no small molecules are released to form the crosslink sites, and

therefore none need be added back in to affect the reverse reaction. While not suited for

the synthesis of linear polymers, these methods have been used to form thermally

reversible crosslinks of a non-covalent nature.

The ionic crosslinks are somewhat anomalous: while some of the work done

involves aggregations of ionic salts and therefore are physical crosslinks, certain

reactions, notably quatemization of tertiary amines with alkyl halides to form quaternary

ammonium salts, are suitable for polymerization as well and are actually covalent in

nature. For cohesiveness, these reactions have been included in this section.

Crosslinking by Physical Phenomena

Physical crosslinking is caused by parts of polymer chains interacting with one

another (without the formation of covalent bonds) to the extent that they behave as

crosslink sites. This phenomenon is common and has been observed in the case of block

copolymers,18 random copolymers,23 and graft copolymers.18 For example, one

copolymer block may act as a crosslink functionality by crystallizing or forming a glassy

phase with other like segments of other chains as shown in Figure 7. Of the types of

synthetic copolymers observed to exhibit physical crosslinking behavior, the case of

block and segmented copolymers is prevalent in the literature. Therefore these types of

polymers only are considered.

The materials that result from physical crosslinking are thermoplastic in nature

because the interactions described above are significant at ambient temperatures, but are

easily disrupted by the addition of heat.18 This is what gives these materials their utility.

They behave as rubbers at low temperatures but become flowable, and thus can be

recycled, at higher temperatures. This also allows for easy processing of the materials.

Soft Segment Hard Segment

Figure 7. Schematic representation of physical crosslinking by phase separation of unlike
segments of a copolymer.

In the case of physical crosslinking in block copolymers, control of the block

architecture is essential for maximum interaction of like segments.18 Through the

development of living anionic copolymerization, a high degree of structure control is

possible. In principle, the same physical crosslinking schemes are available to graft

copolymers; however, their structures are usually less regular than those of block

copolymers and only limited success has been realized.2354-57

Phase separation of segments in block copolymers causes physical crosslinking

when amorphous "soft" blocks are copolymerized with non-flexible "hard" blocks. Very

often the hard segments form crystalline areas, while the soft segments form amorphous

domains. The hard segments are usually present as the minor substituent and, through

microphase separation of the two components, behave as separate "crosslink domains"

covalently bound to the continuous amorphous phase. The term "thermoplastic

elastomeric" behavior describes the elastomeric characteristics of these networks at lower

temperatures and the melt and flow characteristics of the linear polymers at higher

temperatures. Thus crosslinking interactions reverse above the Tg or Tm of the hard

phase. A widely studied example of this behavior is anionically polymerized styrene -

butadiene block copolymer, where butadiene segments serve as the amorphous "soft"

phase, while styrene segments behave as the "hard" phase which serve as the crosslink

sites. A-B-A, (AB)n, and radial (or "star") type block copolymers (simple A-B block

copolymers cannot form physical crosslinks) are known to exhibit this behavior.

As was mentioned earlier, stringent control of the block structure is

essential for effective physical crosslinking. For instance, butadiene 1,2 linkages in

SBR, or any randomness in the copolymerization of monomers, disrupt domain

formation. In fact, even the addition of SBR A-B block copolymer to an A-B-A block

copolymer will disrupt microphase separation.18 The requirements for microphase

separation are discussed theoretically by Meier,58 assuming complete structural purity

(including uniform block and chain lengths), the blocks must be of sufficient length to

afford domain formation. A crystalline domain will be stable in an amorphous matrix if

some minimum degree of polymerization is reached, and microphase separation will not

occur until this requirement is fulfilled. The theory in the case of crystallization of the

hard phase is an extension of the factors governing crystallization in homopolymers,49

which were first discussed by Flory.59

In the physically crosslinked state these materials behave as rubbers in their

macroscopic properties: they exhibit low modulus, high elongation and other properties

associated with rubbers. Rebound and permanent set properties, however, are generally

inferior to those of chemically crosslinked thermosets.1,2,60 Tougher materials, of a more

thermoplastic nature, are synthesized by making the hard segment the major substituent

of the block copolymers.18 In this case, the crosslink sites are formed by the soft or

"glassy" phase. For both crystalline and non-crystalline hard phases, toughness is

imparted by the presence of the soft domain crosslinks.61,62 In other words, the

amorphous crosslinks improve upon the brittleness of the dominant hard phase by

forming a crosslink, and may also impart flexibility via the presence of the soft phase


Several recent reviews and theoretical models are available on the phenomenon of

physical crosslinking via crystallization and glassy phase formation in copolymers.6367

Crosslinking by Other Weak Chain Chain Interactions

In addition to aiding in the formation of crystalline areas in block copolymers,

hydrogen bonding has been cited as a form of crosslinking in and of itself. For

instance, Longworth and Morawetz68 found that random copolymers of ten mole percent

acrylic acid with styrene had doubled viscosity values as compared to styrene

homopolymer (Figure 8). Esterification of acid functionalities with diazomethane

decreased the viscosity of the copolymer to approximately that of branched polyethylene.

Thus the hydrogen bonding interactions served to greatly increase the viscosity of the

linear polymer; however, due to its transient nature, even at room temperature the

viscosity did not increase to the point that the polymer could not flow. Hydrogen

bonding, then, is not effective as a true crosslink in this system.



Figure 8. Schematic representation of thermally reversible crosslinking via hydrogen
bond formation in a polymer functionalized by carboxylic acid groups. [Adapted from
ref. 68]

Other linear polymers have unusual properties that are attributed to hydrogen

bonding interactions. For example, polyurethanes are materials of high mechanical

integrity; their chain stiffness and high tensile strength is presumed to be due to hydrogen

bonding between the amide protons and carbonyl oxygens.1 However, resonance of the

amide functionality also could be responsible for strong dipole dipole interactions and

would explain the observed chain stiffness.

Other systems of linear polymers exhibit the same characteristics of what is

termed thermally reversible "gelation" due to such interactions as hydrogen bonding and

dipole dipole interactions; however, these are generally not high strength networks and

many are crosslinked primarily in the sense that they become insoluble.27-42

On the other hand, a recent study69 indicated that hydrogen bonding is effective in

some cases. Specifically, polyphosphazenes substituted with OH, OEt, and NH2Et

functionalities were found to undergo thermoreversible gelation that was attributed to


hydrogen bonding. Transitions in polymer solubility were linked to hydrophilic -

hydrophobic interactions. These polymers apparently decrosslinked at lower

temperatures; crosslinking reactions proceeded at elevated temperatures. Information on

the physical properties of the systems were not available at the time of writing.

Crosslinking by Ionic Interactions

Ionic interactions have also served as thermoreversible physical crosslink sites in

ionically functionalized polymers. A simple example, illustrated in Figure 9 as described

by Holliday,46 elucidates the use of difunctional and trifunctional (or higher) tertiary

amine "prepolymers" to create ionene network linkages with dialkyl halides. The ionene

networks are thermally reversible for certain values of n and m, due presumably in most

cases to dequaterization of the ammonium linkages. These linkages are actually covalent

in nature and are also included in the review of thermally controlled covalent systems.

X-(CH2)m-X + R2N--(CH2)n-NR2

= alkyl chain
cool A X = halide
polyfunctional amine R = alkyl, aryl

+x X X X X
/v .R NRvw.NR W NR -NR vvvvNRv\


Figure 9. Schematic representation of a thermoreversible ionene network. [Adapted from
ref. 46]

A recent study on the copolymerization of prepolymers of 1,6-bis(dimethylamino)

poly(tetramethylene oxide) with various dihalides showed that ionene formation in the

linear polymers is also thermally reversible.70 The ionenes exhibit good elastomeric

properties and are easily processed after thermal dequaternization (and thus

deionizationn") of the ammonium linkages. Here, then, is a system that follows the

desired design of a thermoreversible polymer system as stated; covalent linkages are

actually formed, so the scheme is applicable to thermoreversible chain extension as well

as crosslinking.

A study that illustrated the use of quaternary ammonium linkages as true ionic

crosslinks, acting by ionic attraction alone, involved the use of polymerized vinyl

aromatic aminimide compounds as shown in Figure 10.71,72 These repeat units, when

present at 4 to 10% in a styrene copolymer, caused the composition to exhibit a

thermoreversible crosslinking interaction.

vmCR-- CHav y
RR = H, Me, halogen
RI, R2, R3 = alkyl, aryl, alkaryl
n= 0 or 1

0 R,
(CF2)r-C-N-- -R2

Figure 10, This vinyl aromatic aminimide compound, when copolymerized at 4-10%
with another vinyl monomer, exhibited thermoreversible ionic crosslinking in a 5%
aqueous solution. [Adapted from ref. 72]

Other types of ionic crosslinks that are thermoreversible include mixtures of

polyammonium cations with polysulfonate anions, which form polyelectrolyte

complexes,46 polysaccharide salts,73,74 copolymers of polydienes, in which the residual

olefinic bond can be thermoreversibly coagulated by a divalent metal cation such as

calcium,75 and ionomers containing polycarboxylate or polysulfonate salts, crosslinked

by divalent cations such as calcium or aggregated with monovalent metals such as


While these systems generally exhibit good plastic properties at lower

temperatures, they are often difficult to melt process due to extremely high viscosities.

For instance, sulfonated polystyrene at only 2% sulfonation has a melt viscosity at 220C

of about 108 poise (sodium salt). The corresponding carboxylated polystyrene has a

viscosity of 105.5 poise under the same conditions, which is still much higher than that of

polystyrene itself.46 Thus, while technically thermoreversible, the ionic crosslinks are

still associated and/or aggregated at high temperatures; the interactions are of a transient

nature and so permit a highly viscous flow. It should be noted that for the sulfonated

polystyrene there exists the possibility of covalent crosslinking between any protonated

sulfonate groups and unreacted phenyl groups to form a sulfone linkage;2 this could

contribute to the observed viscosities.

The physical methods do not yield crosslinks of covalent integrity on microscopic

inspection. In addition, the extent of crystallization and glassy phase formation are

dependent on such variables as thermal history60 and solution characteristics.80'81

Consequently the macroscopic properties of physically crosslinked networks are often

inferior or irreproducible compared to irreversibly crosslinked covalent networks in terms

of stress strain properties.1 Ionic crosslinks offer some advantages for thermoreversible

crosslinking because the ionic bond is much stronger than other physical associations.

However, these bonds often are of sufficient strength to make the crosslinked network

highly viscous and difficult to process in the melt.

The general exception to this rule is the ionenes, which exhibit excellent

elastomeric properties in the quaternized form as well as being easily processed in the

melt. This probably stems from the fact that the quaternary ammonium bond, while ionic

in nature, actually involves a covalent bond between the nitrogen atom of the amine and

the carbon atom of the halide; hence its superior physical properties. When the

dequaternization is acheived, the components are no longer ionic and therefore do not

interact as strongly as the other ionic systems mentioned. This would contribute to their

observed ease of processing. It also makes a good case for the study of thermoreversible

covalent networks for reasons of superior processability.

Covalently crosslinked networks in general offer higher tensile strength,
modulus, tear resistance, and hardness; better rebound properties; lower permanent set

and elongation.1,2,60 But as stated above, three dimensional covalent networks are not

able to be recycled, as they are thermosetting in nature. The exceptions to this rule are

reversible covalent crosslinks. These linkages, such as the ionenes, are known, although

not many systems are to be found in the literature.

Covalent bonding systems also offer the possibility of reversible chain extension,

from which monomer can be regenerated. In principle, this may be of greater utility than

reversible crosslinking since the monomeric materials that are the products of recycling

can be purified and used for other applications besides reforming polymer.

A limited scope of reactions are available for thermally controlled covalent

linkages; the reactions must be adduct forming to prevent the necessity of returning a

small molecule substituent back into the system for the reverse reaction. Additionally, the

adduct should reverse at temperatures below that at which degradation of any of the

system elements occur.

A discussion on the principles and practice of prior art is useful to aid in the

design of a new system of this type.

History of Thermally Controlled Covalent Polymer Linkages

Only a few instances of thermally controllable covalent polymer linkages are

known in the literature. These consist of Diels Alder [4+2] cycloadditions,82-90 nitroso

dimerization,91,92 cyclic anhydride reaction to form ester linkages,9394 and aromatic

urethane formation.95 Of these four systems, none so far are acceptable for commercial
use. The ionenes as covered under reversible ionic bond formation are also included in
this group. In principle, living anionic or cationic polymerizations are also
thermoreversible above the ceiling temperature of the reaction, but useful materials are
obtained by quenching the ionic end of the polymer and at this point the polymerization is
no longer reversible. Therefore, these systems are not included in this review.

Diels Alder 14+21 Cycloaddition
The Diels Alder [4+2] crosslinking and step polymerization reactions are
schematically represented in Figure 11. This reaction offers a wide variety of functional

a. rz + -- oo l

b. cool

Figure 11. Schematic representation of reversible Diels Alder [4+2] a. Chain extension,
and b. Crosslinking of a linear polymer. [Adapted from ref. 96]

groups which are suitable for both polymerization and crosslinking. In addition, the
reaction is well-known as a thermally reversible reaction and therefore would seem to be
ideally suited for polymer recycling applications. However, the lack of success in using
this method for step polymerizations of bisdienes with bisdienophiles has limited the
utility of these systems for reversible polymerizations. In fact, the only instances where
Diels Alder 4+2 step polymerizations have yielded high molecular weight material are in
the cases where the reaction is made to be irreversible via the elimination of a small

molecule such as SO2, CO, CO2, or H2.96

An alternate route which has led to mixed results for chain extension via

thermoreversible Diels Alder adduct formation was studied by Kennedy and Carlson.82

Fractionated prepolymers of Mn ~5,000 to ~10,000 g/mol (Mn is defined as the number

average molecular weight) of the structure shown in Figure 12 were found to be

unreactive towards cyclopentadiene dimerization; trimethylsilylcyclopentadiene, a model

for the prepolymer, was also found to be unreactive compared to cyclopentadiene itself.

Presumably this is due to the facile fluxional isomerization of silylcyclopentadienes,

which is faster than 1,5 hydride shifts and results in 95% of the 5 isomer at room

temperature. The other two isomers account for 5% of the mixture.97


Figure 12. Isobutylene prepolymer used to make thermoreversible high polymer via
dimerization of cyclopentadiene moieties. The molecular weight of the prepolymers range
from about 5,000 to 10,000 g/mol. [Adapted from ref. 82]

These same prepolymers reacted in a stepwise fashion with N,N'-hexamethylene

bismaleimide to form high polymers of Mn = 76,500 g/mol from prepolymers of Mn =

6,700 g/mol. Prepolymers of Mn = 10,600 g/mol gave high polymer of Mn = 102,600

g/mol. The reverse reaction to reform the prepolymers, however, was apparently not

successful. High temperatures and prolonged heating times, leading to thermal

degradation, may have been to blame for the failure to induce reversibility.

Some success has been seen in crosslinking schemes. A reference by Craven in

196983 describes the synthesis of condensation polymers with pendant furan groups,

which are reversibly crosslinked by bismaleimides to yield a tough, rubbery film at 100C

and returned a "remoldable" polymer at 1400C. An example of this system is shown in

Figure 13. No catalysts were used. Properties of the materials were best when a "high"

amount of bismaleimide was used: upon crosslinking, the tensile strength is reported as

o o

n0 +
CH J2 0

SP100C 1400C

Figure 13. Crosslinking by the Diels Alder [4+2] addition reaction. This is one
example of several cited by the author. [Adapted from ref. 83]

6800 psi with 9% elongation and 120,000 initial modulus. Properties after remolding

were not reported.

Kennedy and Castner84 synthesized cyclopentadiene functional butyl rubber, as
shown in Figure 14. These polymers were observed to undergo crosslinking at room

temperature when the samples were dried. Due to the monomer dimer equilibrium of

cyclopentadiene, heating at 2150C in hexachlorobutadiene did not produce soluble
polymer. However, when maleic anhydride was added to this mixture, the polymer

became solubilized; IR studies confirmed that the cyclopentadiene groups were trapped
by maleic anhydride.

A similarly functionalized ethylene propylene rubber was also studied, where
cyclopentadiene groups were present on tertiary rather than secondary carbons (as was

the case for the isobutylene polymer). These polymers required temperatures of 1500C to
crosslink; above 1700C the reverse reaction with maleic anhydride was seen to occur.

230C, dry

Figure 14. Diels Alder / retro Diels Alder reactions in cyclopentadiene functional
polymer, the polymer backbone is polyisobutylene. [From ref. 84]

Recent work on polyphosphazenes functionalized by cyclopentadienylethoxy

groups has proven successful in forming thermoreversible crosslinked networks.85,86

Polyphosphazene, substituted with trifluoroethoxy- or ethoxycyclopentadienyl groups,

was studied by differential scanning calorimetry and changes in solubility. Similarly to

the work of Kennedy and Carlson, and Kennedy and Castner described above, it was

found that at room temperature and at 850-1200C, an equilibrium of crosslinked

cyclopentadiene functionalities existed; insoluble polymer network persisted. DSC

thermograms also showed this to be the case. However, heating in the presence of maleic

anhydride at 850C gave soluble linear polymer. Cycling of the systems was not


The most recent instance of thermoreversible Diels Alder crosslinking, shown in
Figure 15, involves the use of two polymers, one an oxazoline-functional chain, the other

a furan-functional chain.87 The two moieties undergo crosslinking after seven days at

room temperature, with the retro-Diels Alder reaction apparently occurring at 800C after

two hours. The technique for determining whether gelation involved covalent

crosslinking included varying the amount of crosslinkable moieties on the chains, and


N*'+ RT/7 days
800C, ~2hrs.


Figure 15. Diels Alder crosslinking by diene and dienophile functional polymers. Ratios
of n and m were varied to give differentiable swelling in H20. [Adapted from ref. 87]

determining the extent of swell in H20 for the resulting networks. Size exclusion

chromatography (SEC) was also used to show that the same polymers were reformed
after heating the gelled material. Cycling of the gel was observed with swelling properties
similar to the original crosslinking reaction. The networks are nonionic hydrogels, due to
their water swellable behavior.

Other systems that involve the Diels Alder reaction include the observed
thermoreversible gelation of poly(vinyl chloride), in which elimination of HCI has been
implicated as causing the formation of polyene structures which undergo

thermoreversible [4+2] addition.88 Other polyenes may undergo similar reactions. A
catalyzed 1:1 butadiene maleic anhydride copolymer is described by Japanese workers,

wherein a network was formed by heating and linear polymer reformed by cooling.89
Details of the system were not obtained, and it is not clear whether the crosslinking
reaction occurred via the Diels Alder reaction or some other process.

Nitroso Dimerization

The nitroso dimerization reaction has been a fairly successful method for the

synthesis of covalent linkages that are thermally reversible.91,92 Pazos found that nitroso


(R-NO) moieties dimerize at low temperatures and dissociate at temperatures generally

between 1100 and 1500C; thermoreversible linear polymers as well as crosslinked

networks were synthesized. Two examples of this chemistry are shown in Figure 16.

The linear polymers were initially synthesized in a stepwise condensation reaction by the

addition of a difunctional nucleophile (such as a diol or diamine) to a precoupled nitroso

dimer compound such as dimeric 4-nitroso-3,5-dichlorobenzoyl chloride; addition of heat

to the polyester or polyamide resulted in their dissociation to the decoupled nitroso

monomers. the crosslinking reactions, on the other hand, were brought about by the

addition of nitrosyl chloride to a linear polymer containing residual olefinic bonds, such

as styrene butadiene copolymer, crosslinking then occurred at room temperature.

In both of these uses, the dimerization / decoupling reaction was claimed to cycle
several times and useful materials were obtained. After each heat cycle, however, some

a. 2 HZ--C coo

CHz-CH-CH(CI)-CH3 cool

0 0
/ 1\ jI i cool
b. ON- \---NH(CH2pINH-- --NO ____

0 0 0

I- I n
--NH(CH2h2NI+--C--/ \-.- N---/ \-- -O-
Figure 16. Nitroso dimerization reaction to form a. Thermoreversible network, and b.
Thermoreversible linear polymer. [Adapted from ref. 91]

degradation of nitroso functionalities occurred so that after several cycles the material was

no longer useful (J. Pazos, personal conversation with K. B. Wagener). Physical

properties were not thoroughly investigated; some examples are given. For instance, a

terpolymer of ethylene/propylene/1,4-hexadiene (64/32/4 by weight) exhibited tensile

strength of 91 psi and elongation of 665% at break; after treatment with nitrosyl chloride,

the properties improved to 434 psi and 1,040% elongation at break. Properties for the

linear polymers were not reported, but the materials were generally liquids of low

viscosity at high temperatures and developed tough, sometimes flexible, nontacky

properties upon cooling. In some cases adhesive type materials were formed. In general

all these materials are easily thermoformable.

Ester Formation

The reaction of anhydride functional polymers to form ester crosslinks via

nucleophilic attack by polyols has been described.93,94 This ester formation is

schematically represented in Figure 17. One to forty percent of maleic anhydride or

related vinyl cyclic anhydride, copolymerized with another vinyl monomer, formed

thermally reversible crosslinks with several polyols, including di- and triethanolamine.

cool 0



Figure 17. Ester formation with maleic anhydride copolymer and a diol gives thermally
reversible crosslinks due to the tendency of the five membered anhydride ring to reform
upon heating. R is usually an aminoalkyl or alkyl group. [Adapted from ref. 93]

For example, a composition of 7.11% maleic anhydride, copolymerized with

styrene, was crosslinked with butanediol to give a material with mechanical strength of

500 psi and 200% elongation at break. The reverse reaction was affected by heating the

sample to 2600C for 30 minutes, as evidenced by solubility and flow characteristics. IR

studies on these materials indicated that the anhydride functionality is recovered at 250C,

and equilibrium parameters have been calculated for the reaction with dialkyl


Cycling (reversibility) of this system was not as successful. The sample noted

above was originally at 500psi and 200% elongation at break. After one heating and

cooling cycle, re-crosslinking occurred but the material was not as cohesive, breaking at

240psi and 250% elongation. Obviously, degradation of crosslink functionalities is

occurring, presumably at 2600C. Hydrolysis should not be a problem in this system, since

hydrolysed maleic anhydride ring closes to eject water above 1400C in the copolymer.

Other uses of the system include the crosslinking of a hydroxyl functional

polymer backbone by reacting it with pyromellitic anhydride. No claims are made

concerning thermoreversible chain extension chemistry of dianhydrides and diols.

Urethane Formation

Urethane formation has been studied as a means of thermally reversible chain

extension.95 In this work, a series of aromatic model urethanes were found to reverse to

the aromatic isocyanates and phenols in the pure state. The reversibility, evidenced by IR

studies, was dependent on both the aromatic nature of the monomers and also on the

melting point of the urethane adducts, since no reversibility was seen below these

temperatures for any of the adducts. The reverse reaction is dependent on the aromaticity

of the system, presumably due to the resonance stabilizing effect of the phenyl substituent

on the departing leaving groups. The reaction scheme is shown in Figure 18.


cool A


Figure 18. Thermally reversible urethane formation of aromatic diisocyanates and diols.
"PEG" is poly(ethylene glycol) of an undetermined length. This work has not been
published to date. [Adapted from ref. 95]

It should be pointed out that in this case, the polymers themselves have not been

shown to be thermoreversible; however, the model studies have shown that this should

be the case. Figure 18 denotes the envisioned scheme of chain extension /reversibility for

the aromatic polymers according to the model studies and also to the synthesis of

polymers described in the study.

Azlactones in Polymer Chemistry

"Azlactone" is the common name for derivatives of 5(4H)-oxazolone. Other

naming systems used for the heterocycle include 2-oxazolin-5-one, 1,3-oxazolone, and

4,5-dihydro-5-oxo-l,3-oxazole.3.5 Of these systems, 2-oxazolin-5-one is most

commonly used in polymer chemistry, aside from the common name of azlactone.

Azlactones were discovered in the 1880s by Plochl3 but the actual structure was

determined by Erlenmeyer;98 thus synthesis of the azlactone functionality via

cyclodehydration of a-amino acids is commonly referred to as the Erlenmeyer azlactone

synthesis.3 An incorrect assignment of the azlactone structure as that of penicillin5

resulted in considerable research into the chemistry of the azlactone ring; several review

articles are available which discuss both the synthesis of azlactones and their use as

synthetic intermediates in a variety of organic reactions.3'98-101

H Ri P1 H RF R2 N
1 1. NaOH 2 0X-N "C N-r
NH2 2. O 11 I R3 or some other cyclo- N R2
0 oC R3 0 H dehydration agent R
I RuR1

Figure 19. Common synthetic scheme for the formation of 2-oxazolin-5-ones, or
azlactones. R1-R3 may be alkyl or aryl. R1 and R2 may also be an alkylidene or spiro
functionality. [Adapted from ref. 3]

The most common method for synthesizing azlactones is to acylate an a-amino

acid at the amine site, followed by cyclodehydration by an appropriate agent (such as

acetic anhydride, dicyclohexylcarbodiimide, or ethyl chloroformate / triethylamine) to

give the azlactone. A typical reaction is shown in Figure 19. For example,

1-Methyl-2-halopyridinium salts102 and inorganic cyclizing agents such as phosphoryl

chloride / pyridine, thionyl chloride, phosphorus tribromide, phosphorus pentachloride,

and phosgene, have also been used to affect cyclodehydration, to name just a few.3

One reported synthesis, shown in Figure 20, uses simple ketones as the starting

material for a total synthesis of N-acryloyl-2-amino acids (which may then be cyclized to

2-alkenyl azlactones) in one flask.103

o NH2
j + NaCN ,-, R,--- ,,
R, R2 CN
R, 0 R RI 0 R3
CN-- NH-- -.CCH2 aq. HCHOOC-j-NH-C-C=CH2

Figure 20. Total synthesis of N-acryloyl amino acids from ketones. [Adapted from ref.


Key to the stability of the azlactone ring is disubstitution at the 4 site. By

eliminating all protons at this site, facile acid / base chemistry does not occur and the full

scope of reactivity of the ring opening reaction is realized. Deprotonation of azlactones

monosubstituted at the 4 site causes rearrangement of 2-alkenyl azlactones to

2-alkylidene-3-oxazolin-5-ones ("pseudoazlactones"),3 shown in Figure 21.

However, 2-monoalkyl azlactones may be easily functionalized at the 4 site by

facile deprotonation of the azlactone ring, followed by reaction with an electrophile such

as SO2C12. Reaction with a nucleophile completes the synthetic scheme. Alternatively, if

both substituents at the 4 site are H, deprotonation followed with the addition of an

electrophilic species is generally followed by dehydration to form an alkylidene

functionality.3 These reactions are shown in Figure 21.

R2 R2

No'ctH N=R
R1 R1

N -P-H N SR .
R yO RAR,5 R yo -Hp 0R O

Figure 21. Reactions of 2-alkenyl azlactones monosubstituted at the 4-site, 2-alkyl
azlactones monosubstituted at the 4-site, and 2-alkyl azlactones unsubstituted at the
4-site. [Adapted from ref. 3]

The lactone site of azlactones is sensitive to water, and undergoes slow reaction

with atmospheric water at room temperature to yield the open ring N-acyl amino

acid.5 Thus in order to quantify reactions or produce pure materials with other

nucleophiles, care must be taken to exclude adventitious water in the system until the

reaction goes to completion.

Reactions of Bisazlactones

Bisazlactones may be of the 2,2' variety or the 4,4' variety; of the two,

2,2'-bisazlactones (see Figures 22 and 23) have been used almost exclusively in polymer

chemistry. Synthesis of bisazlactones is analogous to that of monofunctional azlactones,

commonly involving acylation by a diacid chloride of two equivalents of an a-amino

acid, followed by cyclodehydration to form two azlactone rings. Several different

bisazlactones are available through this route.9,104-109 Their subsequent reactions are also

analogous to those reactions described above for the monofunctional azlactones.

Bisazlactones have been used in several types of reactions to yield linear polymers

as well as crosslinked networks. Their first such use, reported in 1943, was in stepwise

polymerization reactions (shown in Figure 22) with diamines and diols.104 The reaction

proceeds via nucleophilic ring opening of the lactone functionality to yield linear

polyamides and poly(ester amide)s. The stepwise reaction with primary diamines is

S< 0 0HR4N-RS- NH R6R R

01.R2 0 ORlRR4
RI, R2 = alkyl, aryl, or spiro HHO- R7-OH
R3 = alkyl, heteroalkyl, or aryl
OH H 0

O OR 0 Ri R2
Figure 22. Reactions of 2,2'-bisazlactones with secondary diamines or diols to give
stepwise addition. [Adapted from ref. 104]

facile, producing polyamides in solution with inherent viscosities of 0.5-1.81 dL/g.20

Their degree of crystallinity has been estimated by X-ray diffraction studies at


Secondary amines are less reactive, producing lower molecular weight polymers,
but do not undergo side reactions as described below. Diols are not as reactive as amines

and often require an acid or base catalyst to yield polyesteramides of appreciable
molecular weight. This is significant to the research at hand, in the examination of
bisphenol nucleophiles and bisazlactones for both their forward and reverse reactions.

Polymers derived from 2,2'-bisazlactones and primary diamines or polyamines
and bisazlactones can undergo further reaction after the initial linear polymerization.

Upon the addition of heat, two moles of H20 are lost in a cyclodehydration reaction to

give the corresponding poly(imidazolinone)s,5,110 as shown in Figure 23. Some of these
polymers undergo cyclodehydration with heat alone, while mild basic or Lewis acid
catalysts are required in other cases.

R1 R0 H H 0
SR R2 H2N- R4- NH2 2 R3 N R4
O 0 R R O R1 2 H H /n

R1, R2 = alkyl, aryl, or spiro
R3 = alkyl, heteroalkyl, or aryl

A -H20

Ri R1
Figure23, Formation of a poly(imidazolinone) via R3 R
application of heat to the linear polymer formed by 0 N O
reaction of a 2,2'-bisazlactone and a primary diamine.
[Adapted from ref. 5] I o

Polymerization of 4,4'-bisazlactones with aromatic primary diamines has also

been reportedly with good results: the polymers have inherent viscosities of 0.17 0.51

dLJg and yield is essentially quantitative. These materials are soluble in polar aprotic and

acidic solvents and are thermally stable up to about 300C. a-Amino alcohols have also

been used in a stepwise polymerization with 2,2'-bisazlactones.104 These polyesteramide

linkages contain natural a-amino acid groups and have proven to be biodegradable.

Most recently, 2,2'-bisazlactones have been reacted with difunctional o-phenylene

diamines as a route to benzimidazole containing polymers.112-114 Tetrafunctional

aromatic amines such 3,3',4,4'-tetraaminobiphenyl, among others, react with various

bisazlactones to yield low molecular weight (<10,000 g/mol) polymers with

benzimidazole functionalities, as shown in Figure 24. The degree of cyclization as

opposed to open ring amide formation is near 100% in some cases.

H2 H NH2 + R R/ 2


H2 NH2
0 H H O 0

RI R2 0 O R R2 H n


R2 O O R22 N H/

Figure 24. Addition polymerization of 3,3',4,4'-tetraaminobiphenyl with a
2,2'-bisazlactone, followed by cyclization to yield the corresponding benzimidazole
functionalities. [Adapted from ref. 112]

Bisazlactones may also be used in a similar manner as functionalities for the

crosslinking of nucleophilically functionalized polymers. Cellulose acetate, poly(vinyl

crosslinking of nucleophilically functionalized polymers. Cellulose acetate, poly(vinyl
acetal)s, and casein were reported to be crosslinked by 2,2'-bisazlactones, although the

resulting networks were not well characterized.104 Presumably, hydroxylic moieties

present along the polymer chain form esteramide linkages by ring opening of the

Bisazlactones can also be used to crosslink hydroxylated acrylic polymers and

other hydroxylated resins, as described in patented applications.115,116 The hydroxyl
functionalities in these cases required an acid catalyst to accomplish the nucleophilic ring
opening; "blocked" acid catalysts, which do not have activity below a certain temperature

when a protecting group is ejected. Up to the deblocking temperature of the acid

(1550C-1850C), good mixing is seen but essentially no reaction occurs.

Radical Addition Polymerization of 2-Alkenyl Azlactones

2-Alkenyl azlactones, which were first reported in 1959,117 may be polymerized

by a radical chain mechanism to yield poly(2-alkenyl azlactone)s, azlactone-

functionalized hydrocarbon backbone polymers. This polymerization scheme is shown in

Figure 25. Synthesis of 2-alkenyl azlactones is accomplished in the same manner as the

syntheses outlined above, and several different 2-vinyl and 2-isopropenyl azlactones are

AIBN, -600C

F\ R2 /bulk or in xylene
R,~ or EtOAc RR

Figure 25. Radical chain addition polymerization of 2-alkenyl azlactones. R3 = H or CH3
to obtain appreciable molecular weights. [Adapted from ref. 117]

available.10,118-129 The experimental conditions of the syntheses are generally milder

than for the 2-alkyl analogs since the acid chlorides used to make 2-alkenyl azlactones are


Effective radical addition polymerization is restricted to 2-vinyl azlactones and

2-isopropenyl azlactones, which form polyethylene and polypropylene backbones,

respectively.5 Common initiators such as AIBN (azobisisobutyronitrile) may be used in

this polymerization, for either the bulk or solution reaction. Moderate temperatures, about

600C, are employed.10,123 Properties of the poly(2-vinyl azlactone)s and

poly(2-isopropenyl azlactone)s are similar to those of poly(methyl methacrylate): they are

clear, glassy, and brittle. Glass transitions occur at 920C for poly(2-vinyl-4,4-dimethyl

azlactone), 1650C for poly(2-isopropenyl-4,4-dimethyl azlactone). The homopolymers

are soluble in several solvents but will often precipitate upon the gradual incorporation of

atmospheric water. Precipitation is attributed to the slow hydrolysis of the azlactone ring.

The radical copolymerization of 2-alkenyl azlactones is completely random with

many other vinyl monomers10,117-120,123-128,130 such as methyl methacrylate and

styrene. 2-Alkenyl azlactones are generally liquids and are miscible with several common

vinyl monomers, making them convenient to use in bulk polymerizations. High degrees

of polymerization are attainable because the azlactone functionality itself, when

difunctionalized at the 4 site, does not undergo chain transfer reactions with free radicals.

Thus molecular weights for both homopolymer and copolymers are highly controllable

through the use of chain transfer agents such as tetrahydrofuran.127,131

The azlactone functionality is a useful chemical "handle" for further modification

of poly(2-alkenyl azlactone)s. Through the nucleophilic ring opening reaction, the

azlactone functionality provides a site of reactivity conveniently anchored to the polymer

backbone. The reaction is shown in Figure 26. This use has resulted in several different

n n
+ HXG n

R --H

Fire 26. Modification of a poly(2-vinyl or 2-isopropenyl azlactone), where R1 R3 are
as in Figure 22; X = O, S, or N; G = some group for the modification of the polymer
properties. [Adapted from ref. 99]

applications in polymer modification. Similar utility has been found for oxirane,

anhydride, and isocyanato functionalized polymers.131

The first application of this chemistry was the addition of n-butylamine5,128,130

as the species HXG (X=N, G=n-butyl) to give amide functionality from azlactone. Other

functional groups G, placed on protic nucleophiles HX-, serve to tailor properties by

allowing almost any desired group to be attached to the polymer. Hydrophobic -

hydrophilic properties can be manipulated.132

Reaction with HXG where G contains a vinyl group can be subsequently

crosslinked by irradiation or electron beam bombardment, as shown in Figure 27.

Selective crosslinking through masking techniques allow patterning of the material, as the

uncrosslinked portion is easily washed away. These reactions have found utility in

negative photoresist chemistry and have optical data storage applications as embossable

video disk recording materials.127,133

A protic polyfunctional nucleophile, [HX]nG, can be used to crosslink an

azlactone functionalized polymer via ring opening at the lactone site of the azlactone, as

shown in Figure 28.120 By incorporating a 2-alkenyl azlactone in 0.5-30 weight percent

via radical copolymerization with acrylic or methacrylic monomers, the azlactone moiety

provides a crosslinking functionality for polyfunctional nucleophiles.118-120,134-136

+ VHXvC CH2 RT/ 3-24hr

(usually as a copolymer)



S1. add photoiniiator (5-10 mol % of azactone functionality)
Negative photoresist 2. mask surface with material impervious o radiation
chemistry 3. photolysis
4. unmask surface
5. wash away uncrosslinked polymer

Figure 27. Negative photoresist scheme utilizing poly(2-vinyl or 2-isopropenyl
azlactone). R1 R3 are as in Figure 22; X = O, N, or S. X is attached to some moiety
with a vinyl group for crosslinking. [Adapted from ref. 5]

Again, when amines are used the crosslinking reactions are facile, but hydroxyl

nucleophiles require a strong acid catalyst in order to affect crosslinking.

Self-crosslinking compositions have been synthesized by copolymerizing 2-alkenyl

azlactones with hydroxylated vinyl monomers.119,137

+ HX--R-XH

Figure 28. Example of covalent crosslinking of a poly(2-vinyl or 2-isopropenyl
azlactone) by a difunctional nucleophile. R is alkyl or aryl. [Adapted from ref. 131]

Other Uses

A related application of the chemistry described above is to modify 2-alkenyl
azlactones with HXG before polymerization. Several new vinyl addition monomers have

been synthesized via this route. Recently, monomers of this type have been described

which contain groups susceptible to p-elimination; these have been used in diffusive

transfer photographic film technologies.5,138 This chemistry is shown in Figure 29.

I n In
c==o c==
H30--C--CHa3 1rO-C-CH3

Figure 29. Poly(2-vinyl-4,4-dimethyl azlactone), ring opened with a group capable of
undergoing 3-elimination. This process converts the polymer from alkali impermeable to
alkali permeable for film processing technologies. [Adapted from ref. 5]

Amine or hydroxyl functionalized compounds, including oligomeric or

prepolymeric type molecules, can be reacted with 2-alkenyl azlactones in a nucleophilic

ring opening of the azlactone, yielding a modified pendant acrylamide functionality that

contains the intact vinyl or isopropenyl group. This is shown in Figure 30. The monomer

o 0--c o==c
+ R I I
\+ HN-- N--H N-H
R1 R- R2 R -
0== 0==
Figure 30. Prefunctionalization of 2-alkenyl azlactones to give new vinyl monomers for
radical polymerization. [Adapted from ref. 139]

can then be reacted via free radical techniques.139 Polyfunctional nucleophiles give the

corresponding polyfunctional acrylamide telechelomers. 139,140
The 2-alkenyl azlactones contain a Michael addition site which is susceptible to
nucleophilic attack. So-called "hard" nucleophiles, such as primary amines, react nearly
exclusively at the "hard", or most polarized, electrophilic site. In the case of 2-alkenyl
azlactones, this is the lactone functionality. On the other hand, "soft" nucleophiles such
as mercaptans (either alone or in the presence of an acid catayst) will react preferentially at
the "softer" Michael addition site. This is shown in Figure 31. The latter application has
been useful for the preparation of so-called "multiazlactones," prepolymers which are

useful for further chain extension reaction, from 2-alkenyl azlactones.141,142

RF R3 Ra

2 HS--R--SH 0 S-R4-SR2
R R N N- R2
R1 R1 R,

Figur 31. Michael addition of a difunctional mercaptan to a 2-alkenyl azlactone to form a
"multiazlactone", a variety of bisazlactone. [Adapted from ref. 142]

Secondary amine nucleophiles give a mixture of products which apparently
depends mainly on the steric bulk of the system: the Michael addition site is sterically
more accessible in 2-vinyl azlactones, particularly when a bulky nucleophile is used
and/or bulky substituents are placed at the 4 site of the azlactone. Mercaptans also exhibit
plurality of reaction: when the reaction is performed in the presence of base catalyst of
sufficient strength to form mercaptide anion, nucleophilic addition is observed instead at

the lactone site.127
Two other uses of the azlactone functionality in polymer chemistry are the direct
cationic ring opening polymerization of the azlactone ring, and 2,3-dipolar additions with
electron deficient olefins. In the first application, cationic initiation of 2-alkyl azlactones
monosubstituted or unsubstituted at the 4 site proceeds to form homopolymers via ring

opening, presumably by zwitterionic intermediate. 143,144 Molecular weights up to about
5000 are obtained when methyl triflate is used as the initiator. The polymers, as shown in
Figure 32, are hygroscopic.

0 H F3C-S-OMe
I1 in diglyme 0
R RT/36-72 hrs. k
(R=H or CH3) H R

Figure 32. Cationic polymerization of azlactones. The polymerization does not occur in
the presence of bases or Lewis acids. C4 optical activity, if present, is retained in the
polymerization. [Adapted from ref. 143]

The dipolar nature of the azlactone ring has resulted in its use in 1,3-dipolar

cycloaddition polymerization with vinyl groups. When monosubstituted at the 4 site, a
tautomer equilibrium exists and the molecule can undergo reactions with electron deficient

olefins.101,145 When a monoazlactone is reacted with a difunctional olefin such as a

bismaleic anhydride, as shown in Figure 33, a polymer is formed which possesses high
thermal stability in air.

It can be seen in the literature that azlactones are widely utilized in polymer

chemistry. Their uses for the synthesis of both stepwise addition polymers and network

polymers are known. However, this is the first time that it has been proposed to use

azlactones to form thermoreversible covalent polymers and networks. If the research is

considered successful, then the azlactone phenol system would be one of few known
reactions used for such a purpose.
The merit of studying such systems lies in the fact that "thermoplastic thermosets"

and thermoreversible polymers which regenerate monomers are of high potential utility in

R -

0 0

o 0


Figure 33. 1,3-dipolar cycloaddition reaction of azlactones, monosubstituted at the 4 site,
with bismaleimides. The tautomer which allows the reaction is present only for the
4-monosubstituted azlactone. [Adapted from ref. 145]

the polymer industry. Systems of this nature are recyclable and return small molecules for

use in regenerating the polymers or networks, or other purposes.

Although none of the systems developed in the past have been used for industrial

production, the development of plastics recycling technologies is important for the future,

as solid waste problems continue to grow. It is important for these technologies to be in

place when their use becomes feasible, perhaps in terms of industry regulations on the

use and manufacture of plastics that must be recyclable or degradable.


General Information

NMR spectra (1H and 13C) were recorded on a Varian Model XL-200 200 MHz

Fourier transform instrument. Solvents were as indicated; temperature regulation, where

necessary, was to 0.1 C in the probe. All shifts are reported in ppm (5) downfield from

tetramethylsilane (TMS). Infrared spectra of polymers were run on a Perkin Elmer Model

281 spectrophotometer by casting a film from CH2C12 onto quartz windows. UV spectra

were obtained on a Perkin Elmer Lambda 9 UV/VIS/NIR spectrometer using quartz

cuvettes. Solvent absorbance was subtracted with a blank cuvette.

Size exclusion chromatography (SEC) was performed on a Waters Model 6000A

Liquid Chromatograph equipped with differential refractometer, Perkin Elmer Model LC

- 75 UV detector at the specified wavelength, and SEC columns as indicated. Flow rate

was 1.0 ml/min with THF as the mobile phase. Calibrations and molecular weight

determinations were performed using a Zenith Data Systems PC.

HPLC traces were obtained on a Waters Model 590 instrument equipped with a

Kratos Spectroflow 757 UV detector set at the specified wavelength, Perkin Elmer LC-25

refractive index detector, and a Waters Nova-Pak C18 reverse phase column. The mobile

phase was 60% aqueous CH3CN, at a flow rate of 0.5 ml/min. Peak areas were

determined by cutting and weighing. All SEC and HPLC samples were prefiltered with

0.45mn Millipore filters.

Differential scanning calorimetry (DSC) was performed on a Perkin Elmer DSC 7

instrument equipped with N2 purge, TAC7 data analysis package and PE 7500 computer,

indium and lead were used as standards. Thermogravimetric analysis (TGA) was

obtained using a Perkin Elmer TGA 7 with N2 purge; the instrument was equipped with

TAC7 data analysis and PE 7500 computer. Alumel and Nicoseal (Perkin Elmer) were

used as standards for curie point magnetic transitions.

Service mass spectrometry was performed using a Finnegan MAT 4500

quadrupole mass spectrometer with methane chemical ionization detection. Elemental

analyses were performed by Atlantic Microlabs of Georgia.

Viscometry experiments were carried out in THF with an Ace viscometer,

Ubbelohde type, size 0 capillary. Temperature was regulated by a Haake E52 water bath.

Hexanes, THF, diethyl ether, toluene, and acetonitrile used for reactions and

purifications were distilled under an argon atmosphere immediately prior to use. Toluene

was distilled from a Na K alloy; acetonitrile was distilled from calcium hydride.

Hexane, THF, and diethyl ether were distilled from Na K alloy with benzophenone

indicator. DMF and DMSO were fractionated in vacuo using standard Schlenk

techniques. Other solvents used for reactions were obtained in >99% purity and used

without further purification, unless otherwise specified. Deuterated solvents were 299

atom % D unless otherwise specified; for some quantitative 1H NMR, 100 atom % D

DMSO was used. All solvents (standard and deuterated) were stored in the dry box or in

a dessicator.

Except for syntheses, most reactions were performed in a Vacuum Atmospheres

Dri Lab model HE-43-4 dry box under Argon atmosphere and equipped with a model

HE-63-P pedatrol pressure regulator and large capacity vacuum pump. The atmosphere in

the dry box tested dry to diethyl zinc and titanium tetrachloride and so was maintained at

<10 ppm H20 and 02. Liquid reagents and solvents were taken into the dry box by

sealing the vessels under vacuum before applying vacuum in the antechamber, or
alternatively by securing the unopened bottles with electrical tape before applying
Reactions not performed in the drybox were carried out in sealed vessels or on a
standard Schlenk line equipped with argon and vacuum serviced by high pressure

stopcocks. The vacuum pressure was typically 3x104 mm Hg.

Synthesis. Purification. and Characterization of Model Compounds

2-Isopropyl-4.4-dimethvl-2-oxazolin-5-one (1)
Model azlactone I was synthesized in two steps with modifications of literature
methodology. In the first step, isobutyryl chloride was used to acylate 2-methyl alanine

with modifications to the method outlined by Iwakura, et al.121 Modifications to this
procedure were necessary. The reaction is illustrated in Figure 34.

0 0
21. NaOH, leq. 0 0
2 2. o ,3 eq. Y hexane/ reflux
0 C 0 H

Figure 34. Synthetic scheme used to make 1, 2-isopropyl-4,4-dimethyl-2-oxazolin-5-one.

To 28.0 ml concentrated aqueous NaOH solution (0.40g NaOH added per ml

H20) was added 10.3g (0.10 mol) of 2-methyl alanine (2-aminoisobutyric acid) at room

temperature; the mixture was stirred until all of the 2-methyl alanine was dissolved. This
solution was then transferred to a 250ml, three neck roundbottom flask equipped with
mechanical stirrer, thermometer, and pressure equalizing addition funnel.

Three equivalents of isobutyryl chloride were added to the solution via addition
funnel; violent mechanical stirring plus efficient cooling via dry ice/isopropanol bath was

required to maximize yield and prevent overheating of the heterogeneous mixture formed

during the reaction. Optimum yields were realized by adding the acid chloride as quickly

as possible, maintaining a temperature of 300C + 100C for the duration of the addition.

After the addition was complete, the reaction mixture was stirred vigorously for an

additional hour. Overheating was sometimes observed during the first 15 minutes after

the completion of the isobutyryl chloride addition; this was controlled by the cold bath.

Acidification of the reaction mixture was accomplished by addition of 7.1ml

concentrated HCI, or sufficient HCI to give a pH of <2. Stirring was continued for

another fifteen minutes after the addition.

Deionized water was then added slowly with stirring, enough to dissolve all

solids and form two layers in the reaction flask. Slow magnetic stirring of the layers for

12-24 hours resulted in all of the product being located in the top layer. The layers were

separated and the top layer added directly to boiling toluene; heating was continued until

all the material was dissolved. At this point any excess H20 and H20 soluble impurities

could be removed by pipetting from the hot toluene solution. Slow cooling gave efficient

purification of the resulting crystals. A second dissolving in boiling toluene gave 20%

yield of pure N-isobutyryl-2-methyl alanine by melting point (1500-1520C), 1H NMR

(D20: 81.00, d, 6H; 81.35, s, 6H; 82.20, s, 1H; 82.39, sept., 1H; 87.90, s, 1H), and

elemental analysis (Calc. [Found] C 55.5 [55.4], H 8.7 [8.7], O 27.7 [27.7], N 8.1


Cyclodehydration of the N-isobutyryl-2-methyl alanine was completed via

reaction with 1.1 equivalent dicyclohexylcarbodiimide (DCC) in hexane. Fifty ml hexane

per gram N-isobutyryl-2-methyl alanine was sufficient to allow magnetic stirring of the

heterogeneous mixture. In a typical reaction, 250ml hexane, 5.00g (2.89x10-2 mol)

N-isobutyryl-2-methyl alanine, and 6.55g (3.18x10-2 mol) DCC were added to a 500ml
roundbottom flask equipped with a condenser and argon blanket. The mixture was
refluxed gently for two hours.
After cooling, the azlactone / hexane solution was filtered away from dicyclohexyl
urea. Hexane was then removed by rotary evaporation and the product, a liquid,
fractionated in vacuo using standard Schlenk techniques. Immediate transfer to the dry
box prevented hydrolysis of the azlactone ring. Yield was typically 75-80% of crude
product for the cyclization step.

Characterization was accomplished by 1H NMR (CDC13: 81.17ppm, d, 6H;

81.32, s, 6H; 82.74, sept., 1H), 13C NMR (CDCl3: 818.6ppm, CH3; 824.5, CH3;

829.0, CH; 865.1, C; 8167.8, C; 8181.6, C) and by elemental analysis (Calc. [Found] C
61.94 [61.95], H 8.39 [8.44], O 20.65 [20.60], N 9.03 [8.99]).

Substituted Phenol Compounds (IIa-e)
Phenol compounds IIa-e (shown in Figure 33) were all >99% pure as ordered
from Aldrich. They were used without further purification in the model studies. In
preparation for use, all compounds were either vacuum dried and taken into the dry box,
or taken into the dry box before opening the reagent bottle.

Adducts of I and la-e (IHIa-e)

Q + hexane /reflux X=CF3
Ilua aei X=F

Figure 35. Synthetic scheme for compounds HIla-e.

The synthesis of IlI is outlined in Figure 35. I was used without purification in
the hexane solution from its synthesis (see above) for the synthesis of adducts 11i and g;

distillation was not necessary. Assuming 75% yield from the cyclization of L 3.36g

(2.17x10-2 mol) is present in the hexane solution. Addition of an equimolar amount of

the desired phenol IIa-e plus 0.167g (1.10x10-3 mol) 1,8-diazabicyclo-
[5.4.0]undec-7-ene ("DBU") was followed by reflux for 2-3 hours under argon

atmosphere. Purification was by rotary evaporation of hexane, followed by two to three

recrystallizations from EtOH/H20. The products were dried in a vacuum oven for 12

hours at 200C 500C, depending on the melting point of the adduct. Although all adducts

HI were stable to air, they were stored in the dry box to prevent adsorption of water.

The synthesis of adducts lTa.d and e required rotary evaporation of hexane from

I before the addition of Ia.d and e, respectively. The reactions were run neat, at 800C for
24 hours and then at room temperature for another 24 hours in order to maximize yields.

In addition, Ille was synthesized with a 15% excess of I to avoid the

homopolymerization of lie with the concomitant release of HF. IIIa.d and e required

four to five recrystallizations from EtOH / H20, which caused large apparent losses of


Yield of the purified adducts ranged from 75% for IIlb and c to 5% for lila.

Characterizations were via 1H NMR and elemental analysis as shown in Tables 1 and 2.

Table 1. Elemental analysis, melting point and yield data for adducts IIIa-e.

Compound Atom % Calc. % Found m. OC %Yield (pure)
Ilia C 57.14 57.02 105 5
H 6.16 6.21
0 27.18 27.22
N 9.52 9.51

Table 1. Continued.





Atom % Calc. % Found









Table 2. 1H NMR chemical shifts for compounds IIIa-e.

Compound Site



Chemical Shift, ppm



Multiple Integration



Table 2. Continued.

Chemical Shift. ppm




Multiplet Integrtion







Synthesis. Purification. and Characterization of Compounds
for Polymerizations and Crosslinking

2.2'-Tetramethylenebis(4.4-dimethvl-2-oxazolin-5-one). IV


Figure 36. Bisazlactone 2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one), I.
used in step polymerizations with bisphenols.

Compound IV, 2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one), shown in

Figure 36, was used in step polymerizations with bisphenols. The synthesis is found in a

Schotten Baumann type procedure outlined in the literature9 and is analogous to the




preparation of I. Compound IV was recrystallized from toluene instead of benzene in a

change from the literature procedure. The amino acid acylation step for IV. where

2-methyl alanine (2 equivalents) was added to adipoyl chloride, was found to have a yield

far lower than that claimed by the authors. Thus the open ring analog of IX was obtained

in about 15% yield (49% claimed) for the amino acid acylation step. Cyclization

proceeded as claimed, in 80% yield. Characterization was via 1H NMR, melting point,

and elemental analysis.

Poly(2-vinyl-4.4-dimethyl-2-oxazolin-5-one). V




Figure 37. Poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one), Y, to be used in crosslinking
studies with bisphenols.

Poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one), V, shown in Figure 37, was

obtained by literature methods.10 Bulk polymerization of freshly distilled

2-vinyl-4,4-dimethyl- 2-oxazolin-5-one at 500 20C with AIBN resulted in a bimodal

SEC trace. Using TSK G3000 and TSK G5000 SEC columns (THF, 1.0ml/min), the

molecular weights of the two peaks were determined by polystyrene standards of M,,

or weight average molecular weight, values of 17,500 to 800,000 g/mol. Refractive

index detection was used. Characterization of azlactone functionality was via IR and

confirmed by known shifts as described in the literature.10 The polymer was stored in the

dry box to prevent hydrolysis of azlactone moieties.

Lower molecular weight polymers were achieved using THF as a transfer agent

as described in the literature.127 In this case, one peak was seen in the SEC trace, without

evidence of a bimodal distribution. Using 20g of freshly distilled monomer, 80g freshly

distilled THF, and 1.0g AIBN, a polymer of Mw = 75,000, Mn = 36,000 g/mol was


Bisphenols VIa-VId

via 1-4- fVic fjI-
R )- a OH R: o o o

V "b -f-i- Vid -S-(CF2)4-S+
0 0 0
Figure 38. Bisphenols VI for step polymerizations and crosslinking reactions.

Bisphenols Vla-d were used for step polymerizations with bisazlactone IV as well
as for crosslinking of poly(2-vinyl azlactone), Y. The four bisphenols YI are shown in

Figure 38. Bisphenols Via (4,4'-isopropylidene diphenol or Bisphenol A), YVI

(4,4'-sulfonyl diphenol or Bisphenol S), and Vie (4,4'-hexafluoroisopropylidene

diphenol or Bisphenol F), were used as obtained from Aldrich. They were brought into

the dry box after applying vacuum overnight to remove adventitious water.

Bisphenol VId, 4,4'-(1,4-disulfonylperfluorobutane)diphenol, was not
commercially available, and was synthesized using an original procedure. The precursor,

the corresponding fluorophenyl compound, was obtained from E. I. DuPont De
Nemours and Company [Dr. A. Feiring at DuPont is gratefully acknowledged for

supplying the starting material and basic scheme for synthesis of YId].
The synthetic scheme, shown in Figure 39 and described below, is a modified

procedure as obtained from DuPont; the monofunctional phenol had previously been

made. Therefore, the transformations were accomplished using the same synthetic

strategy with variations as necessary. In fact, the procedure was actually simplified as

compared to the original synthesis as outlined by DuPont.

\- / S-(CF2)4S F
o o

1. MeONa / MeOH / reflux
2. ppt. with H20; CH2CI2 recryst.

o o
MO-O / S-(CF-)4- SOM
o o

1. BBr3 / CH2Cl2 / reflux
2.pH 11, then 2
3. MeOH/H-2 recryst.

o o
HO-( -S-(CF2)4- OH V
0 0
Figure 39. Synthetic scheme for the transformation of 1,4-bis(4-fluorophenylsulfonyl)-
perfluorobutane to 4,4'-(1,4-disulfonylperfluorobutane)diphenol, VId.

The starting material was converted to the anisole derivative by adding 8.35g

(0.016 mol) of the fluorophenyl compound to a solution of 80 ml MeOH, freshly

distilled from CaH2 under argon, and 1.35g (0.059 mol) freshly cut Na metal which had

been stirred under argon overnight. This mixture was refluxed in an oil bath at 900-950C

for 2-3 hours. After cooling, sufficient H20 (approximately 70ml) was added to the

mixture to cause precipitation of the anisole. The precipitate was filtered and sufficient

CH2CI2 added to dissolve the solid with warming. Refrigeration (-320C) of the solution

caused precipitation. A second recrystallization followed by drying in vacuo overnight

gave pure white to colorless needles (7.19g, 83% yield after two recrystallizations) that

were characterized by melting point (1610-163C), elemental analysis (Calc. [Found]: C

39.85 [39.69], H 2.58 [2.54], S 11.81 [11.73]), and 1H NMR (CDC13: 87.9, m, 4H;

87.1, m, 4H; 83.9, s, 6H).
In the second step, a 100ml three neck roundbottom flask, equipped with

mechanical stirrer, condenser column with Argon blanket, and septum, was purged with

argon and flame dried before adding 20ml 1.OM BBr3 solution in CH2Cl2 via syringe.

The solution was cooled in a dry ice/acetone bath. A solution of 3.5g (6.5x10-2 mol) of

the anisole in CH2CI2 was formed by heating and adding solvent as necessary. The warm

solution was then drawn up into a syringe and slowly injected into the cold BBr3

solution. After 5-10 minutes with good stirring, the cooling bath was removed and the
heterogeneous mixture allowed to warm to room temperature. A yellow solution formed

as warming took place.
The mechanical stirring apparatus was quickly removed at this point and a clean,

dry stir bar added, along with a small flash distillation apparatus. Approximately 75% of

the solvent CH2C12 was flashed from the mixture by heating the flask under argon

atmosphere in an oil bath regulated to 750-800C. The remaining material was then refluxed

-15 hours. After cooling the mixture, which had turned orange to brown, 10-20ml of

water was added to the flask slowly with manual stirring. Concentrated NaOH solution

was used to bring the pH to 211, then concentrated HCI added to acidify to pH <2.

Additional water was added as necessary to maintain stirring.
The mixture was filtered (coarse grade fritted glass filter) and the insoluble
material added to reagent grade MeOH. Material insoluble in MeOH was filtered off and
water was added slowly to the resulting solution until precipitation was observed. After

standing 1-2 hours at room temperature, VId was collected and recrystallized two more

times from MeOH/H20 before drying overnight in a vacuum oven at 500C.

After two recrystallizations, VId, an off-white powder, was obtained at 86% yield

for the second step. Purity was determined by melting point (2040C), 1H NMR (CD3OD:

87.8, m, 4H; 87.0, m, 4H [no phenolic protons observed]), and elemental analysis (Calc.

[Found]: C 37.35 [37.42], H 1.95 [1.96], S 12.45 [12.35]). A third recrystallization
after elemental analysis resulted in higher purity with a small loss of yield.

Model Studies

Model studies on compounds I, IIa-d, and IIIa-e can be divided into two parts:
the forward reaction of I and IIa-c, which was studied by extent of reaction at 24 hours,
by acid catalysis, and dilute and concentrated solution kinetics; and the reverse reaction
from MIa- to I and Ila-e, which was studied by TGA and DSC (described above), extent
of reaction at 24 hours, and by HPLC analysis of the products of volatilization during a

constant temperature ramp. In a separate set of experiments, equilibration reactions of

bisphenols YVb-d with two equivalents of I were monitored via HPLC.

The Forward Reaction of I and ITa-c

Observation of the extent of reaction of I at various times for the three phenol
species Ila-c was accomplished by making solutions 0.10M0.01M or 0.51M0.03M

equimolar in various solvents. A typical solution was made in the dry box in the
following manner.

Reagent Amount
I 0.166g (1.07x10-3 mol)
IIa 0.148g (1.06x10-3 mol)
CDC13 to 2.1 ml

The solutions were then examined by 1H NMR. For non-deuterated solvents, the

solutions were doped with deuterated solvent; for instance, a few drops of d6-benzene

were added to chlorobenzene solutions.

The solution was then transferred to a vial equipped with a magnetic stir bar and

stirred for a measured amount of time. Temperature was maintained in the dry box to

1oC over several weeks and temperature fluctuation over 24 hours appeared to be less

than 0.5C. Where elevated temperatures were used, an oil bath regulated to 1oC was

used inside the dry box. After reaction inside the dry box, the solution was transferred to

an NMR tube. Tetramethylsilane was added before capping the tube and removing it from

the dry box; the NMR spectra were run within 30 minutes of removal.

The ratio of the integration of protons due to I and III was used to measure the

reaction, as shown in Figure 40. In this manner, the relative concentrations of the two

species could be quantified. In some experiments, acetonitrile was added directly to the

reaction solution or added to the NMR tube in a capillary; the shift of the single

absorbance due to CH3CN was close to that of the protons measured, so the

disappearance of I alone could be measured. This required the measurement of the initial

ratio of I to acetonitrile. Within experimental error, it was found that the two methods of

measurement were equally good. The internal standard method is illustrated in Figure 41.

All the 1H NMR experiments described for the model studies were analyzed using

the techniques shown in Figure 40 or 41.

The shift of the measured septet of ring opened adducts II does not differ much
from that of the ring opened hydrolysis product of L so phenyl protons were also

examined to make sure that the shift of these protons was concomitant with the observed

product formation. In other words, if product III was apparently forming without the

expected shift in phenyl protons, then hydrolysis was interfering in the reaction.

Ha (82.75ppm)

Figure 40. H NMR spectra of the 0.51M reaction of I and Ha after 10 days at 27010C.
Proton Ha appears at 2.75ppm, proton Ha' at 2.46ppm.


3.0 PPM 2.4 8


2.9 PPN 2.3


Figure 41. Reaction of I and IIa. Top: t = 0; bottom: after 18 hours reaction time in
DMSO, in the presence of CH3CN as internal standard.

The solvent d6-DMSO, which was extensively used in these experiments,

displays protic impurities at about 82.5ppm. This interferes with the septet measured for

several of the species III. In these cases, the absorbance of the impurity was subtracted

from the absorbance due to III by obtaining a t=0 spectrum. Alternatively, the internal

standard method eliminated the need to measure the signal due to III.

In an analogous set of experiments to those described above, acid catalysts such

as anhydrous trifluoroacetic acid, fuming sulfuric acid, CD3COOD, and

AICl3(anhydrous) were added to 0.51M equimolar solutions of I and IIa-c which were

mixed as above. CDC13 was used as solvent in these experiments since side reactions of

the catalysts with d6-DMSO might have interfered with the reaction of I and II. In all

cases, the desired number of equivalents was weighed out, with the catalysts being the

last substituent added to the mixtures, and solvent was then added to the correct volume.

In the case of "catalytic amounts," one drop of trifluoroacetic acid was added from a

disposable nine inch Pasteur pipette; for the solid AIC13, a few crystals were added via

spatula such that the weight corresponded to less than 0.1 equivalent.

Another set of experiments were run, where long term kinetics of the 0.10M and

0.51M equimolar solutions of I and Ila-c in d6-DMSO (uncatalyzed) were measured.

These reactions were measured over a period of one week to three months. In these

experiments, larger amounts of the same solutions were made up and aliquots removed

periodically from the dry box in NMR tubes for analysis. Zero point spectra were
required in these experiments.

Kinetic runs with 1.OM equimolar I and Hib, and 0.10 and 0.51M equimolar I and

IIa-c in d6-DMSO were also examined during the initial stage of the reaction, which is the

first 15 hours or less. In these experiments, a single sample (mixed as described above)

was removed from the dry box in an NMR tube at t = 0 and immediately placed in the

NMR probe. The probe was equilibrated at the desired temperature for 20 minutes prior

to mixing the reagents. Analyses were set up on the spectrometer such that a set number

of transients were accumulated as separate experiments for an array of time delays. A

typical experiment involved accumulations of 48 or 64 transients for each spectrum, with
intervals between spectra of 0, 300, 300, 600, 600, 1200, 2400, (3600 x 4) and (7200 x

4) seconds. Analysis of each spectrum was then carried out as shown in Figure 40 or 41.

Pseudo first order kinetics146 involved reactions with non-equimolar amounts of

the reagents I and IIa at 27.00C in d6-DMSO; in these cases, Ha was added at ten times

the number of equivalents of I. These reactions were performed for 0.10M solutions of I;

0.51M solutions of I with 3.68M IIa were used for concentrated solutions, as 5.1M IIa

was not attainable in d6-DMSO.

Finally, a set of experiments using IIa and its sodium salt were undertaken.
Solutions of 0.51M I with a total equimolar amount of fIIa + sodium salt of Ia] were

mixed, and the ratio of IIa and the salt were varied. All experiments in this case were run

in DMSO solution without acetonitrile internal standard.

The Reverse Reaction of I and II: Reactions of IlI

Synthesis of the purified adducts of I and II allowed examination of the reverse
reaction. Adducts III, whose synthesis is described above, were used to determine the

temperature at which the reverse reaction to give I and II is observed to occur. No further

mechanistic work was attempted with these adducts.

In the first set of experiments, 0.10M d6-DMSO solutions of adducts IIIa- were

mixed in the dry box as shown below for a typical experiment. In some cases,

trifluoroacetic acid or AIC13 were added in catalytic amounts to analogous solutions.

Solutions 0.51M of IIa-c in d6-DMSO were also used. A typical solution was mixed in

the following manner.

Reagent Amount
Mlia 0.030g (1.02x10-4 mol)
d6-DMSO to 1.0ml

These solutions were then placed in heavy walled glass vessels as shown in

Figure 42. A 12" syringe needle (18g) was used to draw up the solutions and place them

in the vessels. The closed vessels were removed from the dry box and immediately sealed

at the constriction. Sealing was accomplished on a standard Schlenk line with an

oxygen/acetylene torch, after three freeze pump thaw cycles. After allowing time to

warm to room temperature, the sealed vessels were placed in a sand bath regulated at the

desired temperature +5C.

14/20 ground glass joint

--- ground glass stopcock

> heavy walled glass

Figure 42. Apparatus used for sealing solutions of III. The stopcock is used for
removing the solutions from the dry box under argon atmosphere; the apparatus is then
attached to a Schlenk line for sealing via the glass joint.

The sand bath consisted of a ceramic heating mantle filled with washed, ignited

sand attached to a variable voltage transformer in order to regulate temperature. A

thermometer bulb was placed at the depth of the glass vessels by marking both the

thermometer and the glass vessels at the same height. The vessels were placed in the

sand bath for 24 hours30 minutes; at that time they were removed from the bath and

quenched in ice water. The tubes were then opened and 1H NMR samples removed and

run immediately. Samples were staggered in the time they were initially exposed to heat,

so that there would be minimum delay between quenching and spectrum accumulation. In

this manner, solutions of IIa-c were run at 1500C, 2000C, 2600C and 3000C.Adducts

Ia-e were also examined for reversibility via DSC and TGA. In the DSC experiments,

the temperature ramp used was 20C/min from 400C to 5000C. N2 sample purge was used

for the neat adducts MIl removed from the dry box in capped vials and opened in the DSC

sample chamber, which was also purged with N2. The traces were examined for

exothermic processes.

TGA runs were performed with N2 sample purge at a temperature ramp of

200C/min from 1000C to 5000C. Since the adducts were exposed to air during the sample

loading process, they were preequilibrated for one hour in the furnace at 700-1000C in

order to remove adsorbed H20.

In a final set of experiments, neat adducts III were placed in a microdistillation

setup inside the dry box. Approximately 0.1g was used in each experiment. The setup is

shown in Figure 43. The apparatus was then removed from the dry box and immediately

attached to a Schenk line, and exposed to vacuum for about ten minutes at ambient

temperature. With the apparatus still under vacuum, the catch flask was immersed in a

dry ice / isopropanol bath. The flask containing the neat adduct was then placed in a sand

bath which had been heated to 500C. A thermometer bulb was placed at the height to

where the reaction flask was immersed. Glass wool was placed between the reaction

flask and the catch flask to maintain the temperature difference. A temperature ramp of

about 5C/min was then applied to the reaction flask with the sand bath while

maintaining a vacuum on the apparatus, and the reaction flask was carefully observed

for disappearance of the neat adduct

10/20joint -
S to vacuum

reaction flask .-
sand bath catch flask dry
ice / isopropanol

Figure 43. Microdistillation apparatus used for the heating of neat adducts Ill in vacuo.

After noting the temperature range where the adduct had disappeared, the entire

apparatus was disassembled after cooling. The reaction flask, distillation head, and catch

flask were washed separately and quantitatively with acetonitrile. The contents of the

wash were then analyzed by HPLC with UV detection at X=215nm. At this wavelength, I

absorbs weakly and II and III absorb strongly. Coinjections of I, II. III, and the

hydrolysis product of I were used to identify the products of the distillation.

Reactions of and VIb-d

Model studies were carried out in the dry box to determine the reaction behavior

of bisphenols Vlb-d with two equivalents of I in acetonitrile. The reagents were mixed as


Reagent Amount
I 0.222g (7.16x0-4 mol)
VIb 0.090g (3.56x10 mol)
CH3CN to 0.80ml

Reagent Amount
I 0.222g (1.43x10-3 mol)
VIc 0.240g (7.14x10 mol)
CH3CN to 0.90ml

Reagent Amount
I 0.lllg (7.16x104 mol)
VId 0.183g (3.56x104 mol)
CH3CN to 0.75ml

In each case, solutions were made as concentrated as possible in CH3CN. These

were placed in small heavy walled glass vessels equipped with high pressure stopcocks,

as shown in Figure 44. These were allowed to equilibrate at various temperatures until it

was confirmed that equilibrium had been reached. Temperature was regulated by oil bath

between 270C and 1200C; at temperatures over 1200C, the reaction vessel was removed

stopcockk closed) from the dry box and placed in a salt bath regulated at the desired

temperature 30C. The salt bath consisted of 40% NaNO2, 7% NaNO3, and 53%

KNO3147 which was placed in a glass dish and heated, with magnetic stirring, on a

hotplate. This mixture melts at about 1400C.

Figure 44. Small capacity (1 2ml) reaction flask made from heavy walled glass.
Stopcock is designed to stay sealed at high pressures.

The nature of the equilibria were determined by HPLC analysis. At appropriate

intervals, approximately 15pl of the reaction solution was removed through the stopcock

via syringe and injected into 1-1.5ml of freshly distilled CH3CN. The solution was

subjected to reverse phase HPLC with UV analysis at X=215nm, as VIb-d and their

reaction products with I absorb strongly in this region. Thus when the concentrations of

unreacted Y, monosubstituted YI ("I-YI"), and disubstituted VI ("I-Y-I") were constant
over at least two days, it was determined that equilibrium had been reached.
The reaction products I-VI and I-YI-I were determined by monitoring the

reactions by HPLC as a function of time. The peak due to bisphenol YI was determined
by coinjection and the products I-VI and I-YI-I were easily discernable because I-VI
always "grew in" first, followed by I-YI-I. In each case, the uncatalyzed reaction at room

temperature produced the expected peaks over time and no others. The azlactone I and its
reaction products were best detected by refractive index. Again, only two products were

observed to grow into the HPLC trace by RI detection, as expected.

Separation of the products in CH3CN solution was attempted by the addition of

predicted non-solvents such as H20 and THF; however, the products could not be

separated by this method. Both IR (solution) and NMR (CD3CN) spectroscopy indicated

mixtures of compounds. In one instance, a precipitate that formed from the reaction of

VIb and 21 was isolated and examined by IR spectroscopy and was thought to be the
disubstituted reaction product of VIb (due to the absence of hydroxyl functionality).
Similar reactions employing I and bisphenols VIa-c were used in a qualitative

manner to determine the reversibility of adduct formation and the effects of anhydrous

AIC13 and fractionated DBU on both the forward and reverse reaction. In these

experiments, solutions were mixed as above and allowed to stir at 260-270C for a specific
amount of time so that: first, measurable reaction had occurred; second, reaction time was

the same for all the bisphenols employing the same catalyst (or no catalyst). In this
manner, comparisons between samples could be made. At the specified time, an HPLC
trace was recorded; HPLC parameters and sample preparation are described above. The
reaction mixture was then heated to 2000C for 5-6 hours and a second HPLC trace was
recorded for each sample.

Polymerization Reactions

Stepwise polymerizations with bisazlactone IV and bisphenols VIa-d were
examined for both the forward and reverse (depolymerization) reactions. The forward

reactions were studied in solution; the reverse reaction was studied neat and in solution.

The solutions also were examined for cycling, or reaction after depolymerization.

Forward Reaction

Stepwise addition polymerizations were carried out in the dry box at 260-270C;

CH3CN and CH3CN:THF (4:1 to 8:1) solutions were used. A typical solution was

mixed in the dry box as follows.

Reagent Amount
IV 0.100g (3.57x10 mol)
Vie 0.121g (3.60x104 mol)
CH3CN 0.40ml
THF 0.05ml Total volume: 0.55ml

Solvents were measured by syringe. In every case, the solutions were made as

concentrated as possible. For some reactions, AlC13(anhydrous) or DBU fractionatedd)

were added in catalytic amounts to the reaction solutions before adding the solvent to the
indicated volume.

SEC was used to monitor the reactions and determine molecular weights. This
data was obtained by removing 15l aliquots of the reaction solutions from the dry box

and diluting with 1 1.5ml of freshly distilled THF. This solution was passed through

the SEC with UV detection at X=215nm. Three columns were used: Phenogel 100A and

500A columns, and a TSK G3000 column were connected. It was found that separation
by this system of columns was sufficient to isolate VI from the other substituents

(oligomers or polymers) of the reaction solutions (1_ absorbed only very weakly).

Therefore, column calibration was carried out using both polystyrene standards

(molecular weights 726 to 7820 g/mol) and the bisphenol monomers themselves. In this

manner the presence of bisphenol monomer in the solution could be unequivocally

determined. Molecular weights of the reaction products of IV and VIa-e were also

determined using these calibrations. The presence of I was confirmed by RI detection.

The polymer formed from Vie and IY was characterized by standard techniques

of dilute solution viscometry148 in THF and 1H NMR in d8-THF (mixed outside of the

dry box; TMS standard) as well as by SEC. Equipment used for viscometry is described

in the General Information section. Additionally, 13C NMR analysis was attempted for

this polymer (in dg-THF).

Reverse Reaction (Depolymerization) and Cycling

The depolymerization reactions were examined by several different methods.

Heating of the neat polymers were carried out by differential scanning calorimetry (DSC)

and in a microdistillation apparatus in vacuo. Solution reactions were carried out in sealed

vials in the dry box or in heavy walled glass vessels equipped with high pressure

stopcocks. Cycling reactions, to determine whether polymer would reform after

depolymerization via heating, were carried out by cooling the depolymerization solutions

and allowing them to stand.

Neat polymer samples were obtained by removing the reaction solutions from the

dry box and evaporating the solvents. The samples were then used without further

purification. DSC runs were obtained for the neat polymer synthesized from IV and Vic,

using a temperature ramp of 100C/min from 400-3000C. This ramp was repeated several

times for the same sample, after cooling at the maximum rate in between runs.

Heating the neat polymer mixtures in vacuo was accomplished using the

microdistillation apparatus pictured in Figure 43. Before the experiment, an SEC sample

was removed from the polymerization solution as described above, for comparison to the

products of heating. The procedure used for the microdistillation was the same as

described above, except the reaction flask was charged with the polymerization solution,

not the neat polymer, the application of vacuum served initially to evaporate the solvent

from the system. This allowed the apparatus to be removed from the dry box without

exposing the reaction solution to air. After heating this mixture from room temperature to

3000C at about 50C per minute, the apparatus was cooled and the contents of the reaction

flask, distillation head, and catch flask were washed separately and quantitatively with

THF. The wash solutions were then passed through the SEC (UV and refractive index

detection) for comparison to the samples obtained before heating.

Depolymerization solutions were the same solutions as used for the forward

reaction. A sample for SEC analysis was removed prior to heating, and the solution was

heated either in a capped vial or in the apparatus pictured in Figure 44. Heating of

solutions at the desired temperature was accomplished in the dry box, either with a sand

bath or oil bath. Six hours reaction time was sufficient in most cases to obtain complete

reversibility, although for some experiments longer heating times were used.

Extent of depolymerization in all cases was then determined by SEC analysis,

using the three column system and calibration method as described earlier.

Cycling of the polymer solutions was examined by allowing the reaction solutions

to remain in the dry box after heating to affect depolymerization. After several weeks, a

sample for SEC was removed and the solution analyzed for any formation of polymer. If

none had formed, the solution was kept for a period up to three months to see whether

polymer could reform under dry box conditions. SEC traces of the solutions were run

approximately every two weeks to check for polymer reformation.

Crosslinking Reactions

Crosslinking of Y, poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one), was

accomplished using bisphenols Va-d. Both the forward and reverse reactions were

studied. Two polymers, Ya (Mw = 2x106 g/mol) and Vb (Mw 75,000 g/mol) were used

in the experiments. Their synthesis is described above. All reactions were run in THF,

DMF, or DMSO; these solvents were found to dissolve both polymers as well as all

bisphenols VIa-d.

Crosslinking Forward Reaction

Polymers Va-b were dissolved in THF, DMF, or DMSO along with the specified

bisphenol Va-d as shown for a typical solution mixed inside the dry box.

Reagent Amount
Va 0.070g
Vie 0.025g (0.15 equivalents; 15%)
DMF to 1.0 ml

In some cases, several grams of polymer were initially dissolved in solvent;

crosslinking solutions were then measured in milliliter aliquots before addition of the

crosslinker. "Number of equivalents" was determined by dividing the weight measured

of either polymer by 139, which is the molecular weight of one repeat unit. Assuming no

hydrolysis of azlactone functionality, this gives the total moles of azlactone functionality.

The desired number of equivalents of VI was then calculated based on the number of

moles of repeat units of V. Percent VI is this number multiplied by 100. Solutions with

0.02-0.03, 0.15, and 0.50 equivalents (2-3%, 15%, and 50%, respectively) of YI were

measured in this fashion.

These solutions were allowed to stir in vials or roundbottom flasks inside or
outside the dry box. Reactions performed outside of the dry box were run either under

argon or open to the atmosphere. Gelation was determined by the point at which the stir

bar could no longer stir the solution, and swirling of the vial revealed a swelled, insoluble

material. Control solutions of polymer in the absence of crosslinker were run alongside

all crosslinking solutions to determine whether gelation could occur in the polymer


Confirmation of gelation was made by filtering the insoluble material from

solution and washing several times in fresh THF. The contents of the THF wash were

analyzed by SEC to determine soluble materials. The insoluble material was then dried in

a vacuum oven at 500C overnight before reversibility tests were run. In some cases

reversibility tests were run in the original solvent, without separation of solvent and

insoluble material.

Reverse Reaction (Decrosslinking) and Cycling

The reverse reaction was examined in several ways. TGA and DSC runs were

performed on the materials in the absence of solvent, using a temperature ramp of

100C/min from 400C to 4000C or 4500C under N2 blanket The washed and vacuum-dried

networks were also returned to the dry box and examined in sealed vials or in the high

pressure vessels shown in Figure 44. The networks were heated at the desired

temperature in a sand bath or salt bath, either neat or in solvent. The neat reactions were

cooled after heating and then extracted with solvent; the wash was then examined by


The gels swelled with solvent were also observed for the disappearance of the gel

upon heating: both the linear polymer V and the bisphenols YI are soluble in the solvents

used. Therefore the mixtures were examined for formation of solution during heating,

and the persistence of the solution thus formed upon cooling to ambient temperature.

Another method, used to show cycling of the crosslinking solutions in DMF,

involved partitioning the crosslinking mixture into three aliquots immediately after

mixing. The first aliquot was immediately dissolved in THF and analyzed by SEC for the

presence of polymer, crosslinker, and solvent. The second aliquot was analyzed after

gelation by washing out soluble materials with THF and again analyzing by SEC.

Finally, the third aliquot was heated after gelation was observed in the mixture, and the

resulting solution was dissolved in THF for a third SEC measurement.

No SEC measurements were made after the end of the first cycle

(decrosslinking). Rather, the observation of insoluble materials after heating the network

sufficiently to decrosslink the first time was taken as evidence that the network was

indeed forming again.

Cycling experiments were performed for all insoluble networks observed to

undergo the reverse reaction after initial crosslinking. These experiments were carried out

either in the dry box or under argon blanket. The time required for the mixture to form a

gel was noted; the solution was then heated for the amount of time necessary to reform a

solution which persisted after cooling the flask to ambient temperature. The solution was

allowed to stir at ambient temperature until gelation was again observed. Cycling was

repeated until gelation was not observed two weeks after heating.

For some cycling experiments, gelation was carried out in air. Heating of the

networks to obtain soluble materials, however, was performed under argon to prevent

failure to recrosslink after solution formation. In some of these experiments, dried

networks were obtained between gelation and reversal by drying the insoluble network in

a vacuum oven. The networks were then resuspended in solvent to affect reversal.

Error Analysis

Error analyses were performed by traditional methods for determination of

standard deviations.149 Error was determined for concentration of solutions, reaction

coordinates (concentration of products divided by concentration of reactants), and rate


Errors in Concentration

"Concentrated solutions", which refers to all 20.5M solutions, generally had a

larger percent error than dilute (0.1-0.5M) solutions because dilute solutions were usually

mixed in larger amounts. Since the errors for these two cases were found to be the same

for all solutions, one example for each type of solution is presented here.

A concentrated solution of isopropyl azlactone (D was mixed by measuring

0.160g of I to an accuracy of 0.001g into solvent to give a total volume of 2.0 ml with

an accuracy of 0. ml. To calculate the number of moles, the error in the accuracy of the

weight is used to determine the relative standard deviation. Thus ([0.001/0.160]2)1/2

yields a relative standard deviation of 6.25x10-3; multiplication by 1.03x10-3 moles

(molecular weight of azlactone I is 155g/mol) yields an error of 6.45x10-6 moles. Thus

the values of concentration reported for concentrated solutions are in error by 0.01x10-3

moles and the above value would be 1.03x10-3 + 0.01x10-3 moles.

In a similar fashion, the error in concentration of the concentrated solutions was

performed for the calculation where number of moles (1.03x10-3 0.01x10-3 moles) is

divided by the measured total volume of solution (2.0 0. Iml). The relative standard

deviations for number of moles and number of milliliters are additive; multiplication of

this value by the calculated concentration (0.52M) yield an accuracy of 2.6x10-2M;

therefore the value as reported is 0.52 0.03M.

Added to this error is the error in measuring concentrations by 1H NMR for

reactions measured in this manner. Precision in the measurement was determined by

taking a single solution of I and measuring its concentration several times by NMR, as

described earlier. The precision of the measurement was determined in this case to be

0.005M. This error does not significantly change the amount of error found previously

by measuring the starting concentrations of reagents.

The accuracy of the 1H NMR measurements was determined by comparing the

ratios of proton measurements in the cases where internal standard (CH3CN) was used to

measure disappearance of I. The ratio of L:CH3CN was calculated based on both the

expected value from weighing the two constituents of the solution, and by measuring the

actual ratio of their proton integration at time zero (t=0).

The weight ratio of I:CH3CN has a standard deviation of 0.005 by calculations

as carried out above. The 1H NMR measurements have a precision of 0.005M as

determined above. Therefore, a typical solution was weighed out such that the molar ratio

of the measured methine proton of I to (CH3CNx3) would have an expected value of

0.530.01. In the actual integration of the protons in this sample, the t=0 spectrum

yielded a value of I:CH3CN = 0.52. The measured ratio, then, is accurate to within the

experimental error encountered in the weighing of I and CH3CN.

In the model studies, measured values of products are determined by the

disappearance of starting reagent I over some period of time. The value of the

concentration of products, then, has the same error as for the measured concentration of


Model studies using bisphenols were also carried out at high concentrations.

Solutions approximately 0.5M in species VIa-d and 1.OM in I were mixed in the same

manner as described for the model studies of I and II. Error in the accuracy of the

molarities of these bisphenol solutions are therefore the same as for the molarities of

solutions of I and II. Polymerizations and crosslinking solutions were found to have the

same error in molarity, as their concentrations were measured similarly.

Error in Reaction Coordinates and Kinetic Data

Reaction coordinates, determined only for concentrated solution model studies,

were also subjected to error analysis. This value is simply concentration of products over

reactants, in this case ([III]/[J][MI). Again taking the relative standard deviation of the

three measurements and multiplying by the calculated reaction coordinate value, the error

in this calculation was determined to be 2.4x10-11/mol.

Finally, the concentrated solution model studies were subjected to kinetic analysis

and forced to fit a semilog plot. Since the semilog plot could be made to fit the data only

at very early times of reaction, the calculated values of [l]o/[]t were found to be very

small. The error in a typical measurement yielded []Uollt = 1.02 0.09 (the use of an

apparent excess of significant figures is necessitated by the need to differentiate the ratio

when taking its natural log); however, when this same error analysis was applied to the

natural log (In) operation for the value of [DIo/lt, the error was found to be in excess of

the values themselves due to the large variation in the value of the natural log in the

vicinity of In 1. Thus, the values recorded for the apparent first order rate constants are

used only as relative measurements of reactivity, since differentiation of concentration in

the reactions was possible. The rate constants were reproducible over one significant

Dilute solutions presented a different case for error analysis. Since these solutions

were made initially in larger volumes (generally 5ml or more instead of 2ml), the error

inherent in the calculation of concentration was less than for the concentrated solutions.

Dilute solutions were used only in the model studies and concentration values were used

only to calculate apparent second order rate constants. The same methods as used above

yielded the error for number of moles of azlactone as being 1.01xl0-3 0.01x10-3

moles. This amount of azlactone was diluted to 10ml (instead of 2ml), leading to an error

of 1.2x10-3M or a reported value of 1.0x10-1 0.01xl0-1M. In a more typical case,

one half this amount of solution was mixed and the error was found to be 0.02x10-IM.

In this case, the amount of error made by NMR measurement becomes significant,

raising the error in each measurement to 0.07x10-1M; concentrations monitored via

NMR are therefore reported as 0.10 0.01M.

The error in calculating the second order rate constants for the dilute solution

model reactions were performed in the same manner as outlined above. The error in time

measurement was approximated at 240s, since the time for sample removal from the dry

box and positioning within the NMR probe was not counted in the dilute solution

measurements (due to the insignificant amount of reaction found to take place during the

2-4 minutes required to perform these operations). Thus the error was calculated for

1/[Ut 1/[I]o = kt, where t is time and k is the apparent rate constant. The error was

found to be 7.8x10-71/mol s, for typical value of 9.8x10-6 0.8x10-61/mol s.

The accuracy of the HPLC measurements of the reaction products of VI with I

was not determined for the cases where equilibrium constants were calculated. However,

precision in these measurements may be estimated. "Equilibrium" was defined as the

point in the reaction when the measured concentrations of the reaction products varied

randomly over 0.001M. Thus, if it is assumed that equilibrium had indeed been reached

in this case, the variation over 0.001M may be considered to be the precision of the

measurement. This does not add appreciable to the error encountered in the mixing of the

reaction solutions (0.03M). For a typical measurement of equilibrium constant, the

standard deviation of [-I--I] over ([j]2[Y]) was determined to be 4.8 12/mol2. The

value of equilibrium in this case would be reported as 12 4.8 12/mol2.


Can thermoreversible covalent polymer linkages be formed using azlactone -

phenol chemistry? This question forms the basis for the research described herein, which

is summarized below.

Model studies using a monofunctional azlactone and three monofunctional

phenols, shown in Figure 45, were performed to probe the mechanism of the reaction.

This information was used to predict the behavior of both linear and network

polymerizations based on azlactone phenol chemistry, illustrated in Figure 3.

The thermoreversible linear polymerization of bisazlactone with a series of

bisphenols exhibited differences in reactivity when compared to the model studies. These

variations in behavior stemmed from the structural differences of the bisphenols relative

to the monofunctional phenols. The reactivity of the bisphenols was best understood by

studying the reactions of bisphenols with the monofunctional model azlactone.

A thermoreversible covalent network polymer was fashioned based on both the

model and linear polymerization chemistry described above. The same bisphenols were

used as crosslinkers, forming covalent polymer networks which survived several thermal

crosslinking decrosslinking cycles. This system represents one of the first successful

means of recycling a covalently crosslinked network [see Chapter 1, "A History of

Thermally Controlled Covalent Polymer Linkages"].

A Model Study of the Forward Reaction

The initial stage of research involved the study of a model system which mimics

the chemistry of the linear and network polymerization schemes targeted in this research.

In this model study, 2-isopropyl-4,4-dimethyl-2-oxazolin-5-one, L was chosen to mimic

the bisazlactone 2,2'-tetramethylenebis(4,4-dimethyl-2-oxazolin-5-one) as well as the

polymer poly(2-vinyl-4,4-dimethyl-2-oxazolin-5-one), also known as poly(2-vinyl-

4,4-dimethyl azlactone). The reaction scheme for the model study is shown in Figure 45.

o H RT

I la X=NO2 Ilia X=N02
lb X=H IIb X=H
11 X=OMe IIc X=OMe

Figure 45. Envisioned scheme for the reaction of I and 11 to form II, and of III to form I
and II.

The monofunctional phenols p-nitrophenol (Ia), phenol (Iib), and

p-methoxyphenol (ic) were selected to mimic 4,4'-bisphenol monomers of varying

degrees of electron withdrawing or electron donating capabilities. Thus the effect of these

substituents could be determined for both the forward and reverse reactions.

Proton NMR spectroscopy (Figure 46) was chosen as an accurate and

reproducible means of monitoring the forward reaction, where the septet due to the

methine proton of the isopropyl group in both I and III was used for analysis. In the

spectra of solutions containing both reactants and products, only these two signals were

cleanly separated such that integration of their areas could be differentiated and


To ensure that the reactions of I and II were accurately measured, acetonitrile was

included as an internal standard in some of the reaction solutions to measure the

disappearance of I. The measured concentration of I could then be compared to the value

found for the measurement of the ratio [I]:[II. In all cases examined, the values agreed

within 5%, so the ratio []:[Jj] was used to determine the concentration of each species.

Repeated measurements of a single solution of 0.51M isopropyl azlactone (I) and

0.15-0.20M CH3CN in d6-DMSO revealed that the measured septet can be determined to

0.005M using 48 transients per spectrum.

The observation of a shift of phenyl protons concomitant with the apparent
formation of II was used as evidence that the reaction was proceeding as expected. This

check was necessary because adventitious water can react with azlactones. Nucleophiles

such as water compete with II in the ring opening reaction, and cause the 1H NMR shift

of the measured methine proton to be very close to that of the expected adducts III. Since

the aromatic protons of ill were shifted slightly upfield from aromatic protons due to I,

they were used as a means of checking that the desired reaction indeed was occurring.

Reactivity Trends of Phenols IIa-c with Isopropyl Azlactone I

Several reactions equimolar in I and each of the phenols IIa-c were performed in

both dilute (0.10M) and concentrated (0.52M) solutions, using several solvents. In dilute

solutions, less than 5% of the forward reaction had taken place over seven days in most

cases, which made analyses both tedious and less precise. Difficulties in carrying out

dilute solution reactions of I and II prompted the study of concentrated solutions. Such a

study was relevant to the model system at hand, since the envisioned step linear and
network polymerizations require concentrated solutions.
The relative reactivities of Hla-c with isopropyl azlactone I were measured by

examining side-by-side reactions (in concentrated solutions) after 24 hours of reaction

time. Table 3 shows the results of 1:1 reactions of I with each of the phenols II in various

N 400C /24 hr
HO N02 + o0 d4-DMSO



a a' b'

3 a PP" 1=5


b' H 0NO


I i I I I
2 3 PPM
Figure 46. Spectra of the reaction of I and IIa after 24 hours at 400C in d6-DMSO. a.
Proton NMR signals due to I and IIIa. b. Magnification of protons a and a'. Interference
in the a' signal is due to protic DMSO, which was subtracted out for quantification.

Table 3. Observed reaction ratio after 24 hours at 270C for the uncatalyzed reaction of I
with II in various solvents, as measured by 1H NMR. All solutions were initially
0.52M0.03M in I and II.
II Solvent at 24 hours
Ha CDC13 4.7 0.2
S DMSO 1.0 0.2
S Chlorobenzene 9.3 0.2
IIb CDCl3 0.1 0.2
S DMSO 0.2 0.2
S Chlorobenzene 0 0.2
Ic CDCl3 0 0.2
S DMSO 0 + 0.2
S Chlorobenzene 0 + 0.2

solvents, as measured by 1H NMR. The progress of the reaction is expressed as the ratio

of product to reactants at 24 hours.

The data show that the relative reactivity of phenol nucleophiles II is dependent

not upon nucleophilic strength, but rather on acidity of the phenolic proton.

Para-methoxyphenol (Jj.), the strongest nucleophile used in the reactions, did not

undergo measurable reactivity in 24 hours. Instead, the weakest nucleophile (and

strongest acid), para-nitrophenol (lp), reacted to the furthest extent in 24 hours. This

observation was initially unexpected.

A two-step mechanism for this reaction is shown in Figure 47. Initial proton

transfer to the azlactone nitrogen from the phenol is followed by ring opening of I by a

phenolate anion. Thus in uncatalyzed additions, p-nitrophenol (Ia) is expected to be the

most reactive nucleophile since it is the most acidic of the three phenols. However, these

experiments do not address which of these first two steps is rate controlling for the

reaction: is there a fast pre-equilibrium for proton transfer to the azlactone ring, followed

by slow nucleophilic attack, or is proton transfer to the azlactone ring rate controlling?

The obvious mechanistic assumption is that nucleophilic attack (step 2) is the

slow step, since there is little precedent for a rate determining proton transfer (step 1) in

such a reaction. The relatively high amount of product liIa (the 1:1 adduct of isopropyl
azlactone and p-nitrophenol) observed at 24 hours may indicate the presence of a higher
concentration of p-nitrophenolate anion in the solution at equilibrium for step 1, and
hence, a higher amount of product formed in the second step. This has been found to be

the case in similar reactions.150

S-X + o Ostep 1 + +

X step 2 0
+ I Y
HN H 0
H I inX

Figure 47. Proposed mechanism for the forward reaction of I and II.

Experiments with sodium p-nitrophenolate were carried out to determine whether
the concentration of phenolate anion was important in the reaction. If nucleophilic attack
is rate controlling for the reaction, then reaction solutions enriched with p-nitrophenolate
should react at a faster rate. However, the addition of sodium p-nitrophenolate caused a
precipitation of unknown nature immediately upon mixing, and no conclusions could be
drawn from these experiments.

The Lewis acid AIC13 was used to study the forward reaction mechanism. The

proposed mechanism, shown in Figure 48, predicts that "acidification" of the azlactone
nitrogen by the Lewis acid is followed by nucleophilic attack by phenol. Thus the
reaction would be independent of the relative acidity of the phenol nucleophiles II, and
instead would depend only on relative nucleophilicity. Therefore the expected result is
that IIIc would be formed in the highest relative concentration at 24 hours, and Ila would
be formed in the lowest relative concentration.

O HO X step 2 0 H
A 1 + HO _X 0

C13AI s 3 AIC L
0 H 0

Figure 48. Proposed mechanism for the reaction of I and II in the presence of AlCl3.

Proton NMR spectra for the reactions of 0.52M solutions (equimolar in I and 1_) with 0.5

molar equivalent of AIC13 are shown in Figure 49, and it appears that the expected

reversal of adducts IIIa-c occurred at <0.5 equivalents of AIC13. Adducts IMb and III

were formed quantitatively, while Ha reacted to about 50% completion with I.
These experiments show that "acidification" of the azlactone imine site is
important in the reaction, and results of these experiments support the reaction
mechanism as shown in Figure 47. Additionally, the fact that the reactions to form IIIb

and III go to completion in 24 hours by adding sufficient AIC13 is promising for the

corresponding linear stepwise polymerizations, which require near 100% reaction to form
polymers of appreciable molecular weight. In order to fit the additional requirements of
thermoreversibility, these reactions should undergo the reverse reaction at temperatures of
<3000C; this question is addressed in the section dealing with model studies of the
reverse reaction.

A; I

3 4 PPH



i t t i i
-i --- --- I -- l ---


f L

I + Ha 0.5 eq. A1CM -E



ib 0.5 eq. AIC

+ 1 0.5 e. AgAlQ mc


Figure 49. Spectra of I and II (0.52M) as indicated, after 24 hours in CDCl3 at 270C with
0.5 molar equivalent AlC13. The quality of the spectra does not allow quantification.



Reactions with strong protic acids (F3CCOOH, H2SO4, etc.) were also examined

in a similar manner to the reactions employing AlC13. Side reactions of an undetermined

nature, possibly due to the reactions of acid counterions with L interfered with the

formation of the expected products. The side reactions were thought not to be the

formation of ring opened azlactone anhydride, since the side products caused signals of

different shifts in the NMR spectrum when different protic acids were used. Since the

envisioned linear and network polymerizations based on the model must be free of side

reactions, it was concluded that protic acids are not suitable reagents or catalysts in this


Kinetic Studies of the Forward Reaction

An analysis of the forward reaction of I and II-c was conducted under dilute
solution conditions for the study of kinetic data. Similar experiments were examined in

concentrated solutions, which are more accurate as a model of the corresponding

polymerizations. All of these reactions were studied in the absence of other reagents,

such as A1C13, to avoid complications such as the possible dimerization of AICl3 or other

side reactions. Data obtained from the study, such as kinetic order and relative rate

constants for the forward reactions, contributed to a better understanding of the model

reaction system.

Measurements were made by 1H NMR as before. The dilute solution kinetics

(0.10M, equimolar in I and ILa-c) appeared to be second order, though there was wide

variability in the coefficient of correlation for the data due to long reaction times. Days or

weeks were required to collect data. For example, the reaction of isopropyl azlactone (I)

and p-nitrophenol (Ila) produced a second order rate constant of 9.8x10-6

0.8x10-61/mol s (correlation = 0.98) at 270C. This data, shown in Figure 50, required

nearly 60 hours of reaction time to collect. Even at 1000C, negligible reactivity was

observed after 18 hours. Phenols lib and IIk were less reactive than IIa, as expected.

When pseudo-first order conditions were employed in order to accelerate the rate of

reaction and find the kinetic dependence of each species in the reaction, experimental

difficulties prevented analysis of kinetic data.

Fitted Overall 20 Kinetic Data

y= 0.355 R =0.98


5.00e-1-- I-- ----------

-5.00e-1 #
Oe+0 le+5 2e+5
time, s

Figure 50. Fitted overall second order plot for the reaction of I and Ila at 270C. Note the
long reaction time. "C" denotes concentration of I.

The observed overall second order kinetics are reasonable based on the reaction

mechanism proposed in Figure 47. Both of the steps shown involve bimolecular

processes; regardless of which step is rate controlling, first order dependence on each

reagent would be expected if the mechanism is correct.

A kinetic study of the forward reaction in concentrated solution was carried out.

Such a study was relevant as a model for the linear and network polymerizations targeted

in this research, since step polymerizations are done either in highly concentrated

solutions, or without solvent. Additionally, the enhanced reactivity expected in

solutions of higher concentration makes analyses simpler. Therefore, the forward

reactions, as shown in Figure 45, were run in 0.52M d6-DMSO solutions.

[I] vs. time
6.0e-1 -

a I+l II
E< -IL 1 ,h 1 I +llc

0.Oe+0 1.0e+5 2.0e+5 3.0e+5
time, s

Figure 51. Plot of [] vs. time for the reaction of 0.51M I and IIa- at 27C in d6-DMSO.

Figure 51 shows the change in concentration of isopropyl azlactone, I, vs. time

for the forward reaction in concentrated solution. Clearly, p-nitrophenol (UIh) reacts faster

than does phenol (ITh) orp-methoxyphenol (le). But while the reaction ofp-nitrophenol

(II) initially is faster, rate retardion occurs between 15-20% conversion, whereas lib and

IIc continue to react. (It appears as though the reaction of I and IIb may also be

proceeding towards rate retardation.) The forward reaction of I and IIa following the

retardation point proceeds slowly but is not an equilibrium, since further reaction was

observed when the experiment was carried out for a three month period.

Data taken before the rate retardation (Region 1, Figure 52) was forced to fit an

overall first order reaction plot; however, no simple kinetic fit could be found for the

reaction after this point. Figure 52 shows data for the reaction of I and Ila at 270C in

d6-DMSO, applied to a semilog plot, and Table 4 shows rate constant values for the

linear portion of the this plot for each of the reactions with IIa-c. In the case of I and IIa,

only Region 1 data was used. For the reaction of I with lib and iSc, the calculated rate

constants represent <10% reaction (the portion of the data able to be fitted). The data

were only fit to the semilog plot through the initial stage of the reaction, and this

presented a problem in terms of error in the reported values (see Experimental section).

Therefore, rate constants were calculated only as a basis for comparison of relative

reactivity at early times.


0.2 -

0. o~- --- ---------------------------------------

5 0.1
Regi m 1

O.OOe+0 5.00e+3 1.00e+4 1.50e+4 2.00e+4
t, sec
Figure 52. Semilog plot for the reaction of 0.52M I and IIa at 270C in d6-DMSO.
"Region 1" represents the portion of the data that were fit; "[azl]o" is initial concentration
of I.

Table 4. Observed rate constants for the reaction of I and II at 0.51M, 270C in DMSO.

Phenol para kobs, overall Correlation
Substituent 1st order plot coefficient
NO2 (IIa) 2x10-5s-1 0.99
H (IIb) xl10-6s-1 0.99
OMe (Ic) 8x10-7s-1 0.99

Clearly, IIa reacts faster than the phenols IIb and IIc during the initial portion of
the reaction. This is in agreement with measurements of relative concentrations of IIla-c

made after 24 hours of reaction time of I and IIa-c. However, it is difficult to make other

statements, particularly given the unusual behavior observed in the forward reaction of

isopropyl azlactone (I) and p-nitrophenol (IIa).

The forced kinetic fit of the reaction in concentrated solutions to an overall first

order scheme cannot be explained mechanistically. Since the dilute solution kinetics were

satisfactorily fit to an overall second order reaction scheme, it appears that concentrating

the solutions causes complication of the reaction order.

Pseudo-first order conditions were applied to the concentrated reaction in order to

find the kinetic dependence of the reaction on I and ila individually. A solution of 0.52M

I and 3.7M Ha (saturated in Fa) in d6-DMSO was examined by 1H NMR. However, the

signals of the methine protons of I and IIIa were found to overlap such that quantitative

data could not be generated. These studies were not pursued further.

Several facts relevant to the polymer systems being modeled were clear after

examining kinetic reactivity data. In dilute solution, the reaction apparently is second

order overall. This supports the proposed reaction mechanism as shown in Figure 47,

assuming first order kinetic dependence on both I and II. However, the reaction under

dilute conditions is extremely slow and does not proceed to completion in a reasonable

time period, which is unacceptable for a linear step polymerization scheme. While

crosslinking reactions are more tolerant of low reactivity, it is desirable in either case to

maximize reactivity by concentrating the solutions or removing solvent altogether.

As expected, more concentrated solutions do react faster. However, the fast initial

reaction period for I with the most reactive phenol nucleophile (p-nitrophenol, Ia) is

followed by an apparent rate retardation. The retardation was substantial enough that the

reaction appeared to be at equilibrium on a time scale of several hours. Again, the

reactions do not reach completion in a reasonable amount of time, which would indicate

that reactivity of the system is insufficient to yield high molecular weights in the

corresponding step polymerizations.

Conclusions from the Model Study of the Forward Reaction

Some important implications regarding the design of the linear and network

polymerizations can be recognized based on model study data.

Since phenols with acidic protons are kinetically the most facile in the uncatalyzed

reaction with model azlactone bisphenol compounds with electron withdrawing

moieties are expected to react faster than bisphenols without electron withdrawing

moieties in the linear and network polymerizations. The bisphenols used for the

polymerizations are 4,4'-bisphenols; therefore the functionality between phenol moieties

of the bisphenols correspond to para substituents in the model phenols. Reactivities of the

bisphenols are therefore expected to follow the same general reactivity trends as the

model when comparisons of substituents are made.

Due to the generally unreactive nature of the reaction of I and II, the

corresponding polymerization solutions should be as concentrated as possible in order to

produce stepwise polymers. In the model studies concentrated solutions did allow for a

more facile reaction as expected. However, very high yields (>99%) must be realized in

the stepwise reactions in order to produce high polymer.

A solution to this problem may be the use of a Lewis acid such as AIC13, which

produced quantitative amounts of IIIb and III in the reactions of I with lib and li.

respectively. This approach is considered in the reactions to form step polymers.

Additionally, it was thought that Lewis acids would cause bisphenols with electron

donating groups to be the most facile nucleophiles in the forward reactions to form linear

polymers. In order to be useful, however, the reactions that form high polymer must also

be reversible to reform the starting materials monomerss) upon heating. The question of

reversibility in the model system is considered in the next section.

A Model Study of the Reverse Reaction

The reverse reaction of I to form I and 11 (Figure 53) was examined to determine
the viability of using the azlactone phenol reaction to form thermoreversible covalent
polymer linkages. Purified adducts Ea-e were used as the starting materials, and the
temperature at which the "reverse" reaction occurs was measured by several techniques.

CHa H 0 CH3

X A X- A o + H X
0 HC CH3 CH3
Ilia X=NO2 I CH3 a X=NOz
lb X=H Ilb X=H
Ind X=CF3 lid X=CF3
IIle X=F Ile X=F
Figure 53. Schematic outline for the study of adducts II, which were expected to give I
and II upon heating.

Model Reverse Reactions in Sealed Tubes
Adducts nia-c were heated at several temperatures (1000C, 1500C, 2000C, 260C
or 3000C) for 24 hours in sealed heavy-walled glass tubes, followed by quick cooling,

and the products were examined by 1H NMR as before.
The uncatalyzed reverse reaction of adduct IIa becomes significant between
2000C and 2600C, with 3:1 ratio of I:IIa being observed after 24 hours (Figure 54).
Raising the temperature to 3000C increased this ratio to 5:1. Adducts MIb and IIIc did not
reverse at all at temperatures up to 3000C. The reverse reaction of lia provided the first
evidence that the azlactone phenol reaction is thermoreversible in a practical sense,
and these results show that this chemistry has potential to form reversible covalent
linkages for linear and network polymers.


O.OM d -DMSO I 24h / 2WC

3.0 2.5 2.1 ppm


H0 0
10M / d4-DMSO / 24


2 2.6 2.0 ppm

3.0 2.5


' aO.OMI/ -DMSO/24b/26fC

2.1 ppm

Figure 54. Proton NMR spectra of adducts Illa-c after 24 hours at 2600C, in solutions
0.10M in d6-DMSO. The peak at 2.5ppm is due to proton exchange with DMSO.

ht/ 2MC


The data suggest that ejection of the phenol (ie., the reverse reaction to form I and

If) at elevated temperatures is a function of the stability of the phenolate anion leaving
group. This was thought to be true because the nitro substituent causes I&la to reverse

below 3000C while IIIb and Hik do not reverse at all.

Model Reactions Under Vacuum

Additional evidence for the successful reverse reaction of Ila was obtained by

heating the neat crystalline adducts in a "micro" distillation apparatus. The apparatus and

procedure are described in greater detail in the Experimental section. A slow temperature

ramp was applied to each of the neat adducts IIIa-e (Figure 53) under vacuum, and

disappearance of II was carefully observed. The apparatus was cooled and dismantled

after disappearance of III, and the contents of the distillation column, catch flask, and any

residue remaining in the heated flask were washed out with CH3CN. The wash was

analyzed by HPLC (both UV and RI analysis). This technique has the advantage of not

exposing adducts to heat for long periods of time, so that the probability of degradation is

Adduct MIla was observed to disappear from the heated flask between 2350C and

2500C, and the materials in the catch flask were found to be I and IIa, with traces of IIIa.

Coinjection of I and its hydrolysis product (N-isobutyryl-2-aminoisobutyric acid)

confirmed that hydrolysis was not responsible for the appearance of Ila. These data agree

with the sealed tube experiments, which show that ring closure of IIa to form I and IIa
occurs between 200C and 2600C.
Other adducts IIIb-e were examined using this technique, and the results are

summarized in Table 5. These results also agree with the sealed tube experiments using

nIb and IIIc. since lITb and IIIc persisted in the heated flask up to 3000C. HPLC analysis
showed that no reverse reaction had occurred. The fluorinated adducts IId and IIIe