You are on page 1of 53

Prog. Polym. Sci.

26 (2001) 799±851
www.elsevier.com/locate/ppolysci

Isocyanates in polyaddition processes.


Structure and reaction mechanisms
A.A. Caraculacu*, S. Coseri
The Romanian Academy's Institute of Macromolecular Chemistry ªPetru Poniº, Gr. Ghica Voda Alley 41 A,
Iasi 6600, Romania
Received 14 March 2000; accepted 27 June 2000

Abstract
The multiple possible addition processes of isocyanates, in accordance with a new proposed general classi®ca-
tion based on the interatomic bond type changes is presented. Starting from the latest achievement of the physico-
chemical methods, new aspects regarding the isocyanates structure and different hydroxy compound association
forms as part of the addition processes were revealed. Some basic reaction mechanisms presented in the current
research were re-evaluated in accord with recent quantum chemical and kinetic data. The intramolecular hydrogen
bonding in the symmetrical diols was found to determine a signi®cant difference between their two ±OH group
reactivities. Recent kinetic and other reactivity measurements, extended to polyfunctional systems reported in
some macromolecular synthesis, were presented.
The perspective of these ®ndings, in order to obtain in the near future an acceptable mathematical model of the
polyaddition process by involving the multiple new synergetic effects observed in different real complex reaction
mixtures, is analysed. q 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Isocyanates; Alcohol associates; Polyaddition processes; Reaction mechanisms; Structure; Reactivity; Kinetics

Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
2. General considerations regarding the isocyanate structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
3. Considerations regarding nucleophilicity and basicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
3.1. The nucleophylic reactant XH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
3.2. The hydrogen bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
3.2.1. Alcohol self-association structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
3.2.2. Dynamic and quantum aspects of hydrogen bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . 815
3.2.3. The hydrogen bonding in glycols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815

* Corresponding author. Tel.: 140-32-140-919; fax: 140-32-211-299.


E-mail address: coseris@ichpp.tuiasi.ro (A.A. Caraculacu).

0079-6700/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S0 0 7 9 -6 7 0 0 (0 0 ) 0 0 03 3 - 2
800 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

4. Pattern used to elucidate the mechanism of uncatalysed alcohol addition to isocyanate . . . . . . . . . . . . 818
4.1. Quantum chemical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 818
4.2. The kinetic methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821
4.2.1. The kinetic studies of the urethane reaction performed under unperturbated hydrogen-bond
association equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 822
4.2.2. The kinetic correlation with the alcohol average self-association degree mixed alcohol
associations and polyols presenting intramolecular bonding . . . . . . . . . . . . . . . . . . . . . . 825
4.3. The kinetics and other reactivity measurements in polyfunctional and macromolecular systems . 830
4.3.1. The glycols reactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
4.3.2. The diisocyanate reactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
4.4. The kinetic study on the real polyaddition macromolecular synthesis between diisocyanates and
diols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 840
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 848
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 848

1. Introduction

This review is dedicated to isocyanate chemistry research Ð one of the most fascinating sections of
organic chemistry. The development of this study was strongly determined by the fact that isocyanates
represent the most important raw materials in polyurethane synthesis.
In general, the macromolecular structure of polymers can be displayed by the repetition of a single
simple structural unit (usually easily deduced even from the name of the polymer); in the case of
polyurethanes the situation becomes more complicated. Under the name polyurethanes, a practically
unlimited number of structures can be involved. The only necessary condition is, in general, the presence
of the urethane group on the macromolecular chain.
The polyurethanes represent the ®rst example of polymer building using the so-called ªtailoringº
procedure. In this technique one inserts, by a process that involves many steps, molecular fragments with
a high diversity of structures and dimensions into a single macromolecular chain.
Until recently, the participation of the isocyanic group in this process was limited to the well-
known addition reaction of a compound that contained a mobile hydrogen atom of the X±H type,
to the isocyanic NyC double bond, leading to the urethane or urea group generation. Recently, it has
become more important to take into account other possible NCO-group reactions. The variety of
these reactions, included by Ulrich [1] in the so-called ªunconventional chemistry of isocyanatesº,
complicates attempts at systematising isocyanate-generated polymers. In our opinion, isocyanate
reactions, in principle, can be ordered by considering the reactants±partners inter-atomic binding-
type changes.
In our attempts to undertake a good reaction systematisation we considered three general cases: (a)
reactions which were already used in the polymer synthesis; (b) reactions occurring only sometimes as
secondary processes; and (c) reactions not mentioned so far in the polymer synthesis. So the general
situation can be reduced to the follow cases:

p ± p transformation. Characterised by new s bonds appearing by the cleavage of the NyC isocyanate
p bonds.
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 801

(1) Isocyanate homoadditions (Eq. (1))

…1†

(2) Heteroaddition of isocyanates on other p-bound containing compounds (Eq. (2))

‰2±9Š …2†
802 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

(3) Heteroaddition of isocyanates on p conjugated systems. Surprisingly, in this case the additions
are in the 1,2 position of a conjugate diene and not in the 1,4 position [10] as in the case of Dies
Alder reaction (3).

…3†

p ± s transformation. Characterised by new s bonds appearing as a result of the cleavage of a


polarised s bond from an isocyanatophile partner, in parallel with the cleavage of an isocyanate
±NyC p bond.
(1) Atom or group migration to the isocyanate NyC p bond (Eq. (4))

…4†

(2) Cycle opening followed by addition to the isocyanate NyC p bond. In this case an
unexpected isocyanate CyO addition can appear (Eq. (5)).

‰6; 11±13Š …5†

The above scheme constitutes only an overview of the main types of isocyanate reaction.
Extensive reviews of other types of particular isocyanate reactions are published elsewhere [14±
16].
In some cases the synthesis of a de®nite product can be achieved by two different mechanisms. Such
a case of isocyanurate ring formation (reaction type I-1) is also possible to be achieved by a type-II-1
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 803

step-by-step mechanism. This method involves traces of urethane participation [17] as well (Eq. (6)).

…6†

A similar mechanism for the linear polyisocyanate formation has been reported [18].
If one considers all the known isocyanate structures which include also the oxygen-, nitrogen- and
sulphur-substituted isocyanates [19], the picture of isocyanate chemistry becomes even more complicated.
Certainly not all of the above-presented reactions are of a major importance for the polymer synthesis,
but they do represent potential sources of research and they often must be taken into account to
understand better the secondary processes.
An extremely large amount of research related to isocyanate chemistry has been undertaken. A study
of the 1998 Chemical Abstracts data showed that 2855 works in the ®eld of urethanes and polyurethanes
and 234 works in the ®eld of isocyanates were reported. The major interest of the researchers was
directed towards practical purposes illustrated by the fact that patents represented 66% of the references.
Despite all this effort, many problems regarding the fundamental aspects of the high-isocyanate
chemistry still remain unsolved. With regard to the complexity of all the above-mentioned facts, the
present review is dedicated only to some theoretical aspects regarding the new trends in the study of the
isocyanate polyaddition mechanism with special emphasis on polyurethane synthesis.

2. General considerations regarding the isocyanate structure

The isocyanate group has a structure belonging to the heterocumulene class, and represents a strained
structure able to perform addition reactions when interacting with different compounds that can be
generally considered as phyloisocyanaic reaction partners. As mentioned before the addition is
performed, as a rule, at the NyC double bond and only exceptionally at the CyO group [6].
Recently, attempts were made to show the similarity between the isocyanates and other groups such as
ketenes. The purpose of these studies was to ®nd general relations which should be valid for the whole
class of heterocumulenes [20,21].
As in the case of the other heterocumulenes, the isocyanate group reactivity is based on the polarisation of
these groups represented by the following resonance-limited structures [22] (Eq. (7)):

…7†
804 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 1. Schematic representation of p and p orbital in the ±NCO group.

In concordance with the molecular orbital theory, the isocyanic group is expected to present a linear
structure, the two cumulated double bound NyC and CyO lying on the same axis, and the p electrons
from the respective double bond being situated in two different perpendicular planes.
Owing to the higher electronegativity of the nitrogen and oxygen atoms, the electron density should be
shifted towards the ends of this system according to Fig. 1.
Theoretically, the p-electrons of the NyC and CyO groups should not interact with each another, and
the both double bonds should be regarded as isolated groups.
In fact, interatomic distance measurements made on methyl isocyanate showed a shortening of the
NyC and CyO bonds. In this case, a length of 1.19 ^ 0.03 A Ê was found for the NyC group and a length
of 1.18 ^ 0.03 A Ê was found for the CyO group [23]. In cases where these groups are isolated, the lengths
are 1.27 and 1.21 A Ê , respectively [24].
Taking into account the fact that one of the factors that could show the presence of an extended
molecular orbital is the existence of a shortening of the system bonds, it is not out of question that, even
in the case of the isocyanate group, a small orbital interaction between the nitrogen lone-pair electron
and that of the oxygen lone pair occurs, although these atoms are involved here mostly in their hybri-
dised forms [25]. This ®nding stimulated the researchers' interest in the problem of the isocyanate group
electron density distribution.
For this purpose the majority of standard calculations: Parizer±Parr±People (PPP) [26,27] extended
Huckel method [28±31], Mullicken±Volfsberg±Helmholtz (MVH) method [32], SSP MO LCAO in
valential PPDP/2 approximation [33] and non-empirical ab-initio methods developed in different bases
were checked [34±37].
Bondarenko et al. [38] claimed however that all these works failed in molecular orbital calculations
optimisation. Relatively accurate experimental data on the molecular geometry exist only for HNCO and
CH3NCO molecules [39±43], yet even in these cases the value of the NCO angle was not unequivocally
established.
The results of the diverse studies on the geometry of some isocyanates having the general structure X±
NyCyO is shown in Table 1 [38].
The results of the different methods of calculations concerning the electron density distribution in the
NCO group are presented in Table 2 [38].
These tables show that opinions on geometry and electron density distribution differ.
In the case of phenylisocyanate, most of the authors accepted that the highest electron density lies at
the NCO oxygen. Bondarenko et al. [38], using an optimised computing method, claimed that the
nitrogen of this group is more electrically charged than the oxygen. A similar situation also occurred
in the case of the molecular geometry study of the phenylisocyanate. Surprisingly, the ±NyCyO
group linearity was contested. The ab initio method con®rmed this situation, especially in the case of
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 805

Table 1
Experimental and calculated values of distances and angles in HNCO, CH3NCO and PhNCO (ED ˆ electron diffraction, MV ˆ
microwave, IR ˆ infrared) [38]

l a X±N, A l NyC, A l CyO, A /XNC a grad /NCO grad Method Reference

HNCO
1.010 a 1.190 1.190 125 b 180.0 b ED [39]
0.987 1.207 1.171 128.1 180.0 b MV,IR [40]
0.986 1.209 1.116 128.0 180.0 b MV [41]
1.060 1.260 1.230 119.2 172.9 PPDP/2 [32]
0.998 1.200 1.180 126.0 175.0 Ab initio [36]
1.036 1.246 1.183 114.6 169.7 Ab initio [35]
1.000 1.233 1.154 128.7 172.0 PPDP/BU [38]
CH3NCO
1.470 b 1.190 1.180 125.0 180.0 b ED [39]
1.440 1.207 b 1.171 b 140.0 180.0 b MV [42]
1.450 1.202 1.168 140.4 180.0 b ED [43]
1.481 1.233 1.187 125.1 169.6 Ab initio [37]
1.381 1.239 1.150 137.0 169.7 PPDP/BU [38]
PhNCO
1.370 1.207 b 1.171 b 140.6 180.0 b MV [38]
1.380 1.250 1.150 109.6 170.1 PPDP/BU [38]
1.447 ± ± 125.1 ± Ab initio [37]
a
For CH3NCO and PhNCO: X Ð the radical carbon atom bound to the nitrogen atom from NCO± group; for HNCO: X ˆ H.
b
Experimental undetermined values, i.e. their magnitude was postulated or taken from another works.

CH3 ±NCO where an angle of rNCO ˆ 169.68 was found [37], a value that is far different from that
expected, i.e. 1808. In the case of HNCO a value of rNCO ˆ 175.08 [36] or 169.78 was found [35].
There is also some disagreement over the phenylisocyanate structure: Bondarenko claims a completely
planar structure for the whole phenylisocyanate molecule (Fig. 2) [38].
This representation contradicts the opinion of Seacher, who considered that this molecule is non-
planar, the NCO group being situated in a separate plane, perpendicular to that of benzene [25].
Also, when comparing Bondarenko's data to the older ab initio calculations [37] (which unfortunately
gives only incomplete information), an agreement with the CS molecule symmetry but a signi®cant
difference for the R±NyC angle and Ph±N bond length is found (Table 1).
One of the few points on which all the authors agree is the existence of a minimum electron density on
the NCO carbon atom. This fact is a determinant for understanding the nucleophilic reactions of
isocyanates.
With regard to the urethane group formation Kozak et al. [44] (based on a quantum energetic
calculations) concluded that the electronic de®cit at the NCO carbon is not suf®cient to promote the
reaction. For example, the authors presented the reaction between Ph±NCO and CH3OH and suggested
that this reaction can take place only as a result of supplementary co-operative affects with other factors,
which will be presented further on.
Returning to the overall electron density distribution in the NCO group, there is also a large disagreement
between researchers. This element should be very important because it should determine the reaction
806 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Table 2
The charge magnitude to the C, N and O atoms in the NCO± group of different isocyanates [38]

qC 2qN 2qO Method of calculations References

HNCO
1.114 0.651 1.076 RMH [28]
0.136 0.050 0.176 MVG [32]
0.430 0.313 0.263 PPDP/2 (opt) [33]
0.688 0.624 0.373 Ab initio [34]
0.550 0.560 0.360 Ab initio (opt) [36]
0.380 0.380 0.210 Ab initio (opt) [35]
0.902 0.772 0.482 PPDP/BU (opt) [38]
CH3NCO
0.527 a 0.234 a 0.290 a PPP [26]
1.340 0.610 1.050 RMH [31]
1.530 0.810 1.090 RMH [29]
1.369 0.564 1.102 RMH [28]
0.866 0.671 0.485 PPDP/BU (opt) [38]
PhNCO
0.486 a 0.204 a 0.270 a PPP [26]
0.446 a 0.075 a 0.347 a PPP [27]
0.873 0.627 0.480 PPDP/BU (opt) [38]
1.329 0.495 1.096 RMH [28]
a
Data from p-electron density.

mechanism by specifying which of the double bonds, CyN or CyO is the ®rst one affected in the initial
stage of the reaction.
Clarifying the symmetry type of the Ph±NCO molecule may also have important consequences in
understanding the reaction mechanism.
So, according to Sacher [25], in the case of a non-planar Ph±NCO structure in which the NCO group
is situated plane perpendicular to that of the benzene ring, the p-orbital system of the whole isocyanate

Fig. 2. Geometry and electron density in phenylisocyanate.


A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 807

group is shifted by interacting with the neighbouring aromatic ring. This should have an important
mechanistic result, indicating that the lone electron pair on the nitrogen that is closer to the benzene ring
should be less available to new bond formation.
However, in the second situation, namely that corresponding to a totally planar structure, a
pronounced asymmetry in the electron density distribution of the whole aromatic ring of the phenyl-
isocyanate should be induced as indicated in Fig. 2, with consequences in its chemical behaviour.
The NCO group deviation from the linearity although apparently of a smaller importance, induces
the possibility of the existence of two planar isomeric forms corresponding to the cis or trans, structure
(Eq. (8)).

…8†

The possible effects of this isomerism on the energetic states belonging to some supposed structures,
that could appear as a result of the interactions between the NCO methyl isocyanate group and the CyC
methyl metacrylate double bond during their copolymerisation, were evaluated by Nizelsky using
quanto-chemical methods [45].
The complexity of understanding the fundamentals of isocyanate chemistry increases when considering
other particular isocyanate structures. A very large series of isocyanate types exists, unexplored until
now in polymer chemistry, in which the radical bonded to the ±NCO group is not an aromatic or aliphatic
hydrocarbonate. Consequently, the electronic structure, and so the reactivity of these isocyanates, can be
strongly in¯uenced by the nature of the alternative radical attached to the nitrogen atom of the NCO
group. These substituents can display more or less electron-withdrawing or -donating capacities. Many
general papers have been dedicated to the isocyanates having an electron-withdrawing group, such as
sulfonyl [46,47], chlorosulfonyl [48], phosphorous [49] and carbonyl [50,51], attached to the isocyanic
group. The same consideration was also given in the opposite case when the presence of an electron-
donating atom such as oxygen, nitrogen or sulphur, modi®es the chemical behaviour of the NCO system.
Reichen [19] presented an interesting review that illustrates the multiple possibilities of these new
classes as of compounds.
Even though the area of isocyanate chemistry is increasing, a great majority of applications still
remains based on the classic addition reactions leading, in general, to the appearance of urethane or
urea groups. Numerous works have been dedicated to the elucidation of the mechanism of these
fundamental reactions. Despite this, many fundamental problems have remained unclear. Later
we will present the main aspects that were taken into account with respect to understanding the
fundamentals of these very much employed processes.

3. Considerations regarding nucleophilicity and basicity

3.1. The nucleophylic reactant XH

As shown before, the most employed isocyanate reactions in polymer chemistry involve a p±s
transformation in which a polyisocyanate partner with the general structure XH interacts with the
808 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

isocyanate group. In all compounds the X±H bonding should be polarised and X must have a
pronounced nucleophilic character. The nature of this compound can be very different such as: NH3,
RNH2, RRNH, H2NOH, H2NNH2, H2O, H2S, H2O2, RNHCONH2, RNHCOOR, RC(yNH)NH2, H2NCN,
ROH, ArOH, RSH, ArSH, PH3, HF, HCL, HBr, HCN, RCOOH, CH2(COOR)2, CH2(COCH3)COOR,
RCH2NO2, ArH, NaHSO3, R2NOH, R3SiOH and ROOH.
The multiple reaction pathways have been presented in some general reviews [14]. From these studies
it can be concluded that the reactivity of XH enhances with X nucleophilicity. Thus the reaction rate of
isocyanates with amines that display a high basicity is always some order of magnitude higher than that
of alcohols. The nucleophilicity of a compound is often evaluated by its basicity because it is easier to be
determined experimentally. From practice it was yet ascertained that this parallelism is not always valid.
The nucleophilicity is a notion inferred from kinetic considerations and re¯ects the af®nity of the
reactant towards the carbon atom: whereas the basicity, which represents the af®nity of the reactant
towards hydrogen, is a notion derived from thermodynamic considerations. In conclusion, the two
notions, nucleophilicity and basicity, do not mix up totally.
Even if, in general, a comparison between these behaviours may be sometimes useful, there are
situations in which contradictions can appear. So, although the phenoxyde ion C6H5O 2 is more basic
than C6H5S 2, the nucleophilicity of the latter is thousand times higher than that of the former.
Even in the ®rst kinetic study on the isocyanate reactivity Backer [52,53] tried to correlate the
alcohol reactivity to the inductive effect, 1I, of the radical bonded with the OH group, in the hope of
®nding a correlation between the inductive effect and the hydroxyl oxygen nucleophilicity. However,
unfortunately, the study was made in dibuthyl ether, a solvent that is able to act as proton acceptor
by hydrogen bonding. Yet the difference between alcohol reactivity observed in this solvent for the
aliphatic alcohol series vanishes when the solvent is replaced by a solvent with a reduced polarity, i.e.
benzene [54].
For the kinetic study, Baker adopted a bimolecular kinetic treatment. He determined enough de®nite
rate constants throughout the entire reaction pathway. A major discrepancy towards the classical
bimolecular mechanisms was established, however, when an unexpected dependence between the
second-order rate constant magnitude and the starting reactant concentration was found [52].
Backer assigned this to two factors: the possible catalytic effects of the alcohol (or of the newly
formed urethane), and the alcohol hydrogen-bonding in¯uence [52±54]. So far other authors have not
reached an agreement.
From the above considerations it may be concluded that it remains dif®cult to establish an a priori
quantitative criterion with regard to the XH reactant nucleophilicity and consequently concerning
reactivity towards the isocyanate group. For the same compound the nucleophilicity, re¯ected by the
reactivity towards the isocyanic group, is not a well-de®ned function; it depends strongly on the possible
association state of the reactants, which is determined in turn by the concentration and by the nature of
the solvent. The reaction capacity of a compound of the XH type can be unambiguously established only
after kinetic measurements, and is valid only for a nucleophile±electrophile reactant pair in the concrete
working conditions.

3.2. The hydrogen bonding

All along many authors tried to predict more accurately the in¯uence of hydrogen bonding on
the reactants reactivity and on the urethane reaction mechanism [20,25,55]. The complexity of the
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 809

hydrogen-bonding systems determined by their high instability and the lack of proper information on
the nature and number of associated alcohol species, however, maintained thus far a high degree of
uncertainty.
From the point of view of macromolecular science, and especially the isocyanate chemistry, the main
problem resides in the elucidation of the hydrogen-bond associated structure, especially in the case of
diols and polyols as they ®nally represent the products involved directly in macromolecular synthesis.
Certainly a similar question was also raised in the case of diamines leading to the urea groups formation,
but the subject would become too large to be presented here. This is the reason why the present review
refers only to the case of the hydroxylic compounds.

3.2.1. Alcohol self-association structure


From IR and cryoscopic measurements it has already been known for years that in unpolar solvents
and at very low concentrations, below 0.01 mol l 21, an alcohol is practically unassociated. Once the
concentration increases the alcohol tends to self-associate. As presented in many monographs [56±58], it
is possible to follow this process employing different physical methods like IR spectroscopy, cryoscopy,
vapour tension, dielectric measurements, heat of dilution and NMR. In parallel with the so-called
ªclassicalº investigation methods, the actual explosive technique development introduces new sophisticated
research techniques mainly oriented towards the dynamic aspects of this phenomenon. The developments
in modern IR spectroscopy, like the measurements in liquid Ar matrix [59], predissociation [60,61], IR
ultrafast dynamic [62] or pulsed IR spectra [63], together with Raman spectroscopy [64,65] could be
mentioned.
In parallel other physical methods were also developed, like the dielectric [66] and dielectric relaxation
[67], acoustic relaxation [68], pulsed spollution neutrons [69], microscopic dynamic by neutron quasi-
elastic scattering spectra [70] and molecular dynamic simulation [71]. If the problem is extended to the
study of the alcohols in solid state or to alcohols that are present at a maximum in the electronic spectra
the number of cited methods is even larger.
Some of these results were analysed comparatively in recent reviews [69,72±76]. From the very many
available pieces of information we used only those which could be related to the isocyanate addition
reactions. We also have taken into account the fact that the addition reactions are generally achieved in
condensed phase, i.e. solution or melt state.
One of the reasons which explains the sensible delay in information transfer from the actual modern
disposable physical methods to the polyurethane synthesis interpretation is the non-existence of co-
operation between different speciality researchers, like those organised under the IUPAC auspices
through the Special Working Party Groups. The variety of synthesis conditions and reactant nature is
very large and it is dif®cult to ®nd multiple disposable sophisticated co-operative data referring to the
one concrete situation. Under the given conditions, for the time being, it is very dif®cult to build up a
better ensemble picture of the problem, even if the recent developments in the structure investigation
techniques could allow that.
In principle even in the early studies it was demonstrated that the alcohol association degree increases
once with the augmentation of concentration. The association process seems to be, however, not a simple
linear step-by-step addition process.
On following the IR spectra evolution versus stoichiometric alcohol concentration (EtOH)5 increase,
besides the initial monomer (EtOH)1 maximum at 3640 cm 21, there appears another small maximum at
3540 cm 21 assigned to the dimer alcohol (EtOH)2 form [21,77]. On increasing the concentration one
810 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 3. IR spectra of ethanol in iso-octane solution. ± ± c ˆ 0:05 mol l21 ; Ð c ˆ 0:1 mol l21 : KBr cell, d ˆ 1 mm:

does not however observe a sudden increase of the 3540 cm 21 maximum, but rather a progressive
increase of another new large maximum between 3100 and 3500 cm 21 attributed to the higher cluster
alcohol associations (EtOH)n (Fig. 3):
Studies of similar systems like ethanol±hexane, by the dielectric polarisation measurement using the
Onsager method [56,78] also evidenced an unexpected course of the process (Fig. 4).
At very low alcohol concentrations, below 0.01 mol l 21, corresponding to the monomer and linear

Fig. 4. Dielectric polarisation p1 of ethanol in hexane versus c1 ˆ ethanol molar fraction.


A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 811

Fig. 5. Cryoscopic average self-association degree f of ethanol in benzene ( p ) at 5.58C and cyclohexane (1) at 6.58C versus
concentration; d ˆ maximum deviation from linearity.

dimer forms, the molecular polarisation dipole moment p 0 1 tends to equal the corresponding value
observed in the gaseous state. In a ®rst step, once the alcohol concentration increases the polarisation
dipole moment value decreases indicating a tendency of appearance of low polar associated forms which
can be attributed to some cyclic structures. If the ethanol concentration increases over the limit of ,0.05
molar fraction corresponding to ,0.4 mol l 21 the polarisation factor p 0 1 begins to rise progressively,
indicating a gradual increase of the number of more polar ethanol associates. A similar behaviour was
also observed in the case of 1-butanol [79].
Recent measurements of the average alcohol association degree function f based on a cryoscopic-NMR
combined method [80] con®rmed the high tendency of EtOH to form trimers (EtOH)3 and tetramers (EtOH)4
with possible cyclic structures. The alcohol concentration region on which this tendency is higher is
characterised by a maximum deviation from the linearity of the curves observed in Fig. 5.
Considering the case of the most unpolar studied system, i.e. ethanol±cyclohexane, presented in Fig.
5, the range of this concentration region of ,0.4 mol l 21 is the same as that found from dielectric
measurement in the case of ethanol±hexanol system, Fig. 4.
There could be some criticism concerning the validity of the laws built on the fundamentals of these
methods, especially at high alcohol concentration, i.e. the limit of the Debye±Onsager equation validity
(whose theoretical fundament was deduced from gaseous state) and the validity limit of the Raoul law at
higher alcohol concentration. Yet, taking into account that our discussion deals mainly with the low
alcohol concentration solutions in the most unpolar solvents, the validity of these laws should have a
higher applicability [81].
812 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 6. Dependence of the chemical shift d (ppm) (TMS) of ethanol OH proton, versus average association degree f in
cyclohexane (1) and benzene ( p ) at freezing point of solutions.

It is possible to accept even that the cryoscopically determined f value calculated with the aid of
the Beckman equation [80] does not represent exactly, at higher alcohol concentrations, the alcohol
associated average molecular weight; in any case it represents a proportional function of it. Thus this
value can give valuable information on the alcohol association state in solution, and offers a possibility
to compare, at least qualitatively, two different systems, i.e. EtOH±benzene and EtOH±cyclohexane,
offering reasonably different conditions for the hydrogen-bond capacity formation. The unequivocal
meaning of co-determined f value was appreciated by comparing the NMR chemical shift d value in
different systems at different concentrations. Thus, it was remarkable to ®nd that both different pair
series of values f and d characteristic to different solutions of EtOH in benzene or cyclohexane fall on the
same curve [80] (Fig. 6).
The corroboration of all the above-mentioned results leads to the conclusion that in the range of 0.01±
0.1 mol dm 23 alcohol concentration and 20±708C, the alcohol proves to have a higher tendency to
form trimers (EtOH)3 [21,60,66,77,79,83], tetramers (EtOH)4 [60,84±88] and pentamers (EtOH)5 [85]
clusters.
Basing on these conclusions many authors are tempted to reduce the reaction mechanism representation
by taking into account only two alcohol hydrogen-bonded equilibrium forms, i.e. monomer±trimer [77]
and monomer±tetramer alcohol association equilibrium forms [89].
Even if theoretical models of such species pairs in equilibrium in good agreement with the
experimental data regarding some physical cumulative properties like vapour pressure or heat of mixing,
etc. were possible [89], it should not be forgotten that these association forms coexist with other
molecular forms like dimers and higher polymer clusters [66,80,82].
With regard to the co-operative effects which could appear in the case of the mixtures of many
structural equilibrium forms, we should note that the elucidation of the questions related to the chemical
reactivity and the reaction mechanism differs considerably from that of a physical partition function
determination. In this case the solution cannot be displayed simply by calculating an average value,
when the participation of the small concentration forms are practically neglected.
In the case of the reaction mechanism study, if we take into account a possible high difference between
the chemical reactivity of different association forms [90] and the possibility of a very rapid equilibrium
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 813

Fig. 7. Methanol dimer structure.

restoration, we cannot exclude any possible association form, from the reaction mechanisms, even if
such forms were present in the reaction mixture only in a small concentration.
Moreover, in the macromolecular synthesis the working conditions are sometimes different from that
presented in the previous physical±chemical studies. So there are numerous cases when the polyaddition
process proceeds without solvents, using neat glycols, and are performed in larger temperature range, i.e.
220±1208C.
Under the conditions the necessity to possess a better image of the entire range of alcohol clusters is
evident.
The complexity of the situation is evidenced even in reference to only one single self-association type, i.e.
the alcohol dimer. The multitude of the observed isomers is re¯ected, for example, in the results obtained by
Shriner et al. [91] and Schrems [92]. Thus, though a linear structure is generally admitted for the alcohol
dimer [83], the N2 matrix isolation measurements evidenced the existence of four proton-donor bands
corresponding to four proton-acceptor bands in the case of methanol. The quadruplet structure is explained
by the coexistence of the different con®gurations in equilibrium with the dimer [83,91±94] (Fig. 7).
As regards the trimer and tetramer structure even though most of the authors choose the classical
cyclic representation [60,66,79,83,84,88,95], recent studies consider these associates as dynamic spatial
structures in lattice-¯uid associated solution [96], H-bonding lattice-¯uid [97], inde®nite forms in which
there appear bifurcated and trifurcated bonds [65] or the so-called bidental bonds [98] in which the O and
H atoms can adopt transient forms participating concomitantly into two hydrogen bonds as part of a
three-dimensional dynamic network [99].
For highly unpolar solvents an amphiphilic structure of the cluster can be also supposed within which
the associates adopt a micelle conformation (Fig. 8). Thus, the OH groups are located in the centre of
these aggregates facilitating a rapid and continuous proton interchange. The hydrocarbonate alcohol
remainder is directed towards the exterior of micelle, i.e. towards the unpolar solvent.
Concomitantly it also performs a permanent exchange of individual alcohol molecules between
micelles or with other monomer alcohol from the solution.
In the efforts to correlate the hydrogen-bonding structure of the alcohol to their chemical reactivity
towards the isocyanates, Satchell et al. [21,27], based on some kinetic measurements, reduced the
mechanism of urethane formation to only a isocyanate±cyclic trimer alcohol interaction. The reason
for this correlation will be discussed in a later section.
Starting from the experimentally obtained Satchel's data [77] we tried to corroborate the average
814 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 8. Alchyl alcohol amphiphilic structure in unpolar solvents.

association degree f function determined by us [80] for the same system with the evolution of the
ratio between alcohol monomer concentration towards the total alcohol concentration, i.e. the
(EtOH)1/(EtOH)s value, Fig. 9.
Concomitantly on the same graph is represented the evolution of the molecular polarisation p 0 1 taken
from the ®rst part of the curve represented in Fig. 4. Under the reasonable supposition of the similitude
between alcohol hydrogen-bonding state in the following unpolar solvents: isooctane, heptane and
cyclohexane, it can be concluded that the minimum of polarisation, p 0 1 reaches an (EtOH)s concentration
of ,0.4 mol l 21 and can be correlated to the majority formation of the cyclic associate structure. It must
be noticed that even at this concentration there still exists an appreciable quantity of monomeric alcohol
(,45%). Following the effect of a further alcohol concentration increase of the f and p 0 1 values is

Fig. 9. Some characteristics of ethanol solutions at different concentrations: % ˆ monomeric ethanol from alcohol stoechio-
metric concentration (iso-octane 24.58C); p1 ˆ dielectric polarisation (hexane 308C) and f ˆ average self-association degree
(cyclohexane 258).
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 815

observed, as expected, a decrease of the monomer fraction (EtOH)1/(EtOH)s, whereas both p 0 1 and f
functions begin to augment. This fact appears to be in disagreement with Satchel's hypothesis of the
preeminent unpolar cyclic trimer formation.
Thus it is also necessary to admit that once the alcohol stoichiometric concentration (EtOH)s increases
there appear gradually some larger and more polar alcohol cluster forms [83].
This phenomenon should be emphasised at lower temperatures (,68C) when the association degree
enhances appreciably [80]. Under the conditions the appearance of the linear or quasilinear associate
forms is favoured [66,79,99] in transition to the ªin®nite linear alcohol self-associationº, characteristic
for the crystalline state [100]. Wong and Drago [99] hypothesised that in the inert solvent alcohols there
may also appear behaviours which even does not exist in the neat alcohol state.

3.2.2. Dynamic and quantum aspects of hydrogen bonding


Besides the physical methods some authors undertake the problem of the hydrogen bonding from the
quantum point of view. Different orbital molecular calculations were applied [101±104], yet unfortunately
at the present time the ab initio calculations are insuf®ciently developed. So there appears a reasonable
discrepancy between some of the calculated and experimentally determined spectral data [105].
In most modern reasoning regarding the hydrogen-bonding phenomenon, the main accent was shifted
from the old global equilibrium characterisation, achieved in the case of some particular association
transient structures forms, towards the dynamics of the process [96,106±109].
Panayiotou et al. [96]attempted to develop an uni®ed statistical thermodynamic model for hydrogen-
bonded ¯uids and presented an interesting equation of state approach. The basic assumption is the
division of intermolecular forces into physical (van der Waals) and chemical (hydrogen-bonding) forces.
A general formalism is presented as available for multicomponent systems of molecules with any
number of hydrogen (proton) donor and acceptor groups. The model is also applied to self-associated
substances and may be used to describe the thermodynamic behaviour of these systems. The equation of
state approach was veri®ed and compared to the experimental values for the vapour pressure, speci®c
volume and heat of mixing and good results were obtained.
It was pointed out that the main aspect to be considered is the equilibrium between the formation and
rupture of every hydrogen bond and not the ensemble equilibrium between various associations. Veytsman
[110] considers the systems of molecules with one type of proton-donor group and one type of proton-
acceptor group and proposes an approximate combinatorial expression for the number or ways to form
hydrogen bonds that does not invoke the existence of associates.
An illustrative picture of the dynamic of this process illustrated by the isotopic hydrogen±deuterium
exchange for different alcohols is presented in Table 3 [111].

3.2.3. The hydrogen bonding in glycols


As already known, in the macromolecular synthesis glycols represent one of the most important raw
materials well illustrated by polyurethane chemistry. Like the usual alcohol, the glycols display a high
tendency of self-association but in this case it is possible that intramolecular hydrogen bonding also
appear, depending on the molecular structure facilities.
Robertson and Stutchbury [112] draw attention to the importance of this kind of association and
inferred it for some observed kinetic misunderstandings.
By studying the IR glycol spectra, Kuhn [113] observed that in diluted unpolar solvents a series of
diols presents two n OH bands, while others present only one. The higher frequency band at 3630±3644 cm 21
816 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Table 3
Comparison of proton transfer times and tautomer lifetimes in several ROH/ROD solvent pairs (293 K). Reprinted with
permission from J Phys Chem 1991;95:10359. q 1991 American Chemical Society [111]

Solvent a Proton transfer times b Tautomer lifetimes b

t pt (H) t pt (D) r pt (D/H) t ¯ (H) t ¯ (D) r ¯ (D/H)

MeOH/MeOD 124 368 3.0 663 883 1.3


EtOH/EtOD 173 471 2.7 771 1021 1.3
2-PrOH/2-PrOD c 234 (990) (4) 902 (1300) (1.4)
1-BuOH/1-BuOD 213 520 2.4 966 1264 1.3
(CH2OH)2/(CH2OD)2 333 691 2.1 749 1040 1.4
a
The solvents are MeOH ˆ methanol, EtOH ˆ ethanol, 2-PrOH ˆ 2-propanol, 1-BuOH ˆ 1-butanol and (CH2OH)2 ˆ
ethylene glycol. Deuterated solvents were substituted only at the hydroxyl positions.
b
Proton transfer times and tautomer lifetimes (ps) were taken to be the rise and decay times measured at 550 nm. The
numbers reported are averages of 2±3 independent measurements. The standard deviation in repeated measurements was
^15 ps. The r (D/H) are the ratios t (D)/t (H).
c
The deuterated sample of 2-propanol contained signi®cant amounts of a ¯uorescent impurity and these values should be
regarded as subject to larger uncertainties than the rest.

was attributed to the free n OH and the lower 3478±3590 cm21 to the intramolecular band. At higher glycol
concentrations a third band appears at 3185±3415 cm 21 attributed to intermolecular associations. The
difference Dn OH between the free n OH and that of intramolecular bands n OH ass represents a measure of the
hydrogen-bond force. Kuhn et al. [114] suggested that in 1,4-diols there exist a higher chain ¯exibility and
the intramolecular hydrogen bonding becomes very strong. This fact is characterised by a high n OH value
of 100±160 cm 21. The steric factors which determine the intramolecular hydrogen-bonding force are, on
the one hand, the possibility of the reciprocal approaching of the two glycol OH groups below the limit of
3.3 AÊ and, on the other hand, the facility of the O±H´´´O structure to adopt a linear form.
The hydrogen-bond strength in diols varies in the following order [115]:
1; 4-diols . 1; 3-diols . 1; 2- and 2; 3-diols:
A correlation between this situation and the reactivity towards phenylisocyanate was made recently
by Caraculacu et al. [116].
The high reactivity of 1,4-buthanediol (BD) in the ®rst step of the reaction, when it reacts only with
a single OH group, was explained by the existence of an enhanced nucleophilicity of the oxygen
from the OH proton-donor group which participates in the intramolecular hydrogen-bonding formation.
The nucleophilicity enhancement is determined by the electron displacement which was calculated
quantitatively in other similar hydrogen-bond systems [117].
Recently, the problem of the glycol intramolecular bonding was considered again by Park and Tasumi
[105] who reinvestigated the IR-induced conformational isomerisation of 1,2-ethane diol in low-
temperature Ar matrices parallel with the reverse reaction in the dark. They used new experimental
facilities and a power superior Fourier transform spectrometer. There were results observed in the region
of 3700±3600, 3000±2870, 1500±1400, 1180±1150 and 1110±1030 cm 21.
According to the behaviour under irradiation, it was possible to make a classi®cation of the different
types of intramolecular interaction. The molecular structure of the 1,2-ethanediol conformers are
depicted in Fig. 10.
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 817

Fig. 10. Six conformers of 1,2-ethanediol and respective IR OH-stretching band.

The relative energy corresponding to the six conformers was obtained from the ab initio calculation
Table 4 [105].
The authors showed that the stretching bands of the intramolecular hydrogen-bonded OH are found at
3624.1 cm 21 (t G 1 g 2) and 3634.2 cm 21 (g 1 G 1 g 2) whereas the corresponding free OH stretching
bands are located at, respectively, 3663.2 cm 21 (t G 1 g 2) and 3665.1 cm 21 (g 1 G 1 g 2).

Table 4
Relative energies for six conformers of 1,2-ethanediol obtained from ab initio calculations of different levels. Reprinted with
permission from J Phys Chem 1991;95:2757. q 1991 American Chemical Society [105]

Ab initio level Relative energy (kJ mol 21)

t G1 g2 g1 G1 g2 g2 G1 g2 tTt t T g2 g2 T g2

4-21 G a 0.00 3.68 9.54 10.67 13.35 14.02


6-31 G p 0.00 2.69 5.37 8.47 9.87 10.19
MP2/6-31 G pp 0.00 0.83 5.08 13.50 13.59 12.51
a
Reference: Van Alsenvy C, Van der Enden L, Schafer L. J Mol Struct. 1984; 108:121±128.
818 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

The fact that the hydrogen-bonded band t G 1 g 2 is located about 10 cm 21 lower than the band of the
hydrogen-bonded g 1 G 1 g 2 indicates that the hydrogen bond corresponding to the t G 1 g 2 structure is
slightly stronger than the other.
Since the ®fth conformer IR bands are generally weaker than that of the others it seems reasonable to
assign these types of band to the conformation g 2 G 1 g which is less stable. According to the MP 2/6-31
G pp calculation the g 2 G 1 g conformation should be 5.08 KJ mol 21 energetically less stable than the t
G 1 g 1 type.
The question regarding the possible simultaneous existence of two intramolecular hydrogen bonds in
this conformer still remains unclear for the time being [105].
The results of the kinetic measurements in the case of different glycol±phenylisocyanate reactions
furnished by Caraculacu et al. [116] con®rmed the supposed correlation between the strength of
hydrogen bonding expressed by Dn OH and the chemical reactivity. Thus the reaction rate constant of
1,2-ethanediol which presents a Dn OH of 31.1±39.1 cm 21 is about 15 times lower than that of the ®rst
rate constant of BD when Dn OH presents a value of 156 cm 21.
In conclusion, for the reaction rate evaluation it is not enough to know whether the reactive OH groups
are involved in any hydrogen bonding, it is also necessary to have some information on the strength of
this bonding.
Unfortunately, in the most illustrative case from the glycols series represented by BD recent detailed
information is not available like in the case of 1,2-ethanediol.
Another principal problem which remained unclear so far is the in¯uence of the hydrogen-bond
associate form, i.e. cyclic or linear, on the chemical reactivity. Such factors should affect, respectively,
the molecular polarisation and the reactivity of the species.

4. Pattern used to elucidate the mechanism of uncatalysed alcohol addition to isocyanate

4.1. Quantum chemical methods

Starting from 1973 Tiger et al. [118] and many others authors [38,90,119,120], tried to introduce
quantum chemical methods in the efforts to elucidate the mechanism of urethane reaction.
The advantage of this way resides in the fact that by using this method the whole system of electro-
phile and nucleophile partners is assessed as a unit in the interaction, in contrast to the older disparate
information regarding only the individual reactants' activity in the absence of the reaction context.
In comparison to other methods like those of the kinetics, the quantum chemical methods start
generally from different presumptive patterns of reactions and offer a possibility to decide which of
the supposed mechanisms is more probable. As a consequence this modality of treatment will be
advantageous only if the real unknown mechanism is present between the different hypothetical
alternatives that are considered.
The precision of the calculations also depends on the possibility of performing a good optimisation,
which in turn needs available information on the geometry of molecules, data which unfortunately often
fails.
Even if at present this method has many imperfections, its perspectives for the future are very large.
No ®nal decision regarding any mechanism deduced through other methods would be possible without a
subsequent quantum chemical veri®cation.
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 819

In order to clarify the problem of hydrogen-bond in¯uence on the alcohol nucleophilicity, the
extended Huckel method [118] was used as a ®rst step. The quantum chemical calculations of the
electron density in the linear methanol dimer shows that with the hydrogen-bond formation, the electro-
negative charge on the oxygen atom from the proton-donor OH group grows, which determines an
enhance of the alcohol reactivity in the nucleophile addition [118].
By using the PPDP/BU method combined with the geometry optimisation, Bondarenko et al. [38]
perform calculations for a series of possible donor±acceptor complexes in the case of two different
isocyanates, i.e. HNCO and CH3NCO, when they interact with some electron donor partners like
CH3OH, H2O, (CH3)2O and C5H5N. Such complexes should play an important role in the nucleophile
addition to the NCO group. It was found that with the appearance of this complex the valence distance
NyC of the NCO group enhances more than that of the corresponding CyO one. Contrary to the
situation presented earlier for the isolated RNCO molecules, the NCO angle in the complexes becomes
equal to 135±1508, close to that corresponding to sp 2 hybridisation characteristic of the ®nal urethane
group.
A similar effect regarding the alcohol activation by hydrogen bonding was found by other authors
when using a quasiempirical CNDO/BW method [90].
The energetic state difference calculations regarding the transition state are also illustrative compared
to possible patterns of the reaction between methyl isocyanate and methanol. In these patterns the methyl
alcohol was regarded either in its monomeric or in its dimeric form, Fig. 11 [90].
Thus the higher reaction capacity of alcohol dimeric form was once more con®rmed.
A new trend of research regarding the isocyanate±alcohol reaction based on quantum chemical
methods was achieved when Kozak et al. [119,120] proposed the use of frontier molecular orbital
(FMO) energy determinations. By applying the PPDP/2 method [120] the authors studied the
reaction capacity of different structural forms of the dimer alcohol to the NyC bound to some iso-
cyanates. Based on an MO combined qualitative analysis the FMO energy as well as the reaction
capacity index (RCI) were calculated using Klopman's formula [121]. Starting from the previous
surprising conclusion which claims the impossibility of the existence of a reaction between the
monomeric form of the CH3OH and PhNCO [119] the authors considered that the necessary condition
for this reaction is that the methanol should adopt a structure which should posses an active electron
donor centre. For the RCI determination in the case of the interaction between an isocyanate and an
alcohol (or of different alcohol associated forms) the following energetic criterion was used [120],
(Eq. (9)):
D A A D
uEHOMO 2 ELUMO u , uEHOMO 2 ELUMO u …9†
D A A D
where: EHOMO ; ELUMO ; EHOMO and ELUMO are the energies of highest occupied and lowest unoccupied
orbital of probable donor D and probable acceptor A.
The electron donor capacity value (EDA) was also calculated with the aid of Eq. (10):
A D D A
EDA ˆ uEHOMO 2 ELUMO u 2 uEHOMO 2 ELUMO u …10†

where the positive value of EDA corresponds to the electron-donative nature of the OH-containing
component in the alcohol±isocyanate interaction, and the negative one corresponds to an electron±
acceptor character.
820 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 11. The hydrogen-bond role and energetic effects in methyl isocyanate±methyl alcohol interaction.

The following alcohol dimeric form was taken into account (Eq. (11)):

…11†

The results are presented in Table 5 [120].


Based on the data presented in Table 5 the authors concluded that the alcohol dimerisation process
enhances its electron donor capacity towards the isocyanates re¯ected by any of the mentioned values of
EDA. Relative to the nucleophile opening of the bound NyC the different methanol dimer forms display
different reaction capacities. In the reaction with alkyl isocyanates, the reactivity of methanol decreases
with the different RNCO substituents as HNCO . CH3NCO . C2H6NCO. The reaction of CH3OH with
AlkNCO is conformational independent, but in the case of PhNCO the conformation of the molecule can
play an essential role. The donor±acceptor roles in the alcohol±isocyanate pair may even change on the
reverse order. The authors concluded also that the change in ROH reactivity as a result of dimerisation
has the same tendency as the in¯uence of an electron donor catalysator on the monomeric alcohol [120].
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 821

Table 5
FMO energy values and electron donor capacity (EDA) p of molecules and molecular systems [121]

Molecule FMO energy Molecule FMO energy

2 EHOMO (eV) ELUMO (eV) 2 EHOMO (eV) ELUMO (eV)

HNCO 13.98 3.29 CH3OH 15.14 7.05


CH3NCO 12.89 3.83 (CH3OH)2 trans 14.35 6.29
C2H5NCO 12.66 4.05 (CH3OH)2 cycl 14.91 6.96
PhNCO unplane 12.05 3.03 (CH3OH)2 cis 14.43 6.52
PhNCO plane 11.46 3.34 CH3OH 9.00 4.50
Probable acceptor A EDA (eV)

Probable donor D

CH3OH (CH3OH)2 (CH3OH)2 (CH3OH)2 CH3OH

HNCO 2.60 2.63 2.66 2.78 6.19


CH3NCO 0.97 1.09 1.11 1.12 4.56
C2H5NCO 0.52 0.55 0.66 0.70 4.11
PhNCO unplane 0.90 0.93 1.04 1.08 4.49
PhNCO plane 20.04 0.06 0.17 0.21 3.62

At present though extensive computer development has determined an enormous interest towards the
orbital computation methods, the results must be regarded with prudence. All the present quantum
chemical methods [120] involve some approximations. Using different MO calculations frequently
leads to contradictory results. This is the reason why the MO calculations can serve mostly for
new hypotheses building. The obtained results should be always checked by using an appropriate
experimental method.

4.2. The kinetic methods

The kinetic measurements were the ®rst methods used for the study of the isocyanate±alcohol addition
mechanism, and thus far they remain the most utilised methods. Starting from 1949 with the fundamental,
extensive work of Backer [52±54,122] there were numerous studies published concerning the
kinetics of isocyanate and diisocyanate addition, both with alcohol and glycols. The systematisation
of the principal results was made by Frish and Saunders [22] and though their monographs are old,
they have remained quite useful till date. From the beginning, the second-order kinetics treatment
was used by most of the authors despite the lack of concordance between the requirements of the
second-order mechanism and the observed reaction rate constant dependence on the initial reactants
concentration.
Many authors tried to obtain a concluding correlation between the reaction rate and the alcohol
association state; however, until now no decisive satisfying mechanism of reaction has been obtained.
In concordance with the macromolecular chemistry requirements, when bifunctional reaction partners
are employed, usually, it appears necessary to perform preliminary measurements regarding the individual
reaction rate constants determinations related to the two reaction steps of these bifunctional partners.
822 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Other studies based on some assumed approximations were directed towards the kinetic measurements
regarding the whole macromolecular polyaddition synthesis system. Parallel researches were directed
towards the problem of the secondary reaction which can accompany the polyaddition process. All these
questions will be brie¯y presented later in this review.
With regard to the fundamental elemental addition reaction, majority of the authors agreed with the
general presumption that the uncatalysed urethane reaction performs apparently by a bimolecular
mechanism [22,52±54,80,116,123±129], even if some particularities should be considered.
One of the most important problems which remains unclear till now is the nature of the favourite
reactive form of alcohol associate for the reaction. The difference between the reactivity of the different
types of alcohol associate is evidenced better in the case of the alcohol±isocyanate reactions performed
in the unpolar media [21,55,77,80]. Though this problem apparently regards only some theoretical
aspects, modern macromolecular chemistry cannot overlook this question because the multiple ways
of polyurethane synthesis effected in solution or melt involve certainly different types of hydrogen-
bonded association of the hydroxyl containing partner [116]. Obviously the molecule segment alternation
order in the ®nal macromolecular chain is controlled by the reactivity of every partner and can have
repercussions on the physical±mechanical properties of the polymer.
In order to provide new explanation to the role of different types of hydrogen-bond associate in the
urethane reactions, two new ways were undertaken in the previous attempt. The ®rst one deals with
kinetic studies of the urethane reaction performed under unperturbated hydrogen-bond association
equilibrium conditions. In the second correlations between the kinetic parameters: (a) the average
hydrogen-bond alcohol self-association value; (b) the simultaneous presence of two different hydroxy
compounds in the reaction capable of forming mixed associations; and ®nally (c) the effect of the
intramolecular hydrogen bond in presence exhibited by some glycols were investigated.

4.2.1. The kinetic studies of the urethane reaction performed under unperturbated hydrogen-bond
association equilibrium
Satchell et al. [21,77] proposed an interesting method in which a very small quantity of reactive p-
nitrophenylisocyanate was reacted with a large excess of ethanol in different alcohol concentration
solutions so to achieve constantly a pseudo-®rst-order reaction condition. This way, the associate alcohol
equilibrium state remains practically unchanged during the reaction.
The reaction was performed in a very unpolar solvent, i.e. isooctane at different temperatures. The
association state of alcohol was followed by the IR measurements. Even though the authors identi®ed
maxima for monomer at 3640 cm 21, for the dimer at 3540 cm 21 and for the polymer at 3100±
3500 cm 21, it was impossible to calculate the individual association concentrations owing to the large
difference between the IR molar extinction coef®cients corresponding to different alcohol associate
types or to the monomeric alcohol. The only quantitative parameter of association state was the intensity
of the maximum IR at 3640 cm 21, which the authors supposed belonged only to the alcohol monomeric
form. In the mentioned works calculated monomer alcohol concentration values are given only for the
alcohol solutions at 24.58C, Fig. 9.
Using the observed ®rst-order reaction rate Kobs constant and both the stoichiometric (EtOH)s and
monomeric (EtOH)1 alcohol concentrations, the authors found some interesting dependencies.
The Kobs graph/[(EtOH)1] against [(EtOH)1] 2 at 24.58C presented by the authors is reasonably
rectilinear over the entire range of [(EtOH)s] values, Fig. 12 [77].
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 823

Fig. 12. Plot of Eq. (13) in the interaction of p-nitrophenyl isocyanate and ethanol in iso-octane at 258C.

Starting from this result the authors suggested a reaction mechanism which is re¯ected by Eq. (12)
kM
3…EtOH†1 …EtOH†3 …EtOH†1 1 NO2 C6 H4 NCO ! NO2 C6 H4 NHCO2 Et
slow

kT
…EtOH†3 1 NO2 C6 H4 NCO ! NO2 C6 H4 NHCO2 Et 1 2…EtOH†1 (12)
slow

The reaction rate equation corresponding to this mechanism is given by relation (13):
Kobs ˆ kM ‰…EtOH†1 Š 1 kt Kt ‰…EtOH†1 Š3 …13†
In agreement with other researchers [80], the authors believed that the monomer is probably a 10 3
order of magnitude less reactive than the trimer. In conclusion the authors postulated as a main way of
the urethane reaction a cyclic transition state starting from the alcohol trimer associate under the
assumption that the intermediary eight-membered hydrogen-bonded rings are the most stable (Eq. (14)).

…14†

As observed, all those considerations are based only on the rectilinearity found by the authors in the
824 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 13. Plot of kobs versus molar ethanol stoichiometric concentration in the p-nitrophenylisocyanate and ethanol excess
reaction; iso-octane, 24.58C. (experimental data were taken from Ref. [77]).

Kobs graph/[(EtOH)1] against [(EtOH)1] 2, (Fig. 12). Unfortunately a similar rectilinear dependence
could be also found if we try to draw a new graph, Kobs against [(EtOH)s], using the same data presented
by the authors (Fig. 13). The consequence of this important second dependence is not however discussed
in the mentioned work.
Indubitably the new unperturbated hydrogen-bond equilibrium method proposed by Satchel et al. [20]
will represent, in future, a valuable way to study the isocyanate±alcohol reaction, but before than that
some important clari®cations should be done:
The ®rst observation regards the problem of establishing to what extent the calculations based on the
intensity of the IR n OH monomer band at 3640 cm 21 could be altered by the existence of a superposition
with the absorption band of the linear associate OH end groups (n OH end), which is located in the same
position [83,130,131]. The authors themselves admitted the presence of the alcohol dimeric form in
some proportions, which is known to exhibit two IR maxima, one of which being situated exactly in the
incriminated n OH monomer absorption zone [83,130,131]. During the exponential correlation, as that
presented by Eq. (13), small errors of concentrations can lead to important deviations [132].
The second observation refers to the fact that in their latest publications the authors omitted from
discussion other high-reactive alcohol associated species, like the linear dimer, in spite of the fact that
the high reactivity of such species was evidenced by the orbital molecular calculation [90,120] and
con®rmed by the kinetics studies effected on different glycols which present more or less intramolecular
hydrogen bonds (similar to the dimeric alcohol structure) [116]. Taking into account the extremely rapid
equilibrium reformation, the dimer route alcohol±isocyanate reaction cannot be easily rejected. The
simple trimer alcohol±isocyanate mechanism proposed by the authors gives no explanations on the
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 825

Fig. 14. Dependence of reaction rate constant k (dm 3 mol 21 min 21) function on initial EtOH concentration for the EtOH±
PhNCO reaction in cyclohexane (1) and benzene ( p ). (CEtOH ˆ 2CPhNCO) at 608C.

observed high reactivity of BD in its ®rst step of reaction [116]. Also this mechanism could not explain
the indubitable catalytic effect of the formed urethane [80].
The authors themselves admitted other explanations for the observed third-order term dependence in
[(ROH)1]. Such a dependence could be rising from a fast pre-equilibrium solvation of the isocyanate
produced by two monomeric alcohol molecules, followed by a slow nucleophilic attack of another
monomer [21].

4.2.2. The kinetic correlation with the alcohol average self-association degree mixed alcohol
associations and polyols presenting intramolecular bonding
Since 1990, Caraculacu et al. [133] have initiated a new series of research using the HPLC method
in order to follow the evolution of reaction composition in the real complex conditions of urethane
synthesis. The method is very laborious and needs numerous model substances, yet ®nally it is possible
to obtain valuable information regarding the evolution of the reaction mixture's whole composition. By
using this method it is also possible to evidence eventual synergetic or inhibiting effects when, as part of
the reaction, different types of interaction like that of alcohol±alcohol, alcohol±glycol, alcohol±amine
and alcohol±urethane can appear. The in¯uence of such interactions could not be evidenced directly
until now by using the older methods.
In the ®rst step of these researches, the second-order reaction rate constant dependence versus the
initial stoechiometric alcohol concentration [EtOH]s and versus the alcohol average self-association
degree function f were followed.
The f values were determined by means of the 1H NMR-OH chemical shifts in the (d OH) value in
connection with a cryoscopically standardised method. This method was applied in the case of the
reaction between PhNCO and ethanol performed in two different solvents cyclohexane (Cy) or benzene
(Bz) [80]. The results presented in Fig. 14 re¯ect the mode in which the initial alcohol concentration
in¯uences the reaction rate in both the solvents. As mentioned earlier, with the alcohol concentration
increase the alcohol reactivity also enhances. The reactivity of different alcohol associates is always
higher than that of the monomeric alcohol. As shown in Fig. 14 the nature of the solvent in¯uences a
noteworthy gradual enhancement of the alcohol association degree.
For the same [EtOH]s a signi®cant difference was found between the reaction rate constants, k, in Bz
and Cy (Fig. 14 [80]). An excellent concordance was found for both situations, Bz and Cy, when the
826 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 15. Dependence of reaction rate constant k (dm 3 mol 21 min 21) function on association degree f for EtOH±PhNCO reaction
in cyclohexane (1) and benzene ( p ). (CEtOH ˆ 2CPhNCO) at 608C.

corresponding reaction rate constant k was represented versus the average alcohol association degree
function f.
As seen in Fig. 15 the points determined by any k and f pair values are independent of the nature of the
reaction solvent and consequently the k±f dependence is represented by a unique curve. The extrapola-
tion of the k value when f ˆ 1 leads to k ˆ 0 in the both cases, which con®rms the Satchel [77]
hypothesis of the very low monomer alcohol reactivity, Fig. 15 [80].
Once the association enhances, the reaction rate constant increases visibly till f ˆ 2±2:5 and then
arises slowly to a constant limit. The validity of the kinetic second-order reaction treatment was
con®rmed for a large degree of conversions by the existence of a quasi rectiliniarity of the graph in
FNexp% versus theRreduced time value d (where Nexp% is the experimentally determined unreacted alcohol
fraction and d ˆ t0 ‰…EtOH†s Šdt†; Fig. 16 [80].
A similar dependence was also established in the case in which one of the monofunctional reactants
was replaced by a difunctional one. This is the case of the reactions between various diisocyanates and 1-
butanol [133±135] or that of the reactions between phenylisocyanate and different glycols performed in
various solvents like Bz, 1,4-dioxane (Dx) and N,N-dimethylformamide (DMF) solution [116].
As already known from the classical kinetic theory, in the case of a second-order reaction, the rate
constant value must not be dependent on the initial reactant concentration. The real existence of such a
dependence (Fig. 16) was attributed to the change of the initial alcohol association state characterised in
every particular case by a de®nite reactivity. The relative constancy of the second-order reaction rate
constant during the whole reaction pathway, when the alcohol concentration decreases gradually, can be
explained only when admitting the existence, in the speci®c molecular association state characteristic of
an initial [EtOH]s, of a compensatory effect between the alcohol which vanishes and the appearing
urethane.
Indeed the activating effect produced by an initial addition of some quantities of urethane in the
reaction is similar to that produced by the equivalent [EtOH]s increase of the initial alcohol concentration.
Under the conditions the overall initial alcohol association states seem to be equivalent and independent
if the alcohol molecules are associated to another alcohol molecule or to that of urethane (Table 6, No. 6
[80]).
An important remark which was also made in the previously mentioned works is that the urethane
catalytic effect is visible only when the hydrogen is present on the urethane group. Its substitution by an
alkyl radical leads to the disappearance of the catalytic effect. Thus the replacement of the additive
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 827

Fig. 16. Dependence of ln FNexp (%) function on d (mol min dm 23) for EtOH reaction with PhNCO, in cyclohexane (1) and
benzene ( p ), (CEtOH ˆ 2CPhNCO) at 608C, at different alcohol concentrations (mol dm 23): 1, 0.15; 2, 0.175; 3, 0.0782; 4, 0.175;
5, 0.625; 6, 1.25; 7, 2.0; 8, 0.625; 9, 2.5; 10, 0.4; 11, 5.0; 12, 1.25.

carbanilic acid ethyl ester (CEE) with N-methyl carbanilic acid ethyl ester (MCEE), which do not posses
more H on the urethane group, produces an inhibition effect on the reaction (Table 6, No. 6 [80]).
Obviously even if the N-substituted introduced urethane is probably inert towards the isocyanate it
can certainly participate in the hydrogen bonding but only as an H acceptor, more probably at the CyO
group, producing an OH-group reactivity deactivation like that determined by the ether groups.
From the above observations it may be concluded that for the urethane reaction rate enhancement it
is not suf®cient that the alcoholic OH group should be included in a hydrogen-bonded associate.
Surprisingly, it was found that it is also necessary to take into account the total number of labile protons
from the system capable of being involved in these associates whether or not some of them are able to
react with the NCO groups in the studied conditions.
Interesting ®ndings were also evidenced in the case of the reactions between the PhNCO and alcohol
mixtures (Table 6, Nos. 13 and 15). In agreement with the above conclusion it was found that for the
reaction rate level, the hydrogen-bond activating effect of the different alcohols present in the mixture is
cumulative, probably due to the relative equivalence of the different alcohols in the mixed hydrogen-
bond formation. Such a relative equivalence has been observed for a long time in the study of the
cryoscopic association degree effected on primary and secondary alcohols in benzene, valid especially
for the low range (,1 mol l 21) alcohol concentration [99,136].
With regard to the kinetic studies performed on the reaction between PhNCO and alcohol±amine
mixtures or between PhNCO and alcohol±phenol mixtures, reciprocal activation of the hydrogen donating
components, i.e. alcohol, phenol or amine were observed. The reaction rate of pure phenol in the standard
conditions utilised in the mentioned work was practically non-existent but if alcohol was introduced in the
828 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Table 6
Apparent second-order rate constant for alcohol±PhNCO reaction in Bz at 608C and the in¯uence of some additives (1-
BuOH ˆ butan-1-ol, 2-BuOH ˆ butan-2-ol, PhOH ˆ phenol, PhNH2 ˆ aniline). Reprinted with permission from Rev
Roumaine Chim 1996;41:725. q 1996 Editura Academiei Romane [80]

No. Alcohol 1 additive Concentration (mol dm 23) k £ 10 3


(mol 21 min 21 dm 3)
Alcohol 1 additive PhNCO

1 1-BuOH 0.175 0.088 16.78


2 1-BuOH 0.175 0.175 16.60
3 1-BuOH 0.35 0.176 33
4 1-BuOH 0.175 0.35 18.4
5 1-BuOH 0.35
1 0.179 78
CEE 0.709
6 1-BuOH 0.35
1 0.176 16.2
CMEE 0.7029
7 EtOH 0.0875 0.043 6.4
8 EtOH 0.175 0.0875 12.86
9 EtOH 0.35 0.175 26
10 PhOH 0.175 0.0875 0
11 EtOH 0.0875 6.4
1 0.0875
PhOH 0.0875 0.35
12 PhNH2 0.175 0.0875 1126
13 EtOH 0.0875 39
1 0.0875
PhNH2 0.0875 2648
14 2-BuOH 0.35 0.176 13.2
15 1-BuOH 0.175 33.4
1 0.176
2-BuOH 0.175 12.8

mixture the reaction with phenol became visible (Table 6, Nos. 10 and 11). Surprisingly the capacity of the
alcohol to is also able to increase the high reactivity of the amine even further.
In another kinetic study [116] the results of the reaction rate constant determination were presented
regarding the two reaction steps of a bifunctional hydroxy compound in the case of different glycol±
PhNCO reactions. Only the use of the HPLC method allowed us to solve this dif®cult kinetic study which
was broached long time ago by Robertson and Stutchbury [112] but still unclear till now.
The elucidation of the multiple-step glycol reaction is important from many points of view. As regards
the macromolecular chemistry, this information offers precious data on the elemental polyaddition step-
by-step reactions which govern the sequence ordering in the macromolecular chain. On the other hand,
this information can lead to a better understanding of the basic knowledge regarding the isocyanate±
alcohol reaction mechanism in organic chemistry.
The glycols represent a special class of hydroxy compounds characterised by an appreciable tendency
to form intramolecular bonding [113,114]. The kinetic measurements on the PhNCO±glycol reaction as
re¯ected in Table 7 [116], aims to assess the effect of the possible intramolecular bonding on the glycol
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 829

Table 7
Apparent second-order rate constants k1 and k2 (mol 21 min 21 dm 3) for glycol reactions with PhNCO in benzene (Bz) at 608C
…Ropt ˆ k1 =k2 ; was determined as described before [132]. 1-BuOH ˆ butan-1-ol, EG ˆ ethylene glycol, BD ˆ 1,4-butanediol,
HD ˆ 1,6-hexanediol, CEE ˆ carbanilic acid ethyl ester). Reprinted with permission from Rev Roumaine Chim 1996;41:539.
q 1996 Editura Academiei Romane [116]

No. Hydroxy compounds Concentration (mol dm 3) k1 £ 10 3 Ropt ˆ k1 =k2 k2 £ 10 3

Hydroxy compounds PhNCO

1 1-BuOH 0.175 0.08816 16.78


2 EG 0.0452 0.0452 11.22 1.44 8.0
3 EG 0.0749 0.0753 20 2.59 7.72
4 BD 0.0438 0.0438 121.6 2.69 46.9
5 BD 0.0876 0.0876 264 5.4 48.8
a
6 HD 0.0438 0.0438 19.4
7 HD 0.0882 0.08826 50.84 1.51 33.26
8 EG 0.0438 21.2 2.3 9.2
1 0.0875
1-BuOH 0.0875 14 ± ±
9 BD 0.0438 298 11 27.08
1 0.0875
1-BuOH 0.0875 19.2 ± ±
10 BD 0.0433 210 2.24 93.34
1 0.0433
CEE 0.1759 ± ±
a
Not calculated due to the reaction low conversion.

OH reactivity. Owing to the very low solubility of glycols in unpolar solvents, all the measurements
were performed only in benzene at low reactant concentrations.
As seen, the reaction rate of BD which proves to have the highest tendency to form intramolecular
bonding, presents in the ®rst step of the reaction (when it acts on the intramolecular hydrogen bonding) a
very high value in comparison to that of 1-BuOH or other glycols. When the solvent benzene is replaced
by a more polar solvent like tetrahydrofurane (THF) or Dx, the reaction rate constants corresponding to
the two glycolic OH groups tend to equalise each other due to the destruction of the intra and inter-
molecular OH self-association in favour of the majority alcohol±solvent associations [116]. The general
decrease of the reaction rate in these solvents indicates that the engagement of OH protons in hydrogen
bonding with different proton acceptor inert partners does not favour the reaction. All of these observations
lead to the conclusion that the mechanism of alcohol hydrogen-bonding activation is in fact not so simple
as it was initially supposed [122,132].
A special situation was revealed when the glycols±PhNCO reactions were performed in DMF. In
this case two opposite parallel tendencies were considered; the ®rst of them was expected to produce a
slow reaction deactivation due to OH proton binding with DMF, like in the case of other proton-acceptor
solvents. The second one regarded the ±NCO group reactivity increase to be a result of the direct
interaction with DMF. As seen from the experimental data, the overall observed effect was a sensible
increase of the reaction rate as a consequence of the preponderance of the ±NCO group reactivity
enhancement towards the opposite tendency produced by the slow ±OH reactivity decrease. A signi®-
cant loss of the selectivity regarding the different OH glycol groups reactivity is also observed.
830 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

4.3. The kinetics and other reactivity measurements in polyfunctional and macromolecular systems

Though the problem of the real mechanism in urethane reaction has not yet been clari®ed the under-
standing of the macromolecular synthesis has determined many workers to effect kinetic and other
reactivity measurements in order to assess the probable route of the reaction pathway during the building
of the multi-step isocyanate-based polymers.
In the reaction of polyurethane synthesis there are two principal types of component, e.g. the polyol
and the polyisocyanate. In the multiple-step polyurethane chain building besides the low-molecular
polyols, there are also higher molecular weight oligomers with ®nal OH groups that are often involved.
In order to obtain clearer information regarding each step of the reaction, the ®rst reactivity study,
related to polymer formation, was initiated on a model synthesis following one by one the reaction
behaviour of each bifunctional monomer in the presence of a corresponding monofunctional contra-
partner. The kinetic equations became simpler in this case than in the case of the global polymer
synthesis conditions, and can be easier solved.

4.3.1. The glycols reactivity


The problem of the glycols reactivity from the literature data has drawn the attention of researchers for
a long time but in an unjusti®ed smaller measure compared with that for diisocyanates, though in
principal both the reactants should have been studied in an equal measure. The reasons would be of a
subjective nature. The problem of the glycolic OH reactivity was probably considered as too banal
inspite of some authors pointed out many years ago the surprises that can appear when phenylisocyanate
reacts with different glycols. The peculiar behaviour of the OH group in the glycol series could not,
however, be clari®ed at that time [112].
Some of the recent results regarding the model kinetic study of the glycol±PhNCO reaction have
already been presented in the previous sections of this review. Yet similar information was reported with
regard to the model kinetic study of the reaction between macrodiols and monoisocyanates as that
performed in the case of glycols.

4.3.2. The diisocyanate reactivity


The ®rst model kinetic studies regarding the polyfunctional monomers involved in the polyurethane
synthesis was performed on diisocyanates [22]. Initially, the kinetic measurements con®ned only to
monitoring the overall ±NCO group consumption on employing chemical and/or IR methods. The
individual ±NCO group rate constants of the polyisocyanates were determined with a questionable
precision using the so-called ªTime Ratios Methodº developed by Frost and Pearson [138]. By this
way the rate of total ±NCO group consumption at a de®ned time ratio was determined and compared to
the ªReactivity Ratiosº values theoretically calculated and tabulated by Frost and Pearson [140].
Recently more precise methods for the kinetic study of the diisocyanate reactivity using HPLC [133±
135], size exclusion chromatography [126], UV [125] and ¯uorescence spectroscopy [125] have been
proposed.
The possible structures of diisocyanates can lead to a large difference between the consecutive
reaction steps of the two ±NCO groups. Thus this fact can determine the existence of two or even
four reaction rate constants in a single diisocyanate±alcohol reaction. In the case of the reaction between
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 831

different types of aromatic diisocyanate and n-buthanol, one of the following situations can occur:

…15†

The abbreviations under the type of the diisocyanates have the following meaning: S ˆ symmetrical
structure, A ˆ asymmetrical structure, N ˆ the two isocyanate ±NCO groups are non-conjugated with
each other, C ˆ the two isocyanate ±NCO groups are conjugated with each other, R ˆ rigid molecule
and V ˆ ¯exible molecule with variable geometry; k1, k2 ˆ reaction rate constants corresponding to the
832 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

®rst and second diisocyanate ±NCO group reactions, respectively. k 0 1 and k 0 2 ˆ reaction in o and p
positions, respectively, after performing the ®rst step reaction of the diisocyanate.
The kinetic studies using complete analyses of reaction mixture by HPLC, which made possible these
determinations, were initiated by Caraculacu et al. [133]. In the beginning the kinetics of symmetrical
diisocyanates, i.e. 4,4 0 -MDI, 4,4 0 -BBDI, 2,2 0 -BBDI, p-phenylene diisocyanate (p-PDI) and m-phenylene
diisocyanate (m-PDI) in reaction with 1-butanol and using benzene as solvent, were performed.
Two different reaction rate constants corresponding to the two reaction steps were evidenced in this
case. As expected, in the case of the symmetrical isocyanates as 4,4 0 -MDI and 4,4 0 -BBDI, the two
consecutive ±NCO reaction steps of each diisocyanate performs with approximately the same reaction
rate …k1 ù k2 †: An apparently unexpected result was evidenced only in the case of a special symmetrical
isocyanate belonging to the SNV diisocyanate type represented by 2,2 0 -BBDI [134]. Thus it was found
that after the ®rst ±NCO group reaction, the second ±NCO group reacts about 200 times faster than the
®rst one. The fact was attributed to an intense intramolecular activating effect produced by the new
formed urethane group on the second unreacted ±NCO group.

The most complicated system represented by 2,4-toluylene diisocyanate which belongs to the ACR
group, was studied by Caraculacu et al. [135] in l991, when four reaction rate constants were determined.
Further on we shall present these cases, in brief, as being illustrative of the analytical and mathematical
treatments. Others kinetic systems can be considered as particular cases of this situation.
The pathway of the reaction between 2,4-TDI and an alcohol is shown in scheme (16):

…16†
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 833

The ªidealisedº kinetic equations are:


dN=dt ˆ 2…k1 1 k2 †NA dM1 =dt ˆ …k1 N 2 k 02 M1 †A dM2 =dt ˆ …k2 N 2 k 01 M2 †A

dB=dt ˆ …k 02 M1 1 k 01 M2 †A dA=dt ˆ ‰…k1 1 k2 †N 1 k 02 M1 1 k 01 M2 ŠA (17)

where: N ˆ molar concentration of unreacted diisocyanate n equal to the molar concentration of


quenched diizocyanate n 0 ; M1 ˆ molar concentration of ortho monoreacted diisocyanate m1 equal to
the molar concentration of quenched the ortho monoreacted isocyanate m 0 1; M2 ˆ molar concentration
of para monoreacted diisocyanate m2 equal to the molar concentration of quenched para monoreacted
isocyanate m 0 2; B ˆ molar concentration of bireacted diisocyanate b, A ˆ molar concentration of alcohol
a and k1, k2, k 0 1 and k 0 2 are the rate constants.
By the HPLC analytical determinations it is possible to monitor only the n 0 , m 0 1, m 0 2 and b stabile
species.
In the particular case adopted by the authors the ratio between the initial isocyanate and the initial
alcohol concentration was ‰NCOŠ0 =‰OHŠ0 ˆ 2: In order to facilitate the mathematical treatment, a new
variable t expressed by Eq. (18) was introduced:
Zt
dt ˆ A dt; so t ˆ A dt …18†
0

in this case
FN ˆ e2…k1 1k2 †t FM1 ˆ R2 …e2k2 t 2 e2…k1 1k2 †t † FM2 ˆ R1 …e2k1 t 2 e2…k1 1k2 †t †

FB ˆ 1 2 FN 2 FM1 2 FM2 (19)

where
R1 ˆ k2 =…k1 1 k2 2 k 01 † R2 ˆ k1 =…k1 1 k2 2 k 02 †
and FN, FM1, FM2, FB are the fractions corresponding to the four reaction species …FN 1 FM1 1 FM2 1 FB ˆ 1†:
By removing the variable t from system (19), the following relations are obtained:
FM1 ˆ ‰R0 =…1 2 R2 †Š…FNR2 2 FN † FM2 ˆ ‰…1 2 R0 †=…1 2 R1 †Š…FNR1 2 FN † …20†
where
R0 ˆ k1 =…k1 1 k2 †; R1 ˆ k 01 =…k1 1 k2 † and R2 ˆ k 02 =…k1 1 k2 †
Using the condition dM1 =dt ˆ dM2 =dt ˆ 0 and relation (17) the maximum FM and FM fractions as well
as their corresponding FN values are obtained:
…FM1 †max ˆ R0 RR2 2 =…12R2 † ! FN ˆ R21=…12R2 † …FM2 †max ˆ …1 2 R0 †RR1 1 =…12R1 † ! FN ˆ R11=…12R1 †
…21†
To test the relation group (19) the calculation of t given by relation (18) was necessary. The
determination of the molar concentration of the alcohol was achieved indirectly, taking into account
that A ˆ N 2 B since A0 ˆ N0 : Numerical integration was achieved by a double square interpolation
834 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 17. Variation of ln FNexp (%) versus d .

method. The dependence of experimental value FNexp versus t is plotted in Fig. 17. It is noticed that this
dependence may be approximated well enough by a straight line. This fact is re¯ected by a relatively
stable value of the sum of the k1 1 k2 constants.
Table 8 displays the experimental values FNexp corresponding to the t values, and the sum of k1 1 k2
calculated with the relation:
k1 1 k2 ˆ 2ln…FNexp †=t …22†
The veri®cation of k1 1 k2 values was achieved by comparing the FNexp value to FNcalc obtained by
Eq. (23):

FNcalc % ˆ 100 e2…k1 1k2 †t …23†


where k1 1 k2 is the arithmetical mean of the obtained value, Table 8.

Table 8
Values of t function, …k1 1 k2 † sum and FNcalc%. Reprinted with permission from Rev Roumaine Chim 1991;36:1135. q 1991
Editura Academiei Romane [135]

No. T (min) FNexp % t (mol min l 21) k1 1 k2 (mol 21 min 21 l) FNcalc%

1 0 100 0.00 100.00


2 60.1 74.58 9.11 0.0322 72.31
3 120.1 58.46 15.96 0.0336 56.65
4 180.1 47.41 21.33 0.0350 46.79
5 239.9 38.86 25.62 0.0369 40.17
6 300 33.31 29.12 0.0378 35.46
7 359.9 30.02 32.10 0.0375 31.89
8 420 28.35 34.76 0.0363 29.01
k1 1 k2 ˆ 0:0356
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 835

Table 9
Optimal values of a and b parameters for two species obtained by search method. Reprinted with permission from Rev
Roumaine Chim 1991;36:1135. q 1991 Editura Academiei Romane [135]

Type of monoreacted species a opt b opt

M1 a1 ˆ R0 ˆ 0:107 b1 ˆ R2 ˆ 0:186
M2 a2 ˆ 1 2 R0 ˆ 0:803 b2 ˆ R1 ˆ 0:044

As expected, a good concordance was found between the FNexp and FNcalc values. Later on, a new
search method for the determination of the constants R0, R1 and R2 from Eq. (20) is presented.
Both FM1 and FM2 relations (20) have a common form as follows:
Y ˆ a‰…X b 2 X†=…1 2 b†Š …24†
where X ˆ FNexp ; Y ˆ FM1 exp ; a ˆ R0 and b ˆ R2 or X ˆ FNexp ; Y ˆ FM2 exp ; a ˆ 1 2 R0 and b ˆ R1
It was noticed that:
Y1 ˆ …X b 2 X†=…1 2 b† …25†
The procedure was the following: different equidistant values were adopted for b in a given interval,
in our case b [ ‰0; 1Š; considering initially an increment Db ˆ 0:01:
For a particular value of b we calculated the average value of a , corresponding to that of the n
experimental data:
X
n
a …b† ˆ ‰Y…i†=Y1 …i; b†Š=n …26†
iˆ1

where Y…i† represents FM1exp or FM2exp values, and Y…i; b† is the corresponding value calculated with the
aid of Eq. (25). The average value is adopted in order to calculate the sum of square mean errors.
X
n
S…b† ˆ ‰Y…i† 2 a …b†Y1 …i; b†Š2 …27†
iˆ1

It was observed that for every series of experimental data corresponding to each of the two mono-
reacted species, the S…b† values display an unique minimum. Firstly, the b optim value corresponding to
the minimum of S was established; then the procedure was repeated in order to obtain a more precise
value of b optim. This time the interval ‰boptim 2 0:01; boptim 1 0:01Š was considered, taking Db ˆ 0:001
as the increment. The procedure was repeated until the desired precision was obtained. This search
method could be applied to all the functions of the type Y ˆ af …x; b†: The obtained data are presented in
Table 9.
On adopting the average value of 0.0356 mol 21 min 21 l for the …k1 1 k2 † sum the following rate
constants, (expressed in mol 21 min 21 l) were calculated, by means of Eq. (20):
k1 ˆ 7:013 £ 1023 k2 ˆ 28:59 £ 1023 k 01 ˆ 1:577 £ 1023 k 02 ˆ 6:604 £ 1023 …28†
Table 9 points out a remarkable thing: the R0 and …1 2 R0 † values which were independently determined
for each of the two FM1 and FM2 data series, respectively, are in good agreement. This situation proves
both the good precision of the experimental measurements and the ef®ciency of the above-presented
836 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Table 10
Comparison of FMexp values with FMcalc values (all in %) [136]

No. FM1exp FM1calc FM2exp FM2calc

1 4.85 4.86 20.57 20.27


2 7.63 7.74 32.54 32.93
3 9.64 9.57 41.36 41.45
4 10.94 10.87 47.88 47.92
5 11.76 11.64 52.14 52.04
6 12.11 12.06 54.47 54.43
7 12.10 12.26 55.49 55.63

method. In Table 10 the experimental data are compared to those obtained theoretically by the substitution,
in Eq. (20), of the values given in Table 9.
A very good agreement can be also noticed between the experimental and calculated values; keeping
in mind that at the end of the reaction FNf ˆ FBf : The parameters R0, R1 and R2 obtained before give the
authors the possibility to calculate the ®nal composition of the reaction mixture using a search method.
For this it starts from the last FNexp value and the composition of the system is calculated using Eq. (20)
for different equidistant values of FN and for an initial increment DFN ˆ 0:01: The calculation ends
when FBf . FNf : At this point the calculation can be continued in order to give a more precise result by
using a smaller increment for FN, i.e. DFN ˆ 0:001: The FN variation interval is now ‰FNf 2 0:01; FNf Š:
Obviously, the new cycle of calculations should give more precise values for the ®nal composition. To
determine a much more precise ®nal composition, the calculation should be continued with a smaller FN
increment.
For the mentioned conditions of synthesis the following values for the ®nal mixture composition were
obtained (all in %):
FNf ˆ 9:47 FM1f ˆ 13:33 FM2f ˆ 67:77 FBf ˆ 9:47 …29†
Two years later, in 1993 Hugo and Wu [139] effected similar researches on 2,4-toluylene diisocyanate
but using the benzyl alcohol as a monofunctional alcohol instead of 1-butanol, and dioxane as a solvent.
The reaction was also monitored by HPLC. Under the conditions a higher difference was found between
the o- and p-NCO groups reactivity.
In the simplest case of symmetrical diisocyanate [132] of the SNR and SNV type, when only two
reaction rate constants must be determined, the mathematical treatment presented above is reduced to the
simplest Eq. (30) which allows the determination of the ®rst reaction rate constant.
k1 ˆ 20:5 ln…FNexp †=t …30†
The second reaction rate constant k2 of the diisocyanates mentioned before was determined by an
indirect method. With this view it was ®rstly necessary to determine a k1 =k2 ˆ Roptim value which should
verify the best, Eq. (31):
FM ˆ …FN1=…2R† 2 FN †=‰1 2 1=…2R†Š …31†
where FN, FM are the molar fraction of unreacted and respectively monoreacted diisocyanate: R the ratio
between the ®rst and second ±NCO groups reaction rate constant k1 =k2 :
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 837

The Roptim value was established by using a numerical searching method. The so-determined k2 value
was veri®ed by comparing the experimentally determined FMexp to the calculated FMcalc. An excellent
concordance was found in all of these cases. The deviations were situated between 1.6 and ^1.4%. This
concordance demonstrates that even if the apparent second-order reaction rate constants can change
more or less during the reaction, the rate between the ®rst and the second reaction rate constants R
remains practically invariable and can furnish precious information with regard to the polyurethane
synthesis.
The obtained results showed that there is no perfect equality between the two steps of the diisocyanate
reaction and so an approximate decrease of the second reaction rate constant was observed during the
reaction. This situation can have many explanations. One of them is the possibility that the compenfry
effects of urethane formation and alcohol consumption, which provide the constancy in the activation of
OH group through hydrogen bonding, should not be quite perfect. On the other hand, it is also possible
that with the reaction development there could appear supplementary NCO groups consumption due to
any secondary reactions with urethane or incidental moisture, leading to the some polymers oligomer
formation.
In the case of diisocyanates the eventual traces of such urea oligomers cannot be detected by HPLC
due to their high insolubility.
Surivet et al. [126] proposed a new technique for the study of diisocyanate reactivity by using the
size exclusion chromatography. The authors intended to compare the change in reactivity when the OH
group belonging to a small and mobile molecule as that of glycols is replaced by an ±OH group bonded
to a larger molecule like that of macrodiols. As model reactions the reactivities of some aliphatic
diisocyanates having symmetrical and asymmetrical structures like 4,4 0 -dicyclohexyl methane diiso-
cyanate (H12MDI), 5-isocyanato-1,3,3-trimethyl cyclohexylmethyl isocyanate and isophorone diisocyanate
(IPDI) were analysed, in reaction with a low-molecular alcohol, i.e. benzyl alcohol (Bz A) and with a
high-molecular alcohol represented by a-hydroxy, v-methyl ether-terminated polyethylene oxyde
(PED) MW ˆ 350 at 808C. The synthesis were performed in bulk state without catalyst. The second-
order kinetic mechanism was con®rmed to also be valid as in the other cases mentioned. The main
purpose of the work was to establish the reactivity ratio of the two NCO groups and not to determine the
individual reaction rate constant values. As regards the reaction conditions chosen by the authors, the
problem is unfortunately complicated by the fact that isocyanates are actually mixtures of isomers and
the authors have not made individual analyses concerning each isomer consumption.
As shown before, in the case of the 2,2-DBDI, the sterical conditions can determine an important
internal reciprocal anchimeric assistance which can be quite different from one isomer to another,
depending on the distance between the two reacting NCO groups. The chromatographic method utilised
by the authors did not succeed in offering separated maxima for every isomer. We should notice that as
mentioned in the work, H12MDI contains three isomers: cis±trans (65%), trans±trans (30%) and cis±cis
(5%); and in the case of IPDI we have actually a mixture of two isomers cis and trans of an unknown
composition. The kinetic equation system used was similar to that mentioned earlier. The rate constants
are obtained by numerical calculations using the Runge±Kutte method and the results are estimated by
comparing the concordance between the experimental and calculated value. The obtained results are
represented in Fig. 18 [126].
The authors found an unexpected difference between the ®rst and the second reaction step of H12MDI.
The explanation given in the work was correctly related to the substitution effect when the isocyanic
group was replaced by that of the urethane. Indubitably this effect would have been evidenced better if it
838 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 18. Comparison between the experimental values (broken curves) and the calculated values (full curves) of monomer
(X, X), monoreacted (Y, A), bireacted (Z, L) and alcohol (A, W) concentration variations for the benzylic alcohol and H12MDI
reaction at 808C in bulk, without catalyst, NCO/OH ˆ 2.

would have undertaken a complete isomeric analysis. In the case of IPDI, the cycloaliphatic isocyanate
group is found to be three times more reactive than the aliphatic one.
In another relatively recent work, Huang et al. [125] reconsidered again the kinetics of 1,5-naphtylene
diisocyanate (NDI) when the reaction is performed with a large excess of 1-butanol in toluene at 50, 60
and 708C. As an analytical method they proposed a combined UV absorption and ¯uorescence emission
spectra method. Contrary to other older results [137] they found that even though the two ±NCO groups
are bonded in some of the aromatic systems represented by the naphthalene skeleton, there is no
reciprocal reactivity in¯uence between the two ±NCO groups.
The electron withdrawing effect of NCO group in this aromatic system was not evidenced and
therefore the corresponding reaction rate constants for the two NDI reaction steps remain practically
equal …k1 ˆ k2 †:
Later on by extending the kinetic measurement on the MDI, Huang found that the reactivity of MDI is
practically equal to that of NDI [125].
The authors tried to apply their results to the polymer synthesis in the case of the MDI-based
polyurethanes. So, in order to follow the MDI consumption during the polyaddition reaction in solid
state, they introduced some quantities of NDI in the initial reaction mixture as a chemical sensor. The
authors proposed a calibration curve which makes it possible to obtain a correlation between the extent
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 839

of the unknown MDI consumption and that of NDI. It is possible to monitor the above-mentioned NDI
concentration in solid state by measuring the speci®c UV absorption and ¯uorescence spectra. In our
opinion much consideration should be given when extrapolating the results obtained in a concrete
situation to another one that is quite different. The dramatic changes for NCO and OH groups reactivities
are illustrative of such situations even when only the nature of the solvents are changed [116,134].
Larger changes are to be expected when passes from the solution to the melt reaction conditions.
In order to obtain an image on the chemical reactivity difference between the two functional groups
belonging to the same bifunctional monomer, other simpli®ed methods were proposed [140]. The simple
analysis of the ®nal reaction mixtures, sometimes, proved to be able to give precious informations. In
principle such a thing becomes possible when a bifunctional monomer is submitted to reaction with a
smaller equivalent quantity of a monofunctional partner [140]. Usually in this case the equivalent
functional group ratio between the bifunctional and the monofunctional partner was taken as equal to
about 2:1.
Caraculacu et al. [140] found a general procedure useful in such cases, starting from the ®nal reaction
mixture composition. The relative k1 =k2 ˆ R ratio can be established with a suf®cient good precision if
one mole of the bifunctional component in question is reacted with one mole of a monofunctional
contrapartner till the complete conversion. With this view, the transcendental Eq. (31) mentioned before
was solved by adopting different R values and using the Traub's algorithms. The so-determined FN value
versus different R was tabulated [140]. In parallel an empirical relation (32) was established which
surpasses the necessity of using any tables. Thus the R value can be easily calculated with the aid of the
exponential of Eq. (32). The standard deviation of the so-obtained R values is less than 6% comparative
to that obtained by using Eq. (31).

R ˆ k1 =k2 ˆ 0:2692FN21:43 1023:838 FN1:884 …32†

For the time being in all of these cases the ®nal reaction composition was determined with the aid of
the analytic HPLC separation of the ®nal products.
In the last time there were also other attempts made to avoid the dif®cult analytic separation of each
individual products from the ®nal reaction mixture. In this case the reaction product composition was
assessed by using spectral measurements on the whole ®nal mixture. Obviously this method is applicable
only in the case in which every reaction species leads to a speci®c signal within the global spectra. The
most promising way was the use of the 13C NMR method due to the fact that the ratio between the
different components can be obtained directly from the intensity of the different signals and does not
need a special calibration.
In connection with this, it is worth mentioning that the studies of Delides et al. [141] who used the 13C
NMR spectroscopy on a series of polyurethane model compounds in order to select their speci®c
characteristic signals. The comparison between the observed chemical shifts and the values predicted
by the Granet and Paul [142] additive theory was in reasonable agreement. The differences found
between the peaks positions of different structures are very useful for the qualitative and quantitative
measurements.
These results were applied by Sebenik et al. [143] in order to follow the polyaddition reaction
performed between ethylene (or 1,2-propylene) glycols and MDI. The rate constant for different parallel
reactions was calculated from the 13C NMR spectra. It was found that the primary OH group of 1,2-
propylene glycol is more reactive than that of the secondary one. Also it was evidenced that ethylene
840 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

glycol is more reactive than 1,2-propylene glycol. At low temperatures and in the absence of a catalyst
no reaction was detected.
Hatada et al. [144] determined the reactivity ratio of the two NCO groups from isophorone diisocyanate
(IPDI) corresponding to Z and S stereo isomers using the above-mentioned method as well. A competitive
reaction between the two IPDI NCO groups with a smaller quantity of 1-butanol representing only 50%
relative to that of the ±NCO group (expressed in equivalents) was used. After the complete reaction, the
obtained mixture was studied by 13C NMR. It was found that the secondary ±NCO is about 1.6 times
more reactive than the primary ±NCO in both Z and E isomers. The reactivity of the E isomer was found
to be slightly higher than that of the Z isomer.
The results are comparable to that obtained by Surivet et al. [126], who used the classical kinetic
method.

4.4. The kinetic study on the real polyaddition macromolecular synthesis between diisocyanates and
diols

In the previous parts of this review we tried to present the new ®ndings re¯ected by the recent
studies dedicated to a better understanding of the elemental process lying at the isocyanate polyaddition
reaction base. The majority of these studies were achieved on model reactions when at least one of the
components was of a monofunctional type.
Working under the conditions the obtained products are in general well-de®ned low molecular weight
substances, able to be isolated and analysed.
In the real polyaddition processes the reactive species grow continuously as a consequence of the
consecutive reaction. Thus the problem becomes more and more complicated. Owing to these special
conditions, relatively few fundamental works have been made in this direction.
The features that determine this complexity can be considered as follows:

1. Even from the beginning of the simplest polyaddition process represented by the reaction between
two unsymmetrical bifunctional monomers, there are four reaction rate constants involved.
2. The number of these constants multiplies even from the earlier stages of reaction. The simpli®ed
presumption that admits the equality of the reactivity between different consecutive polyaddition
steps was con®rmed many times.
3. The hydrogen-bond effects, which proved to play a very important role on the kinetics of the process,
may change during the reaction.
4. The molecular weight increase during the polyaddition determines the gradual changes of the viscosity
and implicitly the molecules' mobility till attaining a quasi-complete immobilisation of the macro-
molecules formed in gels or in solid state.
5. Additional complications appear in the last stage of the macromolecule formation during the so-called
cure and postcure processes. Yet, these processes can play a decisive role in the ®nal polymer
structure and suprastructure formation, and thereafter in¯uence signi®cantly the ®nal physical and
mechanical properties of the material [128].

Going beyond all these inconveniences, partial information was obtained by some researchers
who studied disproportionate polyurethane systems when one of the two bifunctional components
is taken in excess so that the whole reaction should be kept in a liquid phase. The so-obtained
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 841

®nal product often corresponds to an intermediate prepolymer which is prepared usually in the
technological steps of polymer production. In these cases, besides the global NCO group
consumption, it becomes possible to follow the intermediate oligomer evolution by means of
the gel permeation chromatography [145,146], size exclusion chromatography [126] or 13C NMR
spectroscopy [143]. With regard to the 13C NMR method performed without previous quenching
of the reactive groups the kinetic results are affected by an appreciable imprecision: the time of
10 min required by the 600 necessary spectra accumulations for single determination, is too long.
During the 10 min the reaction mixture composition changes remarkably, especially in the ®rst
part of the reaction. We should also note that after 60 min the conversions found by authors was
70% of the ®nal [143].
As regards the kinetic treatment of the obtained experimental results, the authors have signi®cantly
different opinions, as it will be shown further on.
In the case of symmetrical isocyanates and polyols, Szolary et al. [147] adopted simple second-
order kinetic treatment. Sojecki and Ulinska [148,149] also considered a second-order treatment
but admitted that the reactivity of both the groups from the symmetrical polyol could be unequal.
Krol et al. [150,151] adopted a second-order kinetics as well; using the extrapolation of the
kinetic theory of gases to polyaddition systems he tried to correlate the reaction rate to the
decrease of the reactive molecular mobility as a result of the gradual molecular weight
augmentation. Thompson [146] considered that for the system studied, the second-order equations
are inapplicable and proposed an empirical treatment. Surivet et al. [126] supposed that a better
description of the polyaddition process could be achieved by using complex equations in
which both the catalytic effects of the alcohol and urethane are involved, (as Sato had proposed
earlier).
The principal results obtained by applying the all the above-mentioned methods can be listed as
follows:
(a) The applicability of the second-order reaction equation was veri®ed at least in the cases of
the reaction between some polyols and symmetrical SNR-type diisocyanates like HDI and MDI
[128,147].
(b) The nature of the glycol residue from the adipic acid-based polyester chain in¯uences the ®nal ±OH
group reactivity, although apparently the close neighbour structure of the ±OH group is similar to all the used
polyesters: ^ ^ ^ COO(CH2)2OH in PEG, ^ ^ ^ COO(CH2)4OH in PBG and ^ ^ ^ COO(CH2)6OH
in PHS [147]. However this result becomes reasonable if admitting that the terminal ±OH groups are
actually involved in some different cyclic hydrogen-bond self-associations with its proper polyesther
neighbouring chain, which can modify their reactivity.
In the same work it was also found that besides its own structure, the polyesther reactivity also
depends on the nature of the symmetric diisocyanates used. However, surprisingly, the
authors found that the order of polyesther reactivity is not the same for the diisocyanate series
studied.
(c) In the case of low polar systems, represented by oligopolybutane diols when only the ±OH
hydrogen-bond self-associations are favoured, with the increase of the starting macrodiol length
an appreciable diminution of the reaction rate constant was observed. This fact is in agreement
with the dilution effect observed in the urethane reaction of simple alcohol performed in unpolar
solvents. The increase of the molecule chain length of these unpolar macrodiols in this case is a
dilution effect on the terminal ±OH groups concentration.
842 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Fig. 19. The dependence or reaction rate constant k1obs (1) and individual rate constant k1 (2) versus equivalent mass Me in the
reaction of 2,4 toluylene diisocyanate and various oligobutadiene diols.

The case of the reaction between oligopolybutanediol and 2,4-TDI in the melt state is illustrative of
this. (Fig. 19, [152]).
In other usual cases when the ±OH hydrogen-bond self-association is perturbed by the presence of the
large chain of other polar groups like in polyestherdiols, the ®nal ±OH group reactivity does not depend
anymore on the chain length [150].
(d) The reaction rate diminution during polyaddition with the development of the reaction was also
analysed by Krol in the case of the reaction between the diisocyanates and polyols leading to simple
linear polyurethanes (which do not contain hard segments). The author also adopted the second-order
mechanism for the kinetic treatment.
Starting from the fact that the addition reaction results from a collision between two reactant molecules
(A and B), he considered the general equation supplying the number of collisions between the molecules
A and B taking part during 1s in 1 dm 3. With this view he extrapolated the kinetic theory of gases to the
polyaddition process.
As known the equation supplying the number of collisions Z in gaseous state is given by
Eq. (33):

Z ˆ ‰…dA 1 dS †=2Š2 N 2 =106 ‰8pRT…1=MA 1 1=MB †Š1=2 CA CB …33†

where dA and dB are diameters of molecules undergoing collision; N is the Avogadro's number;
MA and MB are the molecular weight of the molecules involved in collision and CA and CB are
the concentrations of species A and B, respectively.
After some mathematical transformations the author obtains for the polyaddition rate constant kAB, the
equation:

kAB ˆ b 0 …1=MA 1 1=MB †1=2 …34†


A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 843

where b 0 is a constant factor characteristic of every concrete reaction conditions.

b 0 ˆ ‰…dA 1 dB †=2Š2 N 2 =106 …8pRT†1=2 e2…Ea =RT† …35†

where Ea is the activation energy.


Such a model was intended to comprise a system of second-order differential equations wherein the
rate constants for successive reactions diminish gradually in accordance with the oligomer molecular
weight, MA and MB, augmentation.
In order to illustrate this theory the author unfortunately selected from its experimental data mainly
the case when a diisocyanate component was used as a mixture: 80 wt% of 2,4- and 20 wt% of 2,6-
toluilenediisocyanate (2,4- and 2,6-TDI).
The kinetic parameters obtained in this case were taken by the authors as average values of the
different elemental reactions of the different ±NCO groups of TDI.
In fact as shown, there exists a high difference between the different TDI±NCO groups reactivity. So
even in the simplest case of a 2,4-pure isomer, the reaction rate constant ratios in non-reacted and
monoreacted 2,4-TDI are kp =ko =kp 0 =ko 0 ˆ 1=0:245=0:23=0:055:
Under the conditions, even in the absence of the supposed gradual molecular weight changing
effect the global reaction rate constant should decrease with the more reactive NCO group species
consumption, thus introducing a supplementary incertitude.
On the other hand, it is dif®cult to make comparisons between different results obtained in the
mentioned works when the starting component nature was changed. This is especially the case
when the reaction of isocyanates with simple alcohol taken as a model reaction was compared to
that of some macrodiols [150]. Within the work, as part of every experiment the authors kept the
same total reaction components concentration as equal to 50 wt%. As a consequence, in these
conditions the OH group equivalent concentration varied as a function of the respective compo-
nent molecular weight value. For example the ±OH group concentration was 0.53 eq l 21 in the
case of TDI-poly(oxypropylene-oxide)glycol M ˆ 1700 reaction, whereas in the case of C6H5 ±
NCO±ethanol reaction, the ±OH group concentration increased at 3.75 eq l 21.
Under the circumstances the hydroxyl group self-association changes and consequently the
reaction rate should be deeply modi®ed. As stated earlier, the reaction rate of ethanol
with phenyl isocyanate (molar ratio 2/1) performed in benzene at 608 enhances from
6.4 £ 10 23 to about 100 £ 10 23 dm 3 mol 21 s 21 when the ±OH eq. concentration varies from
0.0875 to 5 eq l 21.
From the above result it is obvious that even if the kinetic treatment proposed in the work is
attractive, under such conditions every reaction rate comparison should be regarded with special
care.
(e) A new hypothesis in the polyaddition reaction was proposed by Sojecki [148,149] who
presumed that the reactivity of the terminal ±OH group in symmetrical macrodiols could be unequal
as in the case of asymmetrical macrodiols. The experiments were performed on the reaction between
2,4-TDI using different macrodiols like polyethylene adipate M  u ˆ 2200 (PAGE2200) and Mu ˆ 1000
(PAGE1000), poly(ethylene-maleate) M  u ˆ 850 (PMGE850), poly(ethylene glycols) M u ˆ 5000 (PEG5000)

and M u ˆ 1000 (PEG1000) and polyxyprolylene glycol Mu ˆ 2000 (POPG2000). The kinetic measurement
in molten state was performed by measuring the total ±NCO groups consumption by using a chemical
method. Working at low conversions and considering a very high difference between the reactivity of
844 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

p-NCO group relative to the o-NCO group from 2,4-TDI, the author intended to determine the rate
constants k1 and k2 corresponding to the following consecutive reaction:

…36†

The ratio of the initial reactant group concentration [NCO]:[OH] was 1:1, 1:1.5 and 1:2. The reaction
of the ±NCO group from the TDI ortho position was neglected even in the case of the [NCO]:[OH] ˆ 1
ratio. The authors adopted this treatment basing on an old supposition which considers that the reactivity
of the TDI ortho position is 40 times slower than that corresponding to the TDI para position. During the
experiments the conversion was limited only to a maximum 25% conversion.
In order to be able to ®nd the k1 and k2obs values measuring only the decrease of the total ±NCO
group concentration, the author adopted the ªTime Ratioº method proposed by Frost and Pearson [138],
so that

…dx=dt†tˆ0
k1obs ˆ …37†
ab
t2
Zt2 1
2 dx t2 …dx=dt†tˆ0 x2
…dx=dt† dt 2 1
t1 x dt t1 ab 2 t1
k2obs ˆ t2   …38†
x2 …dx=dt†tˆ0 Zt2 2 Zt2
2 x dt 2 …a 2 2b† x2 dt
2 t1 ab t1 t1

where a and b are the initial concentrations of TDI and macrodiols and x is the concentration of the NCO
groups at time t.
After this mathematical treatment the authors found surprising differences between the reactivity of
the two macrodiol ±OH groups even when these are situated at the ends of symmetrical macrodiols like
polyethylene adipate (PAGE) and/or polyethylene glycol (PEG).
Starting from the reaction rate constant presented by the author if transforming the rate constants k obs
21 21 obs 0
expressed by l mol min in the equivalent values k expressed by l/eq. OH min, we have obtained
the following ±OH reaction rate constant ratios for the symmetrical macrodiols:

PAGE508C 0 0
2200 : k 1 =k 2 ˆ 2:15; PAGE508C 0 0
1000 : k 1 =k 2 ˆ 2:1 …39†

In this way, the unexpected difference between the reactivity of the two macrodiol terminal ±OH groups
appears more evidently.
This paradoxical situation probably derived from the highly simpli®ed initial presumption adopted by
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 845

the author regarding the extremely marked difference between the p- and o-NCO group reactivity of
TDI. In fact as it was indubitably proved using HPLC analysis the reactivity difference between the
p- and o-NCO groups from TDI is much more reduced than supposed by Sojecki. The system of
parallel reactions which would be considered as responsible for the total ±NCO group consumption in
time should be much more complicated than that proposed by Eq. (36).
The special case represented by the asymmetrical diisocyanates which display, as shown, two differ-
ent reactive ±NCO groups was also studied by Continho and Cavalheiro [127] in the polyaddition
reaction performed between hydroxy-terminated polybutediene and TDI or 3-isocyanatomethyl-3,5,5-
trimethylcyclodohexyl isocyanate. The authors found in this case a discontinuity in the second-order
plots. The complexity of the problem enhanced even more when the mixture of 2,4- and 2,6-TDI was
reacted with polypropylene glycol. In this situation eight different reaction rate constants should be taken
into account even from the very beginning [146].
The enhancement of the precision of the kinetic measurements in the polyurethane prepolymer
synthesis using the gel permeation chromatographic analysis and UV detectors represents one of the
most recent studies. It was searched to ®nd a new more ef®cient quenching reactant as a derivatizing
agent for the ±NCO groups monitoring so as to obtain an enhanced UV absorption. With this view
Thomson [148] proposed N-4-nitrobenzyl-n-propylamine (PNBP) because using this quenching agent
the molar extinction coef®cient of the obtained products at 269 nm is 7.5 times higher than in the case
when the quenching reaction was performed with simple alcohols.
By using this improved method it becomes possible to follow the evolution of the formation of the
higher molecular weight species when the oligomer reactive NCO group situated on the ends of the
molecule becomes more and more rare.
In gel permeation chromatography there appear different maxima characteristic of different oligomer
species lengths. The evolution of the chromatograms signals during the polyaddition follow the gradual
enhancement of the molecular weight of an ensemble of oligomers which constitute the starting macro-
diols, which in turn is dependent on its initial polydispersion.
The quantitative estimation of their concentration starts from the hypothesis that the registered UV
signals in chromatograms are given only by the ®nal groups of every species. Obviously under the given
conditions, each oligomer from the mixture in question would have different molecular extinction
coef®cients which decreases with the molecular weight increase.
According to the author expectations the signals from the chromatograms represent only the UV
absorption at 267 nm produced by the urea group from the oligomer ends formed as a result of the
quenching reaction between the NCO groups oligomer ends and PNBP.
In fact the absorption at 267 nm is a result of the cumulative effect of both urea and that of the inherent
urethane groups from the molecule chain which also present absorption at the same wavelength.
Even in principle the quenching with PNBP generates urea groups having high UV molecular extinc-
tion coef®cients; with the oligomer length increase (with the NCO end group concentration decrease) the
contribution of the increasing number of newly formed lower absorptive urethane groups from the
oligomer chain becomes more and more important.
Under the circumstances, at higher molecular weight, the calculations regarding the oligomer length
using the so-determined ®nal reactive groups concentration become very disputable.
In our opinion this situation of the difference between the reactivity of the four different ±NCO groups
(from the used in the reaction, 2,4- and 2,6-TDI mixture), also neglected by the authors, leads to the
misunderstanding of the applicability of the second-order kinetic treatment.
846 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

Due this situation, the authors considered that it was possible to use only an empirical treatment of the
kinetic by adopting Eq. (40)
t t b
ˆ 1 …40†
%conversion a a
The a and b terms, characteristic of the studied system, were determined from the slope and intercept of a
plot of Eq. (40).
In recent times new approaches of the urethane kinetic reactions have been proposed. Starting from
the previously observed theoretical incompatibility between the second-order kinetic treatment and the
observed autocatalytical effect of one of the reactants (i.e. alcohol) and also from the observed catalytic
effect of the formed urethane, Cognet et al. [153] reiterated Sato's relation. In the simpli®ed case in
which the reaction performs without any special introduction in the reaction catalyst (the case of the so-
called ªuncatalysedº), the kinetic equation is represented as follows:
dNCO=dt ˆ k1 ‰NCOŠ‰OHŠ2 1 k2 ‰UŠ‰NCOŠ‰OHŠ …41†
In the case of the reaction between 1,4; 3,6-dianhydrosorbitol (DAS) and 4,4-diphenylmethane
diisocyanate (MDI), the solution of this equation was obtained using a computational Runge±Kutta's
numerical method. The results obtained were found to be in good agreement with the experimental data.
Indeed Sato's equation is attractive because it is supposed that this way the kinetics of the process
could probably be better explained under the assumption of the catalytic action of both alcohol and
urethane group. However, as regards to the validity of Sato's equation, some comments should be
made. In some recent works the catalytic action of the alcohol and urethane group was claimed to be
approximately equal, k1 ù k2 ˆ k; under these conditions Eq. (41) becomes:
2dNCO
ˆ k…‰OHŠ 1 ‰UŠ†‰NCOŠ‰OHŠ …42†
dt
On the other hand, due to fact that during the reaction, the ±OH groups are converted in the urethane
group U, the sum of these groups concentrations remains constant during the reaction and
‰OHŠ 1 ‰UŠ ˆ ‰OHŠ0 …43†
where [OH]0 is the initial alcohol concentration.
Thus we return to the well-known second-order kinetic expression
2dNCO=dt ˆ k 0 ‰NCOŠ‰OHŠ …44†
where k 0 is the observed second-order reaction rate constant and
k 0 ˆ k‰OHŠ0 …45†
This situation can justify the observed second-order reaction rate constant dependence on the initial
alcohol concentration as well.
Leaving aside the previously mentioned hypothesis regarding the equality between the k1 and k2, the
validity of Sato's equation can be veri®ed with a higher precision, at least concerning the square
dependence between the ±OH concentration and the reaction order constant re¯ected by the ®rst term
of Eq. (41). Such a veri®cation becomes possible if using the experimental data obtained by other
previously mentioned authors. The experimental data obtained in the isocyanate ultradiluted conditions
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 847

by Satchell (when the alcohol is taken in extremely large excess) can be very useful from this point of
view. The catalytic in¯uence of the little quantities of the urethane formed during the reaction becomes
negligible due to the presence of excess alcohol and Sato's equation (41) can be reduced to the following
form:

d‰NCOŠ
2 ˆ k1 ‰NCOŠ‰OHŠ2 …46†
dt
Taking into account the large excess of alcohol, the OH groups concentration are practically constant
during the reaction and Eq. (46) can be written as a pseudo-®rst order Eq. (47):

d‰NCOŠ
2 ˆ k 0 ‰OHŠ20 …47†
dt
where k 0 ˆ k1 ‰OHŠ20 and [OH]0 is the initial alcohol stoichiometric concentration; the so obtained k 0
should be equal to the kobs determined by Satchell in the his work.
Under the conditions, the ratio kobs : ‰OHŠ20 for different experiments effected in similar conditions but
starting from different [OH]0, should remain constant …kobs =‰OHŠ20 ˆ k1 †:
Unfortunately, the results obtained by using the mentioned Satchel's experimental data do not lead to
a constant value as expected. This very pertinent result con®rms the validity of Sato's equation at least
with regard to the square OH group concentration in¯uence on the urethane reaction rate.
So we should accept the fact that for the time being both the usual second-order treatment and Sato's
kinetic equation represent just two different ways giving us only some simpli®ed mathematical approx-
imations of the process. The real dependence should be more complicated and includes the reactivity and
concentrations at different self or mixed hydrogen bonds and possibly other types of association between
the species displayed in the reaction mixture, which are unfortunately dif®cult to be directly determined.
Thus, the fundamental problem of urethane reaction still remains open, in contradiction with the
opinion of many highly optimistic researchers and anonymous scienti®c journal referees who often
consider this ®eld of research as too old and banal.
Leaving aside the questions related to the fundamental polyaddition process, other studies should
be mentioned, regarding the ®nal structure formation during the solid state cure and postcure processes
in the polyurethane synthesis. It is often very dif®cult to decide whether these reactions are of the
same type as the initial ones or are produced by other processes. These studies also take into account
the possible secondary reactions that are sometimes associated with the atmospheric moisture or
oxygen. Due to the fact that all these processes perform in the solid state, only few of the known
research methods can be ef®ciently used to obtain valuable structural information. Among them, the
spectroscopic techniques have proved to be useful in monitoring the decrease of the ±NCO group
content. The (ATR) FT-IR method provides quantitative information, yet it is limited only to the surface
processes.
Intrinsic ¯uorescence techniques and photoacustic FT-IR spectroscopy were also used in the monitor-
ing the principal cure process regarding the urethane and urea group formation. Unfortunately no
information was given on the other possible secondary reactions in the solid state cure and postcure,
like the uretidinedione, allophanate or isocyanurate groups formation.
At relatively recent times a study presented by Duff and Maciel [154] proved that 15N and 13C solid
state CP/MAS NMR is also a powerful method in monitoring the postcure reaction chemistry of the
848 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

residual isocyanate group consumption in isocyanate resins, and showed that the urea group formation is
the preponderant process observed during the prolonged action of the atmospheric moisture.
Combined methods using both physical±chemical analysis and mechanical kinetics studies were
presented in an extensive work by Prisacariu in order to obtain new information regarding the postcure
mechanisms of the solid state reactions during the casting polyurethane elastomers synthesis.
It was found that in time, within the postcure process, the urea group formation is preeminent over that
of allophanate.
In conclusion we consider that it is dif®cult to discuss all the interesting features of the study of
isocyanate polyaddition in such a limited space. Yet, we hope that this material would be useful to the
researchers who wish to investigate this ®eld.

Acknowledgements

The author wishes to thank warmly Prof. Otto Vogl of The Polytechnical University Brooklyn, New
York, for encouraging in the endeavours to realise this review.

References

[1] Ulrich H. Elastoplastics 1971;3:97.


[2] Richter R. Chem Ber 1969;102:938.
[3] Shozda RI. J Org Chem 1967;32:2960.
[4] Richter R. Tetrahedron Lett 1968:5037.
[5] Richter R. Chem Ber 1969;102:931.
[6] Furukava J, Yamashita S, Murahashi M, Harada K. Makromol Chem 1965;85:80.
[7] Kozak NV, Nizelskii IuN. Ukrainian Polym J 1994;3(1±4):20.
[8] Takida H, Noro T. Kobunshi Kagaku 1965;22:463.
[9] Nicolaev VN, Izeeva MM, Semchicov ID. Vysokomol Soedin A 1980;22:857.
[10] Ohno K, Tsuji J. Chem Commun 1971:241.
[11] Herwech JE, Kauffman WJ. Tetrahedron Lett 1971:807.
[12] Ozaki S, Kato T. J Polym Sci, Part C 1968;23:695.
[13] Dilcone RO. Polym Prepr, Am Chem Soc, Div Polym Chem 1967;9:642.
[14] Ozaki S. Chem Rev 1972;72:457.
[15] Ulrich H. J Polym Sci, Macromol Rev 1976;11:93.
[16] Merten R. Angew Chem Int Ed 1971;10:294.
[17] Kogon J. J Am Chem Soc 1956;78:4911.
[18] Bakalo LA, Lipatova TE, Lipatov IS. Dokl Akad Nauk SSSR 1983;268:879.
[19] Reichen W. Chem Rev 1978;78:569.
[20] Satchell DPN, Satchell RS. Chem Soc Rev 1975;4:231.
[21] Donohoe G, Satchell DPN, Satchell RS. J Chem Soc Perkin Trans 1990;2:1671.
[22] Saunders JH, Frisch KC. Polyurethanes. Chemistry and technology, Part. I. New York: Wiley, 1962.
[23] Mitchell AD, Cross LC, editors. Interatomic distances. London: The Chemical Society, London, 1958 (No. 11).
[24] Nenitescu CD. Tratat elementar de chimie organica. Bucuresti: Editura Tehnica, 1956.
[25] Sacher E. J Macromol Sci Phys B 1979;16:525.
[26] Zabrodin VB, Bagaturijanc AA, Entelis SG. Zh Fiz Khim 1968;42:2324.
[27] Zabrodin VB. Zh Fiz Khim 1971;45:682.
[28] Bondarenko SP, Tiger RP, Borisov EV, Bagaturijanc AA, Entelis SG. Zh Org Khim 1974;10:271.
[29] Litinskii AI, Shreiberg AI, Baliaicus LH, Bolotin AB. Teor Exp Khim 1972;8:807.
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 849

[30] Bondarenko SP, Tiger RP, Bagaturijanc AA, Borisov EV, Lebedev VL, Entelis SG. Izv Akad Nauk SSSR Ser Khim
1977;2:293.
[31] Egorov IuP, Kisilenco AA, Penkovscki VV. Teor Exp Khim 1972;8:613.
[32] Litinskii AO, Nizelskii IuN, Lipatova TE, Rahimov AI, Shatkovskaia DB. Fizica molekul. 5th edn. Kiev: Naukovaja
Dumka, 1977.
[33] Rode YM, Kosmus W, Nachbaur E. Chem Phys Lett 1972;17:186.
[34] Howell JH, Absar I, Wazer van JR. J Chem Phys 1973;59:5895.
[35] Poppinger D, Radom L, Pople JA. J Am Chem Soc 1977;99:7806.
[36] Mc Lean AD, Loew GH, Berkowitz DS. J Mol Spectrosc 1977;64:184.
[37] Poppinger D, Radom L. J Am Chem Soc 1978;100:3674.
[38] Bondarenko SP, Tiger RP, Bagaturijanc AA, Lebedev VL, Entelis SG. Zh Fiz Khim 1980;54:1370.
[39] Eyster EH, Gilette RH, Brockway. J Am Chem Soc 1940;62:3236.
[40] Jones LH, Sholery JN, Shulman RG, Yost GM. J Chem Phys 1950;18:990.
[41] Hocking WH, Gerry MCL, Winnewisser M. Can J Phys 1975;53:1869.
[42] Lett RG, Flygare WH. J Chem Phys 1967;47:4730.
[43] Anderson DWW, Rankin DWH, Robertson A. J Mol Struct 1971;8:435.
[44] Kozak NV, Nizelskii IuN. Teor Exp Khim 1988;2:155.
[45] Nizelskii IuN, Kozak NV, Lipatova TE. Ukr Khim Zh 1987;53:772.
[46] Nicolaev VN, Izeeva MM, Semcikov IuD. Vysokomol Soedin A 1980;22(4):857.
[47] Ramesh S, Radhakrishnan G. J Indian Chem Soc 1997;74(4):347.
[48] Peruchin JW, Archirejew WP, Kusnezow EW. Plaste Kautsch 1975;22:394.
[49] Lipatova TE, Kozak NV, Nizelskii IuN, Krugleak NE. Teor Exp Khim 1984;20:86.
[50] Nesterenko AM, Maslov VG. Zh Strukt Khim 1979;20:949.
[51] Fudjimoto H, Fukui C. Mir 1970:30.
[52] Baker JW, Gaunt J. J Chem Soc 1949:9.
[53] Baker JW, Holdsworth. J Chem Soc 1949:19.
[54] Baker JW, Gaunt J. J Chem Soc 1949:27.
[55] Oberth AE, Bruenner RS. J Phys Chem 1968;72:845.
[56] Huckel W. Teoretische grundlagen der organischen chemie, vol. 2. Leipzig: Akademische Verlagsgesellschaft Geest &
Portig, 1957.
[57] Pimental GC, McClellan AL. The hydrogen bond. San Francisco, CA: Freeman, 1960.
[58] Schuster P, Zundel G, Sandorfy C. The hydrogen bond. Recent developments in theory and experiment. Amsterdam:
North Holland, 1976.
[59] Park CG, Tasumi M. J Phys Chem 1991;95:2757.
[60] Huisken F. Ber Max-Plank Inst Stroemungsforsch 1990;2:114.
[61] Graener H, Ye TQ, Laubereau A. J Chem Phys 1989;91:104.
[62] Graener H, Ye TQ, Laubereau A. J Chem Phys 1989;90:341.
[63] Graener H, Ye TQ, Dohlus R, Laubereau A. Ultrafast phenomenon 6, Springer series in chemistry and physics, vol. 48.
Berlin: Springer, 1988. p. 458.
[64] Edwards HGM, Farwell DW, Jones A. Spectrochim Acta, Part A 1989;45:1165.
[65] Giguere PA, Pigeon-Bosselin M. J Solut Chem 1988;17:1007.
[66] Shinomya T. Bull Chem Soc Jpn 1989;62:3643.
[67] Shinomya T. Bull Chem Soc Jpn 1989;62:3636.
[68] Zhumaev T, Ovlyakulyev B, Shakhparonov MJ. Vestn Mosk Univ, Ser 2, Khim 1989;30:546.
[69] Finney JL, Tomkinson J. J Mol Struct 1990;237:249.
[70] Bermeis FJ, Batallan F, Howells WS, Carlite CJ, Enciso E, Garcia-Hernandez M, Alvarez M, Alonso J. J Phys: Condens
Matter 1990;2:5005.
[71] Matsumoto M, Gubbins KE. J Chem Phys 1990;93:1981.
[72] Mei YP, Van JR, Yan XH, Yon JQ. Phys Rev B, Condens Matter 1993;48:577.
[73] Legon AC. Chem Soc Rev 1990;19:197.
[74] Voytiuk AA, Bliznyuc AA. Zh Strukt Khim 1992;33:157.
[75] Potier A. Chem Solid State Mater 1992;2:1.
850 A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851

[76] Ratajezak H. Stud Phys Theor Chem 1992;78:293.


[77] Satchell DPN, Satchell RS. J Chem Res (S) 1988:190.
[78] Wolf KL, Herold W. Z Phys Chem B 1934;27:58.
[79] Nagata J, Miyamoto K. Fluid Phase Equilib 1990;56:203.
[80] Caraculacu AA, Agherghinei I, Baron P, Timpu D. Rev Roumaine Chim 1996;41:725.
[81] Methoden der organischen chemie, vol. III. Stuttgart: Georg Thieme, 1955. p. 402.
[82] Kolbe A. J Mol Liq 1990;46:141.
[83] Huisken F, Kuleke A, Laush C, Lisy J. J Chem Phys 1991;95:3924.
[84] Iwahashi M, Hayashi Y, Hachiya N, Matsuzawa I, Kobayashi H. J Chem Soc Faraday Trans 1993;89:707.
[85] Fulton LL, Yee Geary G, Smith RD. Supercritical ¯uid engineering science, ACS Symposium Series 1993. Washington,
DC: Amercian Chemical Society, 1993. p. 175 (514pp).
[86] Hufnagel FK, Hanna FF, Ghoneim AH, Saad AL, Turky G. J Chem Soc Faraday Trans 1992;88:1819.
[87] Schulmann EM, Dwyer DW, Doetschmann DC. J Phys Chem 1990;94:7308.
[88] Kohler F, Lui A, Karres L, Gaube J, Wiss Z. Tech Hochsh Carl Schorlemmer Lenna-Merseburg 1989;31:711.
[89] Ghonasgi D, Chapman WG. Mol Phys 1993;79:291.
[90] Thernova EA, Tiger RI, Tarakanov OG. Zh Strukt Khim 1986;27:19.
[91] Schriver L, Burneau A, Perchard JP. J Chem Phys 1982;77:4926.
[92] Schrems O. J Mol Struct 1986;141:451.
[93] Hoffbauer MA, Giese CF, Gentry WR. J Phys Chem 1984;88:181.
[94] Luck WAP, Fritzche M. J Mol Struct 1993;295:47.
[95] Martinez S. Spectrochim Acta A 1990;46:1395.
[96] Panayiotou C, Sanchez IC. J Phys Chem 1991;95:10 090.
[97] Gupta RB, Combes JR, Johnston KP. J Phys Chem 1993;97:707.
[98] Klippenstein SJ, Hynes JT. J Phys Chem 1991;95:4651.
[99] Wong NM, Drago RS. J Phys Chem 1991;95:7542.
[100] Torrie BH, Weng SX, Powell BH. Mol Phys 1989;67:575.
[101] Orozco M, Luque FJ. J Comput Chem 1993;14:587.
[102] Scheiner S. Rev Comput Chem 1991;2:165.
[103] Schmidt PP. J Phys Chem 1993;97:4249.
[104] Shida N, Almlof I, Barbara FB. J Phys Chem 1991;95:10 457.
[105] Park CG, Tasumi M. J Phys Chem 1991;95:2757.
[106] Luck WAP. Angew Chem 1980;92:29.
[107] Sokolov ND, Vener MV, Savellev VA. Fiz Mnogochastichnykh Sist 1991;19:51.
[108] Stasyuk IV, Ivankiv AL. Mod Phys Lett B 1992;6:85.
[109] Graener H, Ye TQ, Laubereau A. Ultrafast phenomenon spectroscopy, Springer proceedings in physics, vol. 49. Berlin:
Springer, 1990. p. 252.
[110] Veytsman BA. J Phys Chem 1990;94:8499.
[111] Moog RS, Maroncelli M. J Phys Chem 1991;95:10359.
[112] Robertson WGP, Stutchbury JE. J Chem Soc 1964:4000.
[113] Kuhn LP. J Am Chem Soc 1952;74:2492.
[114] Kuhn LP, Schleyer PR, Bartinger WF, Eberson L. J Am Chem Soc 1964;86:650.
[115] Fishman EF, Chen TL. Spectrochim Acta A 1969;25:1231.
[116] Caraculacu AA, Agherghinei I, Baron P, Coseri S. Rev Roumaine Chim 1996;41:539.
[117] Kollman PA, Allen LC. Chem Rev 1972;72:283.
[118] Tiger RP, Behli LS, Bondarenko SP, Entelis SG. Zh Org Khim 1973;9:1563.
[119] Kozak NV, Nizelskii IuN. Teor Exp Khim 1988;24:155.
[120] Kozak NV, Nizelskii IuN. Teor Exp Khim 1990;26:155.
[121] Klopman G. Mir 1977:63.
[122] Baker JW, Holdsworth J. Chem Soc 1947:713.
[123] Nguyen AH, Marechal E. J Macromol Sci Chem A 1981;16:881.
[124] Wong SW, Frisch KC. J Polym Sci, Part A: Polym Chem Ed 1986;24:2867.
[125] Huang XY, Yu W, Sung CSP. Macromolecules 1990;23:390.
A.A. Caraculacu, S. Coseri / Prog. Polym. Sci. 26 (2001) 799±851 851

[126] Surivet F, Lam TM, Pascault JP. J Polym Sci, Part A: Polym Chem Ed 1991;29:1977.
[127] Continho F, Cavalheiro L. J Appl Polym Sci: Appl Polym Symp 1991;49:29.
[128] Continho F, Cavalheiro L. Polym Bull 1990;23:341.
[129] Matsunaga K, Yamasita T, Baigaku T. Kugakubu Kenkyu Hokoku 1988;24:57.
[130] West W. Chemical aplications of spectroscopy. New York: Interscience, 1956.
[131] Tucker EE, Becker ED. J Phys Chem 1973;77:1783.
[132] Caraculacu AA, Agherghinei I, Gaspar M, Prisacariu Cr. J Chem Soc, Perkin Trans 1990;2:1343.
[133] Caraculacu AA, Agherghinei I, Baron P, Caraculacu G, Coseri S. Eur Polym J 1996;32:1235.
[134] Caraculacu AA, Agherghinei I, Prisacariu Cr, Cozan V. J Macromol Sci, Chem A 1990;27:1547.
[135] Agherghinei I, Prisacariu Cr, Caraculacu AA. Rev Roumaine Chim 1991;36:1135.
[136] Wolf KL, Harms H. Z Phys Ch B 1939;44:366.
[137] Barbalata A, Caraculacu AA, Iurea V. Eur Polym J 1978;14:427.
[138] Frost AA, Pearson RG. Kinetics and mechanism. 2nd ed. New York: Wiley, 1956. p. 153.
[139] Hugo P, Wu R. Chem Ing Tech 1993;65:317.
[140] Caraculacu AA, Agherghinei I, Gaspar M. Makromol Chem Rapid Commun 1986;7:761.
[141] Delides C, Pethrick RA, Cunliffe AV, Klein PG. Polymer 1981;22:1205.
[142] Grant DM, Paul EG. J Am Chem Soc 1964;86:2984.
[143] Sebenik A, Kastelic C, Obredkar V. J Macromol Sci Chem A 1983;20:341.
[144] Hatada K, Ute K, Oka KI. J Polym Sci, Part A: Polym Chem 1990;28:3019.
[145] Stovbun EV, Lodigina VP, Kuzaev AI, Romanov AK, Baturin SM. Vysokomol Soedin A 1984;26:1449.
[146] Thompson CM, Taylor SG, Mc Gee WW. J Polym Sci, Part A: Polym Chem 1990;28:333.
[147] Szolary C, Farkas F, Kelemenne HA, Gardos G. Magyar Kem Lapja 1988;43:265.
[148] Sojecki R, Ulinska A. Polym J Chem 1978;52:1231.
[149] Sojecki R. Acta Polym 1989;40:715.
[150] Krol P, Atamanczuk B, Pielichowski I. J Appl Polym Sci 1992;46:2139.
[151] Krol P, Gawdzik A. J Appl Polym Sci 1995;58:729.
[152] Stovbun EV, Lodigina VP, Andrianova ZS, Baturin SM. Vysokomol Soedin A 1987;29:2500.
[153] Cognet-Georjon E, Mechin F, Pascault JP. Macromol Chem Phys 1995;196:3733.
[154] Duff DW, Maciel GE. Macromolecules 1991;24:387.

You might also like