Polymer 42 (2001) 981990

www.elsevier.nl/locate/polymer

Relation of glass transition temperature to the hydrogen-bonding degree

and energy in poly(N-vinyl pyrrolidone) blends with hydroxyl-containing
plasticizers. Part 2. Effects of poly(ethylene glycol) chain length
M.M. Feldstein a,*, S.A. Kuptsov b, G.A. Shandryuk a, N.A. Plate a
a

A.V. Topchiev Institute for Petrochemical Synthesis, Russian Academy of Sciences, 29 Leninsky prospekt, 117912, Moscow, Russia
b
Department of Physics, Moscow State Pedagogical University, 1, Malaya Pirogovskaya, 119882, Moscow, Russia
Received 17 February 2000; received in revised form 1 May 2000; accepted 12 June 2000

Abstract
A phenomenological approach has been developed to evaluate a variety of the characteristics of hydrogen bonding in poly(N-vinyl
pyrrolidone) (PVP) miscible blends with short chain poly(ethylene glycol) (PEG), ranging in molecular weight from 200 to
1000 g mol 21. The approach is based on the analysis of experimentally measured composition dependence of the negative deviations in
glass transition temperature, Tg, from weight-average values predicted by the Fox equation. The PVPPEG miscibility is a result of hydrogen
bonding between carbonyl groups in PVP repeat units and both terminal hydroxyls of PEG short chains. Because PEG macromolecules bear
reactive hydroxyl groups only at both chain ends, the PVPPEG complex has a network supramolecular structure. Inuence of blend
composition and PEG molecular weight on the mechanism of hydrogen bonding, the structure and the stoichiometry of the PVPPEG
complex have been studied. The signicance of this work is two-fold. First, the validity of the approach suggested for determining the
stoichiometry, network density and the thermodynamics of hydrogen-bonded complex formation in PVPPEG systems has been conrmed
by the results of independent measurements. Second, the nonequimolar stoichiometry of the hydrogen-bonded complex has been explained
taking into account the counterbalancing contributions of the entropic loss caused by PEG chain immobilization by hydrogen bonding to PVP
repeat units through both PEG chain-end hydroxyls, and the entropic gain due to the increase of the mobility of PVP chain segments between
neighbouring hydrogen-bonded PEG crosslinks in the PVPPEG network. q 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Poly(N-vinyl pyrrolidone)poly(ethylene glycol) blends; Glass transition temperature; Quantication of hydrogen-bonding degree and energy

1. Introduction
Poly(N-vinyl pyrrolidone) (PVP) and poly(ethylene
glycol) (PEG) contain only electron-donating groups in
their repeat units. It is therefore no wonder that PVP has
been shown to be immiscible with high molecular weight
PEG [1]. At the same time, PVP is reported to be soluble in
liquid PEG having a molecular weight of 400 g mol 21 [2]
and this behaviour implies the contribution of proton-donating terminal hydroxyl groups of oligomeric PEG to the
compatibility with PVP. Actually, PVP is immiscible with
dimethyl ether of PEG-400, DMPEG, whose chain-end
reactive protons are replaced by inert methylene groups
[3]. As has been established by FTIR spectroscopy, the
compatibility of PVP blends with PEG-400 is due to hydrogen bonding of PEG terminal hydroxyls to the carbonyls in
* Corresponding author. Fax: 17-095-230-2224.
E-mail address: [email protected] (M.M. Feldstein).

PVP repeat units [3,4]. The nitrogen atom of PVP amide

groups and the oxyethylene units in PEG chains are not
involved in PVPPEG hydrogen bonding [4].
The hydrogen-bonded PVP complex with PEG-400
demonstrates a nonequimolar stoichiometry [57] and
physical properties, which are not typical for unblended
polymers: rubber-like elasticity [4] and pressure-sensitive
adhesion toward diverse substrates [8]. Because reactive
OH groups are located at both ends of the PEG chains,
PEG behaves as a hydrogen-bonding telechelic crosslinker
and the PVPPEG complex reveals a supramolecular
network structure [4]. Due to an appreciable length and
exibility of hydrogen-bonded PEG, the PVPPEG
network exhibits a unique combination of a high cohesive
strength inherent in cured rubbers with an enhanced free
volume, which ensure a liquid-like molecular mobility of
polymer segments in the complex.
Molecular interactions in miscible polymer blends inuence the composition dependence of glass transition

0032-3861/01/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0032-386 1(00)00439-0

982

M.M. Feldstein et al. / Polymer 42 (2001) 981990

temperature, Tg. The contributions of enhanced free volume,

strong favourable interaction and mutual orientation of PVP
and PEG chains are embedded in the compositional dependence of the Tg in PVPPEG blends [4,7,9,10]. Meanwhile,
it is also highly desirable to have a simple scheme for correlating the Tg with the measurements of intermolecular interactions, such as bonding degree and energy. In the rst
paper of this series [6] we have attempted to express in
explicit form the negative Tg deviations from the weightaverage magnitudes described by the Fox equation [11]
through the number of polymerplasticizer hydrogen
bonds. In miscible PVP blends, with the plasticizers bearing
two or three hydroxyl groups in their molecules (PEG and
glycerol, respectively), the Fox equation fails to account for
the large negative Tg deviations, because the original form
of this equation relates Tg to the weight fractions of blend
components, whereas that is the function of appropriate
reactive groups' fractions (hydroxyls of the plasticizers
and carbonyls in PVP repeat units). The tting parameter,
wppl ; introduced into the Fox equation, provides a direct
measure of the negative Tg deviations and has a well-dened
physical meaning, characterizing the weight fraction of
plasticizer molecules that form two or more hydrogen
bonds with the polymer. In this way, wppl connects the
measured Tg with a variety of important quantities, describing the dynamics of polymerplasticizer hydrogen bonding.
The wppl parameter can be easily determined from the known
Tg and requires no complementary data for evaluating the
dynamics and stoichiometry of polymerplasticizer hydrogen bonding and crosslinking.
This paper generalizes our approach proposed in the rst
paper of this series [6] for the case of PVP blends with PEG
having different chain lengths. The characteristics of the
PVPPEG hydrogen-bonded network are analysed using
the Tg composition dependence. The equilibrium constant
and the energy of PVPPEG association into a crosslinked
supramolecular structure of a hydrogen-bonded stoichiometric complex are calculated as well. The computed results
are compared with the data of independent measurements in
order to demonstrate the validity of the method and to gain a
further insight into both the mechanism of PVPPEG
complexation and the complex structure.
2. Experimental
PVP (Kollidon K-90), Mw 1; 000; 000 g mol21 ; and
PEG ranging in molecular weight from 200 to
36,000 g mol 21, were obtained, respectively, from BASF
and Fluka. Dimethyl ether of PEG-400, DMPEG, was
purchased from Sigma. All polymers were used as received.
The complete details of the basic experimental procedures employed in this work were introduced in our recent
paper [9]. The PVP blends with PEG, covering a full range
of compositions, were prepared by dissolving both polymers
in a common solvent (ethyl alcohol) followed by removing

Table 1
Melting temperatures, Tm, and fusion enthalpies, DHm, of PEG employed in
this work
Polymer

Tm (8C)

DHm (J g 21)

PEG-400
PEG-600
PEG-1000
PEG-36,000
DMPEG-400

5.9
15.8
39.0
69.1
1.9

122.1
98.3
137.4
191.8
106.4

the solvent by drying at ambient temperature and relative

humidity (RH) until weight loss was terminated. Removal
of ethyl alcohol from freshly prepared blends was ascertained by FTIR spectroscopy by the absence of methylene
group stretching vibrations at 2974 and 1378 cm 21 using a
Bruker IFS-113v spectrometer with a resolution of 1 cm 21
after averaging from 128 scans. The freshly prepared blends
and unblended PVP were stored at ambient temperature by
equilibrating to controlled pressure of water vapour in desiccators, maintaining RH 50% at 258C over different time.
In the d.s.c. apparatus the samples were rst quench
cooled with liquid nitrogen from ambient temperature to
21008C over 23 min and then heated up at a rate of
208C min 21 to 2208C. Upon heating, a heat capacity jump
followed by single exotherm coupled with symmetric
endotherm, and high temperature endotherm were normally
observed for the PVPPEG blends. These four transitions
were, respectively, attributed to the glass transition, PEG
cold crystallization, melting, and water thermodesorption
[9]. The Tgs were recorded at half-height of the relevant
heat capacity jumps in d.s.c. heating thermograms with a
Mettler TA 4000/DSC 30 thermoanalyser, calibrated with
indium and gallium. The heats of PEG melting peaks were
computed by constructing linear baselines from the peak
onset to completion and numerically integrated with the
appropriate software supplied by Mettler. All reported
values are the averages of replicate experiments varying
less than 12%. Samples of 515 mg in weight were sealed
in standard aluminium pans supplied with pierced lids so
that absorbed the moisture could evaporate upon heating.
An argon purge (50 ml min 21) was used to avoid moisture
condensation at the sensor. The Tg values for PVP and PEG
employed in this research have been presented in Table 1 of
the rst paper in this series [6].
The weight fractions of crystalline PEG in the blends with
amorphous PVP, wcrPEG, have been calculated as the ratios of
appropriate heats of melting, DHm, of blended PEG to the
reference values for unblended PEG, taking into account the
weight fraction of absorbed water, wH2 O :
wcrPEG

DHm blend
:
DHmPEG 1 2 wH2 O

The DHm reference values of unblended crystalline PEG

varying in molecular weight, as well as that of DMPEG,

M.M. Feldstein et al. / Polymer 42 (2001) 981990

are presented in Table 1. PEG-200 and PEG-300 have not

developed any crystallinity during the thermal cycle
employed.
The content of absorbed water in the blends was determined by weighing the samples before and after the d.s.c.
scans using a Mettler Analytical Balance, AE 240, with an
accuracy of ^0.01 mg. Weight loss of the sample after
scanning was compared to the amount of desorbed water
evaluated from the enthalpy change associated with water
evaporation from the sample by d.s.c. Depending on the
relative humidity of the surrounding atmosphere, PVP
hydration ranged from 6 to 8 wt.%, while PEG contained
less sorbed water (01 wt.%).
Wide angle X-ray diffraction measurements (WAXS)
were performed with ltered Cu Ka radiation using a
DRON-3M diffractometer with an asymmetric focusing
curved quartz crystal monochromator of a primary beam,
equipped with a thermocontrolled cooling camera at
230(^1)8C. Diffraction patterns were recorded in transmission mode. The scattering intensity distribution was
measured with a step-scanning device at the step interval
of 0.058, each at the xed time of 10 s, within the range of
twice the Bragg angles from 5 to 508. Film samples of
unblended PVP and PVPPEG blends over a wide composition range were placed between two PET slides of
0.004 mm in thickness by casting appropriate solutions in
ethyl alcohol followed by their drying at ambient temperature. The thickness of the samples was about 1 mm. Similarly, the hermetic PET cells lled with PEG were also used
for liquid specimens. In the cooling camera the samples
were cooled with liquid nitrogen at a rate of 18C min 21.

983

Fig. 1. The involvement of PEG hydroxyl groups into hydrogen bonding

with PVP repeat units as a function of PEG molecular weight at
[OH]:[PVP] 1.132.

3. Results and discussion

interval between wPEG 0:8 and 0.9. The latter nding

implies that a gradual swelling of the crosslinked PVP
PEG complex in an excess of PEG [6] results in a disruption
of the hydrogen-bonded network and the PVP-chains
release only in a relatively dilute PVP solution in PEG
(1020%). In other words, the range of PEG concentrations
wPEG 0:80:9 outlines the border between a swollen
PVPPEG hydrogen-bonded gel and dilute PVP solution
in liquid PEG. This conclusion has been supported by the
results of independent measurements. As has been recently
shown [12], the activation energy for interdiffusion under
PVP dissolution in liquid PEG-400 reveals a rapid drop with
the increase in PEG concentration within the same region, as
is expected for the gelsolution transition.

3.1. PEG chain length effects on the negative Tg deviations

from the weight-average values

3.2. The inuence of PEG chain length on the degree of

hydrogen bonding

As is shown in the rst paper of this series [6], the value

of negative deviations from the weight-averaged Tg, wppl ;
describes in fact the number of reactive groups in plasticizer
molecules, capable of forming two or three hydrogen bonds
with PVP repeat units through the second and third hydroxyl
groups. Because all PEG have two proton-donating hydroxyl groups at chain ends, wpPEG dened by a modied Fox
equation (2) in Ref. [6] is expected to be independent of the
PEG molecular weight. On the other hand, the concentration
of terminal hydroxyl groups in PEG decreases with the
increase in molecular weight. From this line of reasoning,
only ne or negligible molecular weight effects on the wpPEG
values might be anticipated.
Actually, the obtained wpPEG values are similar for all
PEGs ranging in molecular weight from 200 to
600 g mol 21 and follow the pattern shown in Fig. 3 [6] for
PEG-400, varying inversely with the PEG weight fraction in
the blend, wPEG, and approaching zero in the concentration

The wPEG 1 wpPEG =wPEG ratio describes the involvement

of hydroxyl groups at the PEG chain ends in hydrogen
bonding to PVP units ([3, Eq. (6)]). With the decrease in
PEG concentration the fraction of hydrogen bonded OH
groups in PEG molecules increases tending to 2 at zero
PEG content. This behaviour is shown for PEG-400 in
Fig. 5 in the rst paper of this series [6]. Within the PEGoverloaded region, where [OH]/[PVP] . 1, the number of
hydrogen-bonded OH groups in the PEG molecule is only
slightly inuenced by blend composition, considerably
exceeding unity but tending to this limit at an enormous
PEG excess. Since exible PEG macromolecules forming
two hydrogen bonds with PVP units through both terminal
hydroxyls act as a crosslinker, connecting the PVP chains
into the supramolecular network structure, this nding indicates the stability of the PVPPEG network in comparatively dilute PVP solutions in PEG.
Although all PEG chains bear two reactive hydroxyl

984

M.M. Feldstein et al. / Polymer 42 (2001) 981990

Fig. 2. The plot of the percentage of PEG macromolecules crosslinking

PVP units against the number of OH groups available in the blends per
one PVP carbonyl.

Fig. 4. The content of PVP units (in mole percent) crosslinked by PEG
chains in the blend [OH]:[PVP] 1.132, plotted against the PEG molecular
weight.

groups, their hydrogen-bonding activity, expressed in terms

of wPEG 1 wpPEG =wPEG ratio, is inversely related to the
chain length at a xed concentration of OH groups in the
blend (Fig. 1). As the molecular weight increases approaching 1000 g mol 21, PEG ceases crosslinking PVP units. This
conclusion has been conrmed experimentally. The d.s.c.
traces of PVP blends with PEG-1000 have been shown to
exhibit two glass transitions, which relate to PVP and PEG,
respectively. This indicates the incompatibility of PVP with
PEG-1000 at a molecular level [3].

hydroxyls available in blends per one PVP unit (Fig. 2).

Using the wpPEG =wPEG ratio, the dynamics of hydrogen bonding and crosslinking capability of the PVP units has been
described in terms of the quantities outlined by Eqs. (4) and
(5) presented in the rst paper of this series [6]. Comparison
of the number of hydrogen-bonded PVP repeat units for
PVP blends with PEG-200, -300 and -600 with that for
PEG-400, displayed in Fig. 7 in Ref. [6], highlights the
inuence of PEG chain length on the state of PVP carbonyls. Qualitatively, the dynamics of PVP hydrogen bonding
with PEG-200, -300 and -600 follows the pattern shown by
PEG-400 [6, Fig. 7], however the quantitative distinctions
are evident, particularly in the percentage of PVP units
crosslinked through telechelic PEG chains. To make the
comparison more readily illustrative, the concentration
proles of the amount of PVP units, crosslinked through
PEG chains varying in molecular weight from 200 to 600,
are presented in Fig. 3.
The content of crosslinked PVP units is nearly independent of PEG concentration over a wide composition range
for a variety of PEGs ranged in molecular weight, indicating
the stoichiometry of hydrogen-bonded network PVPPEG
complexes. The most stable, toward swelling with an excess
of PEG, is the PVP complex with PEG-400, while other
PEGs reveal a noticeable reduction in the content of crosslinked PVP units with the increase in PEG concentration.
The crosslinking degree of PVP units in the stoichiometric
complexes, shown in Fig. 4 as a function of PEG molecular
weight, follows the pattern demonstrated in Fig. 1 for the
hydrogen-bonding reactivity of PEG. Both the involvement
of PEG chain-end hydroxyls into hydrogen bonding with
PVP (Fig. 1) and the amount of crosslinked PVP units in
the stoichiometric complex (Fig. 4) are found to relate
inversely to the PEG molecular weight, allowing us to
gain an insight into the structure of the stoichiometric supramolecular PVPPEG network complex.

3.3. PEG molecular weight as a factor controlling the

stoichiometry of the hydrogen-bonded PVPPEG network
The ratio wpPEG =wPEG denes the weight fraction of PEG
macromolecules crosslinking PVP chains by hydrogen
bonding through both terminal hydroxyls. The crosslinking
capability of PEG chains increases with the decrease in their
length as well as with the reduction in the number of PEG

Fig. 3. Impact of PEG molecular weight upon the compositional behaviour

of the percentage of PVP units crosslinked through PEG chains.

M.M. Feldstein et al. / Polymer 42 (2001) 981990

985

the amount of bound (noncrystallizing) PEG from the

appropriate intercepts at wcrPEG 0: The binding degree
obtained in this manner is found to be wPEG < 0:526 for
PEG-400 and ,0.364 (g PEG/g blend) for PEG-600 (Fig.
5). The obtained values correspond to ,0.308 and ,0.106
bound macromolecules of PEG-400 and PEG-600 per one
PVP repeat unit available in the blends. Accordingly, in
total ,61.6 and 23.2% of PVP repeat units have been hydrogen bonded to terminal hydroxyls in the complexes with
PEG-400 and PEG-600, respectively. In full agreement
with the data in Fig. 4, determined by the analysis of the
behaviour of wpPEG vs. composition, the binding degrees of
PEG-400 and PEG-600, found from the intercepts in Fig. 5,
relate to each other as the reciprocals of their molecular
weights:
Fig. 5. The weight fraction of crystalline PEG, wcrPEG, in blends with amorphous PVP as a function of PEG weight fraction, wPEG, for the PEG varied
in molecular weight from 400 to 36,000 g mol 21 and for dimethyl ether of
PEG-400, DMPEG.

The stoichiometric composition of the PVP complex with

PEG-400, established via the analysis of the compositional
behaviour of wpPEG ; has been earlier shown to be in reasonable agreement with the data of independent measurements,
such as FTIR spectroscopy and the relation of the amount of
noncrystallizing PEG-400 to blend composition, determined
from d.s.c. heating thermograms [5,6]. In order to give additional credit to the wpPEG -based approach for describing
correctly the stoichiometry of the PVPPEG complex, the
composition dependence of PEG crystallinity in the blends
with amorphous PVP is considered in Fig. 5. PEG-200 and
PEG-300 have been found to develop no crystallinity during
the quench-cooling/heating cycle employed in this research
for d.s.c. measurements.
In Fig. 5 the weight fractions of crystalline PEG, wcrPEG,
are plotted against total PEG content, wPEG. If amorphous
PVP merely diluted crystalline PEG and did not interfere
with its crystallinity through a strong favourable interaction
such as hydrogen bonding, the plot would represent a
straight line connecting the origin and the crystallinity
degree of pure PEG, dened as wcrPEG 1: This ideal dilution behaviour is in fact typical of the PVP blends with those
PEGs which provide no or negligible contribution of endchain hydroxyls (DMPEG-400 and comparatively longchain PEG having molecular weights of 1000 g mol 21 and
higher). In contrast, the disappearance of melting peaks in
the d.s.c. scans due to free PEG-400 and PEG-600 within the
PVP-overloaded blends, and the lack of equivalence
between the crystallinity (wcrPEG) and the total amount of
PEG in the PEG-overloaded systems, are the proofs for
PEG being hydrogen bonded with PVP in the blends
enriched by amorphous PVP. This has been shown earlier
analysing in detail both kinetic and thermodynamic contributions to the depression of PEG-400 crystallinity in the
blends with PVP [7].
The data in Fig. 5 provide a feasible method of estimating

wPEG-600 wcrPEG 0
0:364
MWPEG-400

0:69 <
MWPEG-600
wPEG-400 wcrPEG 0
0:526

400
0:67:
600

From the Tg composition relationship, the total amount

of crosslinked PVP units and that hydrogen bonded to the
PEG macromolecules, having one of the ends free of hydrogen bonding with PVP, can be obtained as the sum of M1 and
M2 quantities, dened by Eqs. (4) and (5) in Ref. [6]. At the
PEG concentration ([OH]:[PVP] < 0.30) corresponding to
the onset of the limiting values of the content of PVP units
crosslinked by PEG-400 (,20%) and PEG-600 (,10%), we
get M1 1 M2 < 62% for PVP bonding to PEG-400 [6, Fig.
7] and M1 1 M2 < 20% for PEG-600. The values found
match closely to that obtained from blend crystallinity.
Consequently, the analysis of blend crystallinity allows estimating the total amount of bound PEG, but provides no
distinctions between strongly (crosslinked) and loosely
bound PEG. The varieties of bound PEG may be however
determined with both the wpPEG -based approach and FTIR
spectroscopy.
The linearity of the plots for PEG-400 and PEG-600 crystallinity in Fig. 5 reveals the invariability of the appropriate
intercepts over PEG-overloaded composition range, i.e. the
stoichiometry of PVP hydrogen-bonded complexes with
PEG-400 and PEG-600. At an excess of PEG, only wpPEG analysis provides an unambiguous measure of the amount of
crosslinked PVP units, dened earlier as strongly bound [6].
At the same time, the FTIR data for PVP blends with PEG400, presented in Table 2 of the rst paper in this series [6],
demonstrate that even in comparatively dilute (15%) PVP
solution in liquid PEG-400 (at [OH]:[PVP] 2.8) both the
total amount of hydrogen-bonded PVP units (56 mol%) and
that strongly bound (26.2%) remain nearly invariant with
changing composition. Thus, the analysis of PEG fusion
enthalpy and FTIR spectroscopy conrm the applicability
of the wpPEG -based approach for describing the stoichiometry
of PVPPEG hydrogen-bonded complexes using the Tg
composition relationship. In a subsequent section of this

986

M.M. Feldstein et al. / Polymer 42 (2001) 981990

Fig. 6. The plot of hydrogen-bonded crosslink density, n H/V0, against the

composition of PVP blends with PEG varied in molecular weight from 200
to 600 g mol 21.

paper we will present the WAXS data supporting the data

shown in Fig. 5 for the crystallinity of PVP blends with
PEG-400 and PEG-600.
3.4. Inuence of PEG chain length on network density in
hydrogen-bonded stoichiometric PVPPEG complexes
Mixing glassy PVP with short chain PEG is a two-stage
process [13]. At the rst stage, when comparatively a small
amount of plasticizer is added to PVP, the Tg falls dramatically over 2208C. This stage may be treated as PVP plasticization due to hydrogen bonding of carbonyls to PEG
terminal hydroxyls. Since the hydroxyls are located at the
PEG chain ends, and because within PVP-overloaded
blends both PEG terminal hydroxyls are found to be
involved in hydrogen bonding by crosslinking PVP units
(Fig. 3), the relatively short PEG chains serve as spacers,
creating free volumes between longer PVP chains [3]. The
fraction of free volume may be evaluated from the blends'

Fig. 7. The average molecular weight Mc1 of the PVP chain segment
between neighbouring hydrogen-bonded PEG crosslinks, and the average
number of polymer repeat units in this segment, NPVP, in the course of PVP
PEG network complex swelling in excess PEG.

Tg using a combination of the Doolittle and Williams

LandellFerry (WLF) equations [14].
The second stage of the PVPPEG mixing process
reveals comparatively insignicant Tg reduction, that is indicative of weaker PVPPEG interaction. This stage has been
dened as a gradual swelling and dissolution of the PVP
PEG crosslinked complex in an excess of PEG [13]. The
boundary between the rst and second stages corresponds to
the stoichiometric composition of the PVPPEG complex.
Based on these arguments, one could expect that the
longer the chains of the PEG crosslinker, serving as a spacer
between the PVP and PEG chains in the network complex,
the larger the free volume, i.e. the greater the Tgs decrease
under PVPPEG mixing. However, this is not the case for
PVP blends with different PEG where the effect of PEG
molecular weight on the blends' Tg has been found to be
comparatively insignicant. The fractional free volume in
PVP blends with PEGs, ranging in molecular weight from
200 to 600 g mol 21, is nearly unaffected by the PEG chain
length. The reason for this behaviour is the compensating
effect of the PEG chain length on the density of hydrogenbonded crosslinks in the PVPPEG network.
Knowing wpPEG and applying Eq. (6), presented in the rst
paper of this series [6], make feasible estimating the density
of hydrogen-bonded entanglements, nH =V0 ; in the PVP
PEG network (Fig. 6). The average critical molecular
weight of the PVP chain segment between neighbouring
junctions of the hydrogen-bonded network, Mc1 ; may be
evaluated using a simple relation:

nH
r
1
V0
Mc

where r is a blend density. The Mc1 behaviour is illustrated

in Fig. 7 along with an average number of PVP repeat units
in the chain segment between hydrogen-bonded junctions,
NPVP. The shorter the chains of the PEG crosslinker the
larger the density of the network. The maximum of the
PVPPEG network density (Fig. 6) corresponds to the
onset of the steady-state content of the crosslinked PVP
units (Fig. 3), which in turn relates to the minimum PEG
concentration providing the formation of a stoichiometric
complex. In full agreement with the two-stage model of the
PVPPEG mixing mechanism, subsequent addition of PEG
results in a gradual decrease of the PVPPEG network
density (Fig. 6), which is typical for swelling gels. Both
the Mc1 and NPVP quantities are the linear functions of the
blend composition (Fig. 7). The longer the PEG chain the
easier the network swells in an excess of PEG, i.e. the longer
the PVP chain segment between neighbouring crosslinks.
The PVPPEG-200 network swelling is accompanied by
relatively slight growth in the length of the PVP chain
segment between neighbouring hydrogen-bonded network
junctions, compared to PEG of higher molecular weight.
The swelling of the PVP complex with PEG-200 results,
however, in a signicant reduction of network density
(Fig. 6). This is a direct consequence of the rise in PVP

M.M. Feldstein et al. / Polymer 42 (2001) 981990

Fig. 8. X-ray diffraction patterns of PVP, PEG-400, and their blends over
entire composition range at 2308C: (1) 100% PVP; (2) 36% PEG; (3) 53%
PEG; (4) 69% PEG; (5) 85% PEG; (6) 92% PEG; (7) 100% PEG.

crosslinking degree as the length of the PEG chain decreases

(Figs. 3 and 4).
The effects of PEG chain length on the parameters of the
hydrogen-bonded network, displayed in Figs. 6 and 7, are
analysed by including the molecular weight terms into Eq.
(6) [6], employed for evaluation of n H/V0, Mc1 and NPVP
using wpPEG : Therefore, it would of great importance to
compare the established PEG chain length effects to the
data of independent measurements. Figs. 8 and 9 demonstrate the contributions of PEG molecular weight to the
structure of PVP solutions in PEG-400 and PEG-600
revealed by X-ray diffraction patterns.
The X-ray diffraction pattern of unblended PVP shows
two broad peaks at 11.3 and 21.38, which correspond to a
period of ,0.718 and ,0.417 nm, respectively [15].
According to Killian and Boueke [16], the rst halo of the
amorphous polymer is most likely due to interchain interference, while the second one corresponds to the scattering
from side chains. Hence, the rst peak characterizes the
periodicity of PVP backbones arrangement, whereas the
second halo relates to the scattering from pyrrolidone
rings in PVP side chains. The X-ray scattering curves of
PEG-400 and PEG-600 at 2308C exhibit two sharp peaks
of crystalline polymer at 19.3 and 23.48. Dilution of crystalline PEG with amorphous PVP results in the depression of
PEG crystallization. No crystalline phase occurs in the
blends containing up to 50 wt.% of PEG-400 and 36 wt.%
of PEG-600, which is in excellent agreement with the data
of d.s.c. measurements presented in Fig. 5. Note, that PEG600 crystallization begins at lower concentration compared
to PEG-400 behaviour. This nding has been assigned to the

987

Fig. 9. X-ray diffraction patterns of PVP, PEG-600, and their blends over a
wide composition range at 2308C: (1) 100% PVP; (2) 36% PEG; (3) 53%
PEG; (4) 69% PEG; (5) 100% PEG.

higher binding degree of PEG-400 in the stoichiometric

complex with amorphous PVP (Fig. 3). The PVPPEG
blending also leads to a rapid decrease in the intensity of
the halo at 11.38, which is an evidence for the loss of
contacts between the repeat units of the neighbouring PVP
chains. In the blend containing 36 wt.% of PEG, where the
stoichiometric complex is just formed, the PVP interchain
contacts are still manifested by a shoulder on the scattering curves around 11.38. Subsequent swelling of the
PVPPEG network in an excess of PEG causes the
decrease in PVP interchain contacts at ,50 wt.% of
PEG-400, whereas these are still observed in the blends
containing 69 wt.% of PEG-600. This is an unambiguous

Fig. 10. The composition dependence of the constant of PVP hydrogenbonded crosslinked stoichiometric complex formation, KH1 (mol g 21), with
PEG varied in molecular weight from 200 to 600 g mol 21.

988

M.M. Feldstein et al. / Polymer 42 (2001) 981990

consequence of a sparse PVPPEG-600 hydrogen bonded

network. The contribution of PEG spacer length to the
suppression of PVP interchain contacts within the network
complex is thought to be negligible compared to that of the
length of the PVP chain segment between neighbouring
hydrogen-bonded network junctions, controlled by crosslinking degree. Actually, the PEG chains are highly exible.
The PEG chain segment has been reported to consist of 45
monomer units [17]. This is twice as low as the total number
of oxyethylene units in the PEG-400 chain, and trice as that
in the PEG-600 chain. Thus, the results for the PVPPEG
network density, obtained from wpPEG ; are consistent with
the structure of the PVPPEG hydrogen-bonded complex
established with the WAXS technique.
3.5. Equilibrium constants and energy of PVPPEG
association into hydrogen-bonded crosslinked
stoichiometric complex
Like an individual chemical compound, the stoichiometric PVPPEG complex demonstrates a composition
independent of the concentration of the parent components
in the blend. This is the case not only at the stage of crosslinked complex swelling and dissolution in an excess of
PEG as is evident from the data in Fig. 3, but within the
entire composition range. The PVP-overloaded blends
(wPEG , 0.36) have been found to separate into the stoichiometric PVP complex and the parent PVP, as is indicated by
the occurrence of two Tgs around 230 and 508C, which
correspond to the complex and hydrated PVP, respectively
[5]. With the rise of PEG concentration up to 36 wt.%, the
change in heat capacity between the rubbery and glassy
state, DCp, accounted for the rst transition grows, whereas
the latter decreases. For the blends with wPEG . 0.36 only
lower temperature glass transition is observed from d.s.c.
traces. Consequently, the process of hydrogen-bonded
crosslinked PVPPEG complex formation may be treated
as a chemical equilibrium, described by the constant KH1
(mol g 21), which can be expressed in terms of wpPEG and

wPEG:
2wpPEG

KH1
2wPEG 2

wpPEG

wPVP
2wpPEG
2
MWPVP
MWPEG

where MWPVP is the molecular weight of the PVP repeat

unit (111.14 g mol 21) and MWPEG is the molecular weight
of the PEG macromolecule (200, 300, 400 and
600 g mol 21).
In accordance with Le Chatelier's principle, a small
change in composition, such as that caused by the addition
or removal of one of the reactants or the product, shifts the
position of the equilibrium accordingly. Consequently, KH1
varies with blend composition, as is illustrated in Fig. 10.
KH1 is a measure of the tendency of the complexation to
occur. If it is a large and positive number, the concentration
of the PVPPEG complex at equilibrium is large compared
to those of active reactants, [OH] and [PVP]. This is the case
for the blends containing an excess of PVP, i.e. at the stage
of PVPPEG complex formation, whereas at the stage of
the complex swelling and dissolution in an excess of PEG,
KH1 drops signicantly tending to zero at considerable PEG
overloading. The effect of PEG molecular weight on KH1 is,
the shorter PEG chain the stronger the association (Fig. 10).
The equilibrium constant of the PVPPEG stoichiometric
complex formation provides a bridge to the standard free
energy change for PVPPEG association into hydrogenbonded crosslinked stoichiometric complex, DG1
H : The
latter relates to the equilibrium constant through the fundamental van't Hoff equation:
1
DG1
H 2RT ln KH :

The compositional proles of the free energy for PVP

curing by hydrogen bonding through short PEG chains are
presented in Fig. 11. Large KH1 values correspond to the
large negative DG1
H magnitudes. The PVPPEG complex
formation is an exothermic process and the gain in energy
is greater at the rst stage of PVP plasticization due to
hydrogen bonding with PEG ([PEG]:[PVP] , 0.5 or
[OH]:[PVP] , 1), than at the second stage of PVPPEG
complex swelling and dissolution in an excess of PEG.
The shorter the PEG chain the larger the gain in energy
under PVPPEG complex formation. This fundamental
conclusion leads to understanding the reason for nonequimolar stoichiometry of PVPPEG hydrogen bonding.
3.6. Molecular structure of stoichiometric complex and
PVPPEG hydrogen-bonding thermodynamics

Fig. 11. Effects of blend composition and PEG molecular weight on the
21
standard free energy change DG1
H ; kJ=mol for hydrogen-bonded stoichiometric network PVPPEG complex formation at 208C.

Every PEG macromolecule bears only two hydroxyl reactive groups at opposite chain ends. The energy of strong PVP
hydrogen-bonds with PEG-400, measured from FTIR spectra,
has been found to be 21.42 kJ mol 21 [4], whereas the data in
Fig. 11 give the values of the order of 1013 kJ mol 21 and
show the DG1
H variation with composition. This apparent

M.M. Feldstein et al. / Polymer 42 (2001) 981990

Fig. 12. The number of PVP repeat units per one PEG monomer unit in the
mesh between neighbouring hydrogen-bonded crosslinks plotted against
the PEG molecular weight. The NPVP/NPEG ratio is evaluated from appropriate slopes of linear plots in Fig. 7.

discrepancy is explicable, because the total free energy of

mixing has been embedded in a specied Tg value, which
includes both enthalpy and entropy contributions [18], as
well as that due to PEG self-association [19]. At the same
time, the FTIR data characterize rather an enthalpy of a
single hydrogen bond. In our particular case, by the denition of the PVPPEG complex formation equilibrium
constant, outlined by Eq. (4), DG1
H represents only a constituting part of the total free energy of mixing, which accounts
for PVP hydrogen-bond crosslinking with PEG. Compared
to the magnitude obtained from the FTIR data, DG1
H appears
to be smaller due to a negative change in entropy under
PVPPEG hydrogen-bonded association into a supramolecular network structure.
Hydrogen bonding of PEG terminal hydroxyls to the
carbonyls in PVP repeat units causes inevitable loss in
entropy due to the decrease in the number of degrees of
freedom of the PEG chains attached to longer PVP macromolecules through two hydrogen bonds at both PEG chain
ends. Unblended PVP is in a glassy state and its segmental
mobility is essentially frozen. The PVP plasticization (due
to hydrogen bonding with PEG) results in the growth of
mobility since the lower the Tg the larger the free volume
and higher the mobility. Within the PEG-underloaded
region ([PEG]:[PVP] , 0.5) the loss in entropy due to
PEG chains immobilization can be still counterbalanced
by the increase in the mobility of PVP chain segments
between neighbouring cross-links, caused by free volume
formation under PVPPEG hydrogen bonding. It means
that PVP curing by PEG short chains is thermodynamically
allowed only if the length of the PVP chain segments
between neighbouring hydrogen-bonded network junctions
is above some critical value which depends on the length of
the crosslinking PEG chains. The longer the PEG chains the

989

larger the loss in entropy due to their immobilization and

greater must be the counterbalancing increase in PVP
segmental mobility. The latter requires a longer PVP
chain segment between neighbouring crosslinks, i.e. a sparser hydrogen-bonded PVPPEG network and, consequently, the lower crosslinking degree of PVP units in the
hydrogen-bonded stoichiometric complex. As PVP chains
have already obtained sufcient mobility in the exible
network of the PVPPEG complex, they become unable
to compensate the further loss in PEG chain entropy during
the course of hydrogen bonding. For this reason the PEG
macromolecules cease to form hydrogen bonds with PVP
units through the second terminal hydroxyl groups, and the
crosslinked complex has achieved its stoichiometric composition. All these phenomena have been in fact observed in
PVPPEG systems as is shown in this study.
Although the length of the PVP segment between neighbouring junctions of the hydrogen-bonded network is
mainly determined by the length of crosslinking PEG
chain, the ratio between these lengths is supposed to develop
some degree of misbalance with the increase in PEG molecular weight due to growing contribution of van der Waals
interactions between neighbouring PEG chains in the hydrogen-bonded network. If merely the entropic contribution
had governed the frequency of PEG hydrogen-bonded
crosslinks, the NPVP/NPEG ratio would be independent on
PEG chain length. However, as Fig. 12 demonstrates, the
ratio of the length of the PVP-segment between neighbouring junctions to that of PEG chain increases monotonously
with PEG molecular weight within the range from 200 to
400 g mol 21. At the same time, comparatively longer PEG
spacers, exemplied by PEG-600, reveal signicantly sparser networks, i.e. the PVP chain segment between neighbouring crosslinks becomes longer. The nonlinear character
of the relationship in Fig. 12, coupled with the linearity of
the dependence of the PVP crosslinking-degree on the PEG
molecular weight (Fig. 4) is indicative of an enthalpy contribution to the process of PEG chains distribution within the
crosslinked stoichiometric complex. Indeed, the longer the
PEG chains and greater the contribution of the PEG interchain interactions, the stronger is the trend of PEG assembling into hydrogen bonding clusters of extended chains
within the PVPPEG network. Hydrogen bonding of PVP
with PEG cannot interfere with PEG clusterization. By this
means, comparatively shorter PEG chains hydrogen bonded
to PVP units through terminal hydroxyl groups appear to be
more or less uniformly distributed along PVP chains. In
contrast, longer PEG chains exhibit a tendency of association. As a result, the PVP macromolecules are crosslinked
by the clusters of extended PEG chains. Such clusters
resemble PEG crystallites, where terminal hydroxyl groups
are exposed outward crystalline grains. At this instance, the
molecular structure of the PVPPEG stoichiometric
complex can be treated as PVP chains adsorbed on the
end faces of PEG crystallites through hydrogen bonding
of the PVP units to the PEG end-chain hydroxyls. This

990

M.M. Feldstein et al. / Polymer 42 (2001) 981990

conjecture has been in fact supported by our nding that

PEG blends with a small amount of PVP demonstrate significantly higher melting temperature than that found for
unblended PEG [20].
4. Conclusions
The observed relation of the negative Tg deviations from
the weight-average values to the weight fraction of PEG
macromolecules, which form with PVP repeat units two
hydrogen bonds, provides a convenient tool to evaluate
the stoichiometry of the PVPPEG hydrogen-bonded
complexes, their crosslinking degree and energy of formation, as well as the density and molecular structure of the
hydrogen-bonded network. The analysis of PEG chain
length effects on a variety of PVPPEG hydrogen-bonding
characteristics allows obtaining an insight into the thermodynamics of polymer interaction over the entire composition range. The nonequimolar stoichiometry of PVPPEG
hydrogen-bonded network complexes is a result of specic
balance between the loss in entropy due to the immobilization of short PEG chains in the course of their hydrogen
bonding to PVP units through both chain ends, and the
entropic gain due to the increase in mobility of PVP-chain
segments between neighbouring hydrogen-bonded junctions
in the network. The competition between these two factors
eventually determines the observed increase in the PVP
crosslinking degree with the decrease in PEG molecular
weight and denes the nonequimolar stoichiometry of the
formed hydrogen-bonded complexes.

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

Acknowledgements
This research was in part made possible by Award No.
RC1-2057 of the US Civilian Research & Development
Foundation for the Independent States of the Former Soviet
Union (CRDF). We express our appreciation to Professors
Alexander Yu. Grosberg, Ronald A. Siegel, Anatoly E.
Chalykh, and Dr Elena Dormidontova for their helpful
discussion and comments.
References
[1] Cesteros LC, Quintana JR, Fernandez JA, Katime I. Miscibility of
poly(ethylene oxide) with poly(N-vinyl pyrrolidone): DMTA and
DTA studies. J Polym Sci, Polym Phys Ed 1989;27:256776.
[2] Buehler V. Kollidon: polyvinylpyrrolidone for the pharmaceutical
industry. 3rd ed. Ludwigshafen: BASF, 1996 (p. 1920).
[3] Feldstein MM, Lebedeva TL, Shandryuk GA, Kotomin SV, Kuptsov

[13]

[14]
[15]
[16]
[17]
[18]
[19]
[20]

SA, Igonin VE, Grokhovskaya TE, Kulichikhin VG. Complex formation in poly(N-vinyl pyrrolidonepoly(ethylene glycol) blends.
Polym Sci 1999;41(8):85466.
Lebedeva TL, Igonin VE, Feldstein MM, Plate NA. H-bonding poly(ethylene glycol) to poly(N-vinyl pyrrolidone) within an adhesive
hydrogel matrix for transdermal drug delivery. Proc. Int. Symp.
Control. Release Bioact. Mater., vol. 24, 1997. p. 4478.
Feldstein MM, Lebedeva TL, Shandryuk GA, Igonin VE, Avdeev
NN, Kulichikhin VG. Stoichiometry of poly(N-vinyl pyrrolidone)
poly(ethylene glycol) complex. Polym Sci 1999;41(8):86775.
Feldstein MM, Shandryuk GA, Plate NA. Relation of glass transition
temperature to the hydrogen bonding degree and energy in poly(Nvinyl pyrrolidone) blends with hydroxyl-containing plasticizers: 1.
Effects of the number of hydroxyl groups in plasticizer molecule.
Polymer 2001;42(3):9719.
Feldstein MM, Kuptsov SA, Shandryuk GA. Coherence of thermal
transitions in poly(N-vinyl pyrrolidone)poly(ethylene glycol)
compatible blends: 2. The temperature of maximum cold crystallization rate versus glass transition. Polymer 2000;41(14):533948.
Feldstein MM, Chalykh AE, Chalykh AA, Plate NA. Quantitative
relationship between molecular structure and adhesion of PVP
PEG hydrogels. Polym Mater Sci Engng 1999;81:4656.
Feldstein MM, Shandryuk GA, Kuptsov SA, Plate NA. Coherence of
thermal transitions in poly(N-vinyl pyrrolidone)poly(ethylene
glycol) compatible blends: 1. Interrelations among temperatures of
melting, maximum cold crystallization rate and glass transition. Polymer 2000;41(14):532738.
Feldstein MM, Bairamov DF. Effects of chain orientation, free
volume and interaction on glass transition in poly(N-vinyl pyrrolidone)poly(ethylene glycol) blends involving a stoichiometric hydrogen-bonded network complex. Polym Mater Sci Engng 2000;82:365
6.
Fox TG. Inuence of diluent and copolymer composition on the glass
transition temperature of a polymer system. Bull Am Phys Soc
1956;1:123.
Igonin VE, Makarova VV, Avdeev NN, Feldstein MM, Kulichikhin
VG. Optical microinterference measurements of polymer miscibility,
dissolution and interdiffusion. Proc. Int. Symp. Control. Release
Bioactive Mater., vol. 26, 1999. p. 3912.
Feldstein MM. A two-stage mechanism of poly(N-vinyl pyrrolidone)
mixing with short-chain poly(ethylene glycol). Proc. Int. Symp.
Control. Release Bioact. Mater., vol. 25, 1998. p. 8489.
Ferry JD. Viscoelastic properties of polymers. 2nd ed. New York:
Wiley, 1970 (chap. 11).
Guinier A. Theorie et technique de la radiochristallographie, 2nd ed.
Paris: Dunod, 1956 (chap. 12, paragraph 3).
Killian HG, Boueke K. Rontgenographische Strukturanalyse von
amorphem Polystirol. VI. J Polym Sci 1962;58(16):31133.
Bershtein VA, Egorov VM. Differential scanning calorimetry of polymers. New York: Horwood, 1994 (chap. 1).
Wu J. The glassy state, ideal glass transition, and second-order phase
transition. J Appl Polym Sci 1999;71:14350.
Coleman MM, Graf JF, Painter PC. Specic interactions and the
miscibility of polymer blends. Lancaster: Technomic, 1991 (chap. 4).
Feldstein MM, Kuptsov SA, Shandryuk GA, Plate NA, Chalykh AE.
Coherence of thermal transitions in poly(N-vinyl pyrrolidone)poly(ethylene glycol) compatible blends: 3. Impact of sorbed water upon
phase behaviour. Polymer 2000;41(14):534959.