Full text of DOI:10.1002/pi.1351

Polymer International
Polym Int 52:1647–1652 (2003)
DOI: 10.1002/pi.1351
Solid-state NMR relaxometry study of phenolic
resins†
Peter Adriaensens, Robby Rego, Robert Carleer, Ben Ottenbourgs and Jan Gelan∗
Limburg University, Instituut voor Materiaalonderzoek (IMO), Departement SBG, Universitaire Campus, Gebouw D, B-3590
Diepenbeek, Belgium
1
Abstract: H wideline and 13C CP/MAS spin-lattice (T1H) relaxation solid-state NMR experiments have
been performed in order to investigate the cure state of phenolic novolac resins. Due to the process of spin
diffusion, a single averaged T1H decay time was found which was sensitive to the cure state of the network
structure. Measurements as a function of temperature revealed that mobile end-groups and branches
determine the most efficient pathway of T1H relaxation. Combined with FTIR measurements, it is shown
that rearrangements of the hydrogen bond structure take place upon heating–cooling. There are strong
indications that the process of disruption–reformation of the hydrogen bonds is not completely reversible.
 2003 Society of Chemical Industry
Keywords: molecular dynamics; curing; hydrogen bonds
resins, probably due to the complexity of interpreta-
INTRODUCTION
1
tion of H wideline NMR data5–9 as the information
Phenolic resins, produced in the reaction of phenol and
formaldehyde, have a wide variety of applications as
impregnants and adhesives.1 Novolac thermoharders
are prepared by thermal curing of novolac prepolymer
resins in the presence of a methylene crosslinker such
as paraformaldehyde and consist of rigid, crosslinked,
insoluble polymer networks.
of all protons of the polymer is superimposed into a
broad envelope. As a result, the analysis of such decay
curves can be very complex. Despite these complica-
tions, in this paper attention will be focused on fast
1H wideline T1H decay time experiments as, in these
rigid novolac networks, the process of spin diffusion
forces the T1H relaxation times to be averaged out to
a single decay time. Moreover, the initial complexity
of interpretation can be strongly reduced by means of
additional 13C cross-polarization magic angle spinning
(CP/MAS) NMR experiments, since its chemical shift
selectivity permits investigations of the relaxation of
different constituents of the network independently.
Attention will be focused on the status of the hydro-
gen bond structure, which plays an important role in
the dynamics of the network structure.
While solution and solid-state 13C NMR spec-
troscopy can be implemented to examine the func-
tional group content and chemical structure of novolac
resins,2,3 the segmental molecular mobility of these
resins, which is also correlated with the physico-
chemical properties, can be investigated by applying
1H wideline solid-state NMR relaxometry. This tech-
nique allows measurement of the resin cure kinetics by
monitoring relaxation decay times as a function of cure
time.4 Due to the high sensitivity of the H nucleus,
1
even at low fields, these experiments are fast enough to
eventually perform at-line or on-line NMR relaxom-
etry. In particular, the spin-lattice decay time T1H is
an especially useful parameter to study the molecular
mobility of novolac resins as a function of the tem-
perature or the cure state. Moreover, due to efficient
spin diffusion in such rigid polymer networks, T1H is
averaged out over a distance of tens of nanometers,
making it a volume property.
EXPERIMENTAL
Materials
The high-ortho content novolac resin used in this study
was prepared by reacting 1500 g of commercial grade
phenol with 685 g of 50% formaldehyde solution in
the presence of a bivalent metal acetate (0.35 wt%
ZnAc2) catalyst in the pH range 4–6. Relaxometry
experiments were carried out on high-ortho novolac
samples which were obtained by mixing and curing
Only a small amount of information is available in
literature concerning proton relaxometry of phenolic
∗ Correspondence to: Jan Gelan, Limburg University, Instituut voor Materiaalonderzoek (IMO), Departement SBG, Universitaire Campus,
Gebouw D, B-3590 Diepenbeek, Belgium
E-mail: Jan.Gelan@luc.ac.be
†Poster presentation: paper presented at the Polymer Chemistry for the Design of New Materials Conference, 19–20 September 2002,
Ghent, Belgium
Contract/grant sponsor: Flemish Institute for the Promotion of Scientific–Technological Research in Industry (IWT)
Contract/grant sponsor: Belgian Government Interuniversitaire attractiepolen (IUAP)
(Received 26 July 2002; revised version received 22 November 2002; accepted 2 January 2003)
Published online 11 September 2003
 2003 Society of Chemical Industry. Polym Int 0959–8103/2003/$30.00
1647

P Adriaensens et al
the high-ortho novolac prepolymer with 11 wt% of
paraformaldehyde at 180 ◦C and 150 bar for different
time periods varying from 2 to 60 min. After cooling
down in a desiccator, all samples were ground
to fine powders and preserved in dry atmospheric
conditions (desiccator).
RESULTS AND DISCUSSION
T1H relaxation decay time analysis
The T1H relaxation time, of the order of seconds,
is sensitive to high frequency molecular motions,
of the order of hundreds of MHz and, reveals
information about the molecular mobility and/or
molecular morphology of a polymer network. Since
the T1H relaxation behaviour of cured phenolic resins
will be discussed at temperatures well below the glass
transition temperature, these high-frequency motions
only arise from segmental chain motions of mobile
end-groups and branched side-chains11 since the
polymer network consists of a rigid crosslinked matrix.
In general, molecular motions are characterized by
a distribution of correlation times τc. As shown in
Fig 1, a minimum appears in the log T1H versus log τc
correlation diagram. At this minimum the spectral
density of molecular motions with a frequency of the
order of hundreds of MHz (proton Larmor frequency)
is optimal. At the left side of this minimum, the so-
called high-temperature side of the T1H minimum,
the higher frequency motions of small molecules,
mobile end-groups and side-chains are situated and
the T1H value is magnetic field independent. For rigid
molecules, the T1H relaxation time is situated at the
right side of this minimum and becomes magnetic
field dependent.
It must, however, be noted that, in rigid polymer
systems, the T1 and T1ρ relaxation times of abundant
nuclei, such as protons, are determined not only
by dynamical phenomena but also by the process
of spin diffusion. In rigid systems, spin diffusion
(magnetization transfer via energy-conserving spin
flip-flops) is very efficient due to the strong dipolar
proton–proton couplings. Because of spin diffusion
the T1H relaxation time becomes a ‘volume property’,
since it becomes averaged out over a distance of several
tens of nm. The spin diffusive path length (L) over
which spin diffusion takes place in a certain time t can
be calculated with the following equation:11,12
NMR measurements
Proton wideline solid-state measurements were per-
formed on a Varian Inova 400 spectrometer in a
dedicated wideline probe equipped with a 5-mm
coil. Samples were placed in 5-mm Pyrex tubes fit-
ted tightly with Teflon stoppers. T1H measurements
were accomplished by placing an inversion recovery
segment in front of a ‘solid echo’ pulse sequence
(180◦xꢀ —t—90x◦ꢀ —τ/2 − 90y◦ꢀ -τ/2-acquire) developed
by Powels and Strange10 to overcome the effects of
dead-time of the receiver, with an echo delay (τ) of
8 µs. A 90◦ pulse width of 2 µs was used and the
evolution time t was varied between 0.001 and 30 s.
The determination of T1H via the carbon resonances
was accomplished by means of inversion recovery
13C CP/MAS experiments on a 200 MHz spectrom-
eter and by using a spinning speed of 3.5 kHz, in
ceramic Si3N4 rotors, a π/2 pulse width of 8.4 µs, a
cross-polarization contact time of 2 ms and a spin-lock
field B1 of 40 kHz. The evolution time t was varied
between 0.001 and 25 s. Ethylene glycol was used to
calibrate the temperature, the magic angle was set with
KBr and the Hartman–Hahn condition was adjusted
using the aromatic signal of hexamethylbenzene. The
latter was also used to calibrate the chemical shift
(132.1 ppm) scale. A preparation time of five times
the longest T1H was always respected. Quantitative
functional group information was obtained by means
of 13C CP/MAS contact time studies in which the
contact time was varied between 2.5 and 30 ms.
The T1H decay times were obtained by analyzing
the signal intensity as a function of the variable
inversion recovery evolution time according to the
following equation:
ꢀ
L = 6 DT1H
(2)
−t
M(t) = M0(1 − 2 × eT
)
(1)
1H
in which M(t) and M0 are the magnetization at time t
and at equilibrium, respectively,
T1
FTIR measurements
The FTIR spectra, taken as a function of temperature
between 20 and 180 ◦C, were recorded on a Bruker
IFS 48 infrared spectrometer with conventional KBr
pellets. All measurements were signal averaged by 32
T2
scans and had a resolution of 4 cm−1
.
-12 -11 -10 -9
-8
-7
-6
-5
-4
-3
-2
-1
TGA measurements
log τc
TGA measurements are performed on a General
V4.1C DuPont 2000 thermogravimetric analyser with
a continuous nitrogen flow of 50 ml min−1 and a
Figure 1. Dependency of the T1 and the T2 relaxation decay times
upon the correlation time τc of molecular motion and the magnetic
heating rate of 10 ◦C min−1
.
field strength (
400 MHz, - - - - 20 MHz).
. . . . . . .
1648
Polym Int 52:1647–1652 (2003)

Solid-state NMR relaxometry of phenolic resins
in which
D
is the spin diffusion coefficient,
bridges (20–45 ppm) as well as the aromatic signals
(110–160 ppm), exhibited the same T1H decay
time. This means that spin diffusion took place
efficiently over a distance of 30–60 nm within the
polymer network. The results of the 1H wideline
measurements, obtained on a 400 MHz spectrometer,
were in complete agreement with those obtained
via the 13C CP/MAS technique on a 200 MHz
spectrometer. The 1H wideline technique, however,
offers the possibility for rapid determination of T1H
relaxation times, saving considerable spectrometer
time. For all specimens only a single T1H decay time
was found by analyzing the intensity of the broad
wideline spectra as a function of the variable time t
according to eqn (1). The fact that the T1H values
are field independent is a strong indication that the
correlation time of the relaxation dominating probes of
these rigid novolac networks are situated at the left side
of the log T1H versus log τc curve (Fig 1). Relaxation
via mobile groups seems to be the dominant pathway
for T1H relaxation in these systems.
10−16 m2s−1 for rigid materials,13–15 and T1H the
relaxation decay time. Spin diffusion data contain
information about the sizes of molecular domains since
all abundant nuclei within the diffusive path length
will obtain the same decay time and has often been
exploited to study heterogeneous systems by wideline
1H NMR.9
The main difficulty coupled with 1H wideline
relaxometry experiments remains the interpretation
of the decay times and their fractions, since the
information of all proton spins appear in a single
broad resonance in the spectrum. In this respect,
determinations of T1H relaxation times through high
resolution solid-state 13C CP/MAS NMR (detection
of the proton relaxation times via the chemical
shift selective carbon nuclei) may be of considerable
interest, as mentioned above.
T1H experiments at room temperature
It has been reported before that the T1H decay time
can be used as a probe to follow the curing progress.3
Table 1 presents the T1H results obtained by means
of 13C CP/MAS and 1H wideline NMR as well as
the degree of conversion (%, content R) for a high-
ortho content novolac that was isothermally cured for
different time periods. The degree of conversion was
determined by means of the formaldehyde/phenol
(F/P) ratio derived from quantitative 13C CP/MAS
spectra.3 All the resonance signals in the carbon
spectrum (Fig 2), those from the aliphatic methylene
T1H experiments as a function of temperature
In order to verify that the correlation times of the
molecular motions of the T1H relaxation dominating
probes are situated at the high-temperature side of the
T1 minimum, T1H measurements have been performed
as a function of the temperature, between ambient
temperature and 100 ◦C. The results for the high-
ortho novolac cured for 6 min (180 ◦C, 150 bar)
are presented in Table 2. Even at these elevated
temperatures, the signal intensity could be analyzed
by a single T1H decay time, indicating that spin
diffusion was still efficient. The results presented
in Table 2 indeed show an increase in the T1H
decay times upon increasing the temperature. This
confirms that the correlation times τc of the relaxation
dominating probes are indeed situated at the left side
of the minimum in the log T1H versus log τc curve.
Mobile network fragments, having motions of several
hundreds of MHz, clearly represent the dominant
Table 1. Degree of conversion R (%) and T1H relaxation decay times
of a high-ortho novolac cured (180 ◦C, 150 bar) for different time
periods
Cure time
(min)
T1H(s)
13C CP/MAS)
T1H(s)
(1H wideline)
R (%)
(
2
8
10
12
60
9.5
45.2
54.8
61.9
73.8
2.72
1.83
1.82
1.72
1.87
2.88
1.92
1.82
1.76
1.84
relaxation pathway for T1H
.
However, the thermal evolution of the T1H decay
times seemed to depend on the experimental condi-
tions. Consecutive measurements as a function of tem-
perature with the same sample clearly resulted in dif-
ferent T1H decay times compared with measurements
The T1H relaxation times were measured by means of solid-state 13C
CP/MAS (200 MHz) and 1H wideline (400 MHz) experiments at room
temperature.
Table 2. T1H relaxation decay times as a function of temperature of a
high-ortho novolac cured for 6 min (180 ◦C, 150 bar), obtained by 1
H
wideline (400 MHz) NMR experiments
a
b
Temp (◦C)
T1H (s)
T1H (s)
20
25
50
75
100
2.31
2.58
4.23
5.72
5.99
2.31
2.58
4.15
6.63
8.83
S
S
250
200
150
100
50
0
ppm
Figure 2. 13C NMR solid-state spectrum (200 MHz) of the high-ortho
novolac resin cured for 6 min (180 ◦C, 150 bar). The peaks assigned
with S are spinning side bands.
a Using a fresh sample for each temperature.
b Using the same sample for the whole thermal range.
Polym Int 52:1647–1652 (2003)
1649

P Adriaensens et al
performed with fresh samples for each temperature.
Table 2 clearly shows that if all measurements were
performed consecutively with the same sample, T1H
increases even more than for a fresh sample at each
temperature.
be very small in cured novolac samples, so explaining
why the corresponding protons are not observed as a
sharp line superimposed on the broad line in the H
1
wideline spectra.
In order to find out the origin of this phenomenon,
spin-lattice relaxation time experiments were per-
formed as a function of temperature by the 13C
CP/MAS technique since it allowed determination
of the relaxation of each carbon resonance selec-
tively. T1H relaxation time measurements, performed
between −35 and 100 ◦C, showed at each temperature
and for all signals of the 13C NMR spectrum, a single
and similar decay time. The T1H results, obtained by
using a fresh sample for each temperature, are shown
in Fig 3. Measuring the whole thermal range with the
same sample also results in significantly higher T1H
values with, as an example, a T1H value of more than
8 s at 100 ◦C. To our knowledge, no explanation has
been given yet for this phenomenon in the literature.
Figure 3 further shows that the correlation times of
molecular motions of the relaxation dominating probes
are situated at the high-temperature side of the T1
minimum in the thermal range between 0 and 100 ◦C.
The T1H minimum was situated around 0 ◦C, which
means that at this temperature the spectral density of
200 MHz motions was optimal. At temperatures below
0 ◦C, the T1H values started to increase again due to
slowing down of molecular motions. These results
confirmed that at ambient temperature the correlation
times τc of the relaxation dominating probes indeed
are situated at the left side of the T1H minimum in the
log T1H versus log τc curve. Since the rigid structure
of cured novolac resins (thermoharder, 3D network)
consist solely of aliphatic methylene bridges and
phenolic rings, only end-groups and free branches can
be responsible for these high frequency motions. The
amount of end-groups and free branches is expected to
Chemical structure as a function of the
temperature
In order to check whether no post-curing (ageing)
occurred in specimens that were used to determine the
T1H decay times up to 100 ◦C, the chemical structure
was investigated by means of a 13C CP/MAS contact
time study although no post-curing was expected at
temperatures below Tg (vitrification).16 Such a study
enabled a quantitative calculation of the number of
methylene bridges per phenolic unit. As expected,
the 13C CP/MAS contact time studies indicated no
structural changes whatever took place during the T1H
experiments at elevated temperatures.
In order to find out whether reorganization
of the secondary hydrogen-bond interactions also
contributed to an increase in molecular motions of
relaxation determining probes upon heating, a FTIR
study was performed between 20 and 180 ◦C. In
tridimensional phenolic networks a large amount
of intra- and inter-molecular hydrogen bonds were
formed between the phenolic hydroxyl groups1 Fig. 4
shows the FTIR hydroxyl stretching band as a
function of temperature for the novolac cured for
6 min. Upon increasing the temperature, the intensity
of the hydroxyl band decreased and the band
became somewhat sharper. This indicated that heating
influenced intra- and inter-molecular hydrogen bond
arrangements and could be responsible for the
increased molecular dynamics of end-groups and
branches. Since their molecular motions are situated
7
6
5
4
3
2
1
4000
3800
3600
3400
3200
3000
2800
2600
Wavenumber (cm-1)
Figure 4. FTIR spectra of the hydroxyl stretching band as a function
of the temperature for the 6 min cured high-ortho novolac. The
temperatures at which the spectra were performed are 20, 50, 70, 90,
110, 130, 150 and 180 ◦C, from the outer to the inner of the spectra,
120 100
80
60
40
20
0
-20
-40
Temperature (°C)
Figure 3. T1H relaxation times as a function of the temperature of the
high-ortho novolac cured for 6 min (180 ◦C, 150 bar), by means of
13C CP/MAS experiments at different temperatures.
respectively. Spectra were all normalized on the signal at 1590 cm−1
corresponding to the elongation of the aromatic ethylene bond
,
=
(C C).
1650
Polym Int 52:1647–1652 (2003)

Solid-state NMR relaxometry of phenolic resins
at the high-temperature side of the T1 minimum,
shorter correlation times result in longer T1H values.
Hydrogen-bond rearrangement explains why longer
T1H values were obtained when the relaxation
experiments as a function of the temperature were
performed with the same sample (Table 2). When
the sample is exposed to consecutive increments
in temperature, the strength of the hydrogen-bond
structure drops systematically. If fresh samples are
used for each T1H experiment this effect will be
smaller probably due to different intermediate states
and because equilibrium is not necessarily established.
It was further observed in FTIR studies that the
rearranged hydrogen-bond structure did not recover
immediately after cooling. FTIR spectra taken after
cooling clearly showed some residual effect on the
hydroxyl stretching band which disappeared only after
several days. This would mean that the disruption
of the initial hydrogen bonds is reversible but that
rearrangement is a slow process. Similar observations
were noticed in the T1H experiments at first sight. In
the T1H decay time still decreased further below its
initial value until an equilibrium value of around 1.8 s
was reached after about 5 days. This was a strong
indication that, after prolonged heating periods, even
at rather moderate temperatures, a slightly different
hydrogen-bond structure arised upon cooling. The
influence of water absorption during the observation
period after cooling from the heating pretreatment
was ruled out by Karl–Fisher and TGA experiments.
These experiments were performed to ensure that
the decrease of the T1H decay time is not due to
slow absorption of water, initially driven off during
the heating of the sample. Karl–Fisher titrations
performed on a high-ortho novolac cured (180 ◦C, 150
bar) for 6 min demonstrated a small amount (0.74%)
of initial water that increased on exposure to open
air (Table 3). However 1H wideline measurements
demonstrated that this increasing amount of absorbed
water did not effect T1H decay times (Table 3). The
TGA experiment (Fig 6) confirmed that this water
was released in the temperature range 90–100 ◦C
and that sample degradation only took place above
250 ◦C. Therefore, the decrease of the T1H decay
time as a function of time after cooling down from
the heating pretreatment could not be attributed to
water absorption but, as stated before, was due to a
rearrangement of the hydrogen-bond structure.
1
order to investigate this in more detail, H wideline
T1H measurements were performed as a function
of time after cooling down from a pretreatment at
different consecutive temperatures up to 100 ◦C. As
mentioned before, consecutive T1H measurements
between 25 and 100 ◦C with the same sample
(Table 2), result in a T1H value of 8.8 s at 100 ◦C.
Figure 5 shows the change in the T1H decay time after
cooling rapidly to ambient temperature as a function
of time. During the first 24 h, the T1H decay time
systematically decreased to a value of around 3.5 s
indicating that the hydrogen-bond structure recovered
slowly. The initial T1H value only recovered 48 h after
cooling. However, the rearrangement of the hydrogen
bonds seems not to be finished after 48 h. After 2 days,
110
0.3
384.21 °C
99.21 °C
0.2
0.1
100
90
96.31 °C
273.68 °C
97.85 °C
9
8
7
6
5
4
3
2
1
0.1
80
70
-0.1
0
100
200
300
400
500
Temperature (°C)
Figure 6. TGA performed on a high-ortho novolac cured (180 ◦C, 150
bar) for 6 min.
Table 3. Amount of water (%) and the T1H decay times of a high-ortho
novolac cured for 6 min (180 ◦C, 150 bar) as a function of time in air
b
Time in air
(h)
H2O content (%)
(Karl–Fisher)a
T1H (s)
1
(
H wideline)
0
3
24
0.74
0.76
1.35
1.39
2.31
2.31
2.32
2.31
0
20
40
60
80
100
120
Time (h)
800
a The water content was measured by means of Karl-Fisher titrations
and the T1H decay times by means of 1H wideline (400 MHz) NMR
experiments.
Figure 5. T1H relaxation decay time (via 1H wideline NMR, 400 MHz)
as a function of time after cooling down from a pretreatment at
different temperatures consecutive up to 100 ◦C of the high-ortho
novolac cured for 6 min (180 ◦C, 150 bar).
b Using the same sample for the whole time range.
Polym Int 52:1647–1652 (2003)
1651

P Adriaensens et al
Minister—Federale diensten voor wetenschappelijke,
technische en culturele aangelegenheden).
CONCLUSIONS
The proton spin-lattice relaxation times of cured
novolac resins is averaged out by spin diffusion.
The correlation times of molecular motions of the
relaxation dominating probes are situated at the left
side of the minimum of the log T1H versus log τc
curve. Mobile end-groups and side-chain branches
seems to determine the most efficient pathway for
T1H relaxation.
REFERENCES
1 Knop A and Pilato LA, Phenolic Resins, Springer-Verlag, Berlin
(1985).
2 Ottenbourgs BT, Adriaensens PJ, Reekmans BJ, Carleer RA,
Vanderzande DJ and Gelan JM, Ind Eng Chem Res 34:1364
(1995).
Despite no post-curing or other chemical reactions
taking place upon heating to 100 ◦C, FTIR and T1H
relaxation experiments demonstrate a rearrangement
of the secondary hydrogen-bond structure. Although
this seemed to be reversible at first sight, there were
strong indications that prolonged heating, even at
these moderate temperatures, results in a slightly
different hydrogen-bond structure after cooling.
3 Ottenbourgs B, Adriaensens P, Carleer R, Vanderzande D and
Gelan JM, Polymer 39:5293 (1998).
4 Neiss TG and Vanderheiden EJ, Macromol Symp 86:117 (1994).
5 Tanaka H and Takagi K, Brit Polym J 21:519 (1989).
6 Charlesby A and Bridges BJ, Brit Polym J 17:645 (1981).
7 Fry CG and Lind AC, Macromolecules 21:1292 (1988).
8 Kaplan S and O’Malley JJ, Polymer 22:221 (1981).
9 Zhang X and Solomon DH, Macromolecules 27:4919 (1994).
10 Powels JG and Strange JH, Proc Phys Soc (London) 82:6 (1963).
11 Laupretre F, Progr Polym Sci 15:425 (1990).
12 Ku W, Liang J, Wei K, Liu H, Huang C, Fang S and Wu W,
Macromolecules 24:4605 (1991).
13 Packer KJ, Poplett IJF and Taylor MJ, J Chem Soc Faraday
Trans 1:84 (1988).
14 Kenwright AM, Kenneth JP and Say BJ, J Magnet Reson 69:426
(1986).
15 Ibbett RN, NMR Spectroscopy of Polymers, Chapman & Hall,
London (1993).
16 Ottenbourgs BT, PhD thesis, Limburgs Universitair Centrum,
Diepenbeek, Belgium (1998).
ACKNOWLEDGEMENTS
This work was financially supported by IWT (Flemish
Institute for the Promotion of Scientific Technological
Research in Industry) and fits in the framework of the
IUAP (Interuniversitaire attractiepolen), financed by
the Belgium Government (Diensten van de Eerste
1652
Polym Int 52:1647–1652 (2003)