MAGNETIC RESONANCE IN CHEMISTRY

Magn. Reson. Chem. 2006; 44: 375–384

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1763

1
H/29Si cross-polarization NMR experiments of
silica-reinforced polydimethylsiloxane elastomers:
probing the polymer–filler interface†
Robbie Iuliucci,1∗ Craig Taylor2 and W. Kirk Hollis2
1
Department of Chemistry, Washington and Jefferson College, Washington, PA 15301, USA
2
Los Alamos National Laboratory, Los Alamos, NM, 87545, USA

Received 2 September 2005; Revised 27 October 2005; Accepted 28 October 2005

Polydimethylsiloxane (PDMS) elastomers reinforced with fumed silica exhibit unusual strength
characteristics that are necessary for their designed applications. The microscopic details of the surface
interaction between the polymer and silica are not well characterized. 1 H/29 Si cross-polarization (cp)
experiments are used to characterize cured and uncured samples of Dow Corning silastic 745. Changes to
the cp dynamics upon curing are evident by the variation in peak intensities in the variable contact-time
spectra of the two samples. Estimates of the cp relaxation parameters are reported for the cured sample.
Additional information can be obtained by expanding the 1 H/29 Si cp to a two-dimensional heteronuclear
correlation experiment. Dramatic differences between the cured and uncured 1 H/29 Si HetCor spectra are
observed that are not visible in the 1D spectra. These changes can be rationalized as a dehydration of the
silica surface and an increased hardening of the polymer after the curing process. Furthermore, isolation
of the NMR signal corresponding to nuclei at or near the polymer–filler interface may be achieved in the
2D 1 H/29 Si HetCor spectrum. Published in 2006 by John Wiley & Sons, Ltd.

KEYWORDS: solid-state NMR; 1 H/29 Si cross-polarization; variable contact time; 2D heteronuclear correlation spectroscopy;
polydimethylsiloxane; fumed silica; elastomer curing

INTRODUCTION approach is challenged by the fact that the NMR signal of

the bulk material must be distinguished from the minor
Cross-polarization (cp) dynamics has been commonly used surface signal of interest. Otherwise, multicomponent mod-
to characterize solid-phase materials such as polymers,1 sil- els must be employed for the nonlinear data analysis. Such
icate surfaces,2 – 6 and polydimethylsiloxane (PDMS)/silica models cannot always be applied confidently. This is further
composite materials.7 – 10 Potentially, one may obtain quan- complicated in systems with polymer surface interfaces that
titative information as well as a measure of heteronuclear suffer from poor signal-to-noise. In PDMS–silica filler sys-
dipolar couplings, which can be related to molecular motion tems, however, a variation in the chemical shift exists for the
and/or nuclear distances, from the transfer profile. The different components of the material. This variation allows
approach recently has been applied toward exploring the isolation of the surface interface in a two-dimensional het-
interface of organic-silicate composite material.11 – 16 In this eronuclear correlation (2D HetCor) spectrum. Polarization
study, we wish to demonstrate the utility of cp spectra to dynamics can be studied in 2D HetCor experiments where
probe the polymer–filler interface in Dow Corning silas- the exchange mechanism is driven by cp.
tic 745, a polysiloxane containing inorganic silica fillers. If The Dow Corning polysiloxane silastic 745, DC745, is
the technique proves valid, it would be of great utility for highly valued for its current military applications. A better
the engineering of fortified soft materials. However, the characterization of DC745’s uncured elastomer resin as
well as knowledge of its aging profile is of great interest.
†
Dedicated to Professor David M. Grant of the University A more detailed understanding of the changes that take
of Utah on the occasion of his 75th birthday, in place during the shelf life of uncured elastomers is also of
recognition of the outstanding contributions he has made general interest to the silicone industry.17 In this regard,
to the methodology of nuclear magnetic resonance and its DC745 serves as a model silicone elastomer. DC745 has
application to a wide range of chemical topics over a long already been studied extensively and a series of pristine and
period of time.
artificially aged samples have been prepared.18 A test group
Ł Correspondence to: Robbie Iuliucci, Department of Chemistry,
for future experiments is available. DC745 consists primarily
Washington and Jefferson College, Washington, PA 15301, USA. of PDMS, where approximately 1 in 20 of the methyl groups
E-mail: riuliucci@washjeff.edu
Contract/grant sponsor: National Science Foundation; is substituted for phenyl rings and, to a lesser extent, the
Contract/grant number: CHE-0216075. methyl is replaced by a vinyl group. Phenyl rings are found

This article is a US Government work and is in the public domain in the USA.
376 R. Iuliucci, C. Taylor and W. K. Hollis

to help improve creep resistance by stabilizing the siloxane how these interactions change during the lifetime of the
bond and decreasing the rate of chain scission reactions. material. During storage, Crepe-hardening can result for
The vinyl serves as the cross-linking group in the polymer uncross-linked siloxane filled with fumed silica.17
when peroxide cross-linking agents are added. In DC745, Macroscopic measurements on filled gels are useful
PDMS is mixed with a high-surface-area fumed silicate that indicators of the reinforcing effect, but are incapable
contains quartz. Upon curing, the material transforms from of providing information on the strength of the local
a puttylike consistency to a firm rubber. interactions that govern the underlying mechanism. Clearly,
It is believed that the material’s unique strength char- adsorption forces must play a central role in the adhesion
acteristic stems from the surface interaction between the of the polymer to the filler and these ultimately control the
polymer and the fumed silica filler.19 – 22 A cartoon represen- macroscopic mechanical behavior. These phenomena occur
tation of the surface interface is displayed in Fig. 1. Physical in a thin layer around the silica particles on a distance
forces, hydrogen bonding, and van der Waals forces dom- scale of the order of the size of a polymer molecule, i.e.
inate polymer–filler interactions for hydrophilic surfaces. 10–100 Å. Such length scales are ideally suited for NMR cp
Covalent bonding, however, may be involved with cur- investigations that depend upon short length scale dipolar
ing. Understanding this polymer–filler interaction on the interactions.
molecular level serves as the primary objective of this study. The employment of heteronuclear dipolar coupling for
Fillers can be a way to improve the physical and magnetization transfer limits cp to motionally constrained
mechanical properties of polymers.23 For example, in plastics and proximate spin pairs. In composite materials that
exposed to mechanical loads and high-heat service, glass contain regions with varying degree of motion, cp can
fibers or mineral fillers can add rigidity and creep resistance. select the confined spin pairs from those undergoing rapid
In general, the addition of inorganic fillers to elastomers movement, which eliminates their heteronuclear couplings.
leads to the mechanical reinforcement of polymers and to In DC745, the overwhelming bulk signal of flexible PDMS
reduced swelling, hence rendering them mechanically and that dominates a 29 Si single-pulse excitation spectrum is
chemically more stable. For silica and zeolites in PDMS, removed from the 1 H/29 Si cp spectrum. Cross-polarization,
such reinforcement has been explained by adsorption of the in 1D NMR, cannot confidently probe the PDMS–filler
polymer chains on the silanol groups of the filler surface. interface. Physical entanglements and cross-links reduce
When the filler interacts well with the polymer, the effective polymer motion in PDMS, resulting in additional polarizable
cross-linking density increases and the swelling degree silicone. The silica that is insulated from the polymer can
of the polymer network decreases. Finely divided filler also be cross-polarized from surface hydroxyl groups and
particles have been used for many decades as reinforcing adsorbed water. Although the dipolar coupling environment
agents in PDMS elastomers and their effect on the resulting of these sites may be different from that of the PDMS–silica
macroscopic elastic properties is well documented.24 – 26 It interface, the chemical shifts will not vary significantly. This
is also well known that the mechanical properties of filled complicates the cp dynamics as the transferred 29 Si signal
systems (elastic modulus, tear strength, etc.) are sensitive becomes a superposition of all possibilities. Biexponential
to the surface treatment applied to the filler. Less well curve fitting has been applied to study the cp dynamics of
understood, however, is the effect of the polymer–filler fumed silica.6 Here, however, one would have to extend the
interaction on the local distribution of polymer around the nonlinear curve fitting beyond two components, where the
filler on a submicroscopic length scale. It is also not known solutions are rather fortuitous.
In composite materials, where the chemical shifts of the
components are resolved, the dynamics can be determined
Silica-filled polydimethylsiloxane elastomer
independently in a two-dimension spectrum. This is the case
C6H5 CH3 C2H3
polymer in siloxane polymers reinforced with inorganic silica. The
[-SiO]x[-SiO]y [-SiO]z 29
Si chemical shift of silicate,
80 to
115 ppm, are clearly
Free polymer CH3 CH3 CH3 resolved from siloxane polymer shifts,
10 to
40 ppm.
Dphenyl D Dvinyl Silanol proton shifts in silica, ¾5 ppm, are well separated
from the methyl proton shifts of siloxane polymers, ¾0 ppm.
silica O The combination of 1 H and 29 Si shifts confidently isolates
Bound Q#Si
Si the signal corresponding to the elastomers surface in a
polymer O O
O 2D 1 H/29 Si HetCor spectrum. Potentially, the magnetization
Q4 Q3 or Q2
transfer dynamics of the surface interface can be studied in a
2D 1 H/29 Si HetCor spectra when cp mediates the exchange.

EXPERIMENTAL
Samples of DC745, cured and uncured gum resin, were
Figure 1. A cartoon depicting the siloxane polymer silica filler provided by LANL. DC745 is cured by mill rolling a 99.5%
system. Hydrogen bonding between hydroxyl sites on the gum resin and 0.5% catalyst blend by weight. A 50% active
surface and polymer are assumed as the primary mechanism Luperco 101XL (2,5-dimethyl-2,5-di(t-butylperoxy)hexane)
for the surface interaction. was used as the catalyst. The milled pads were cured at

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
 1 H/29 Si cross-polarization NMR experiments of silica-reinforced polydimethylsiloxane elastomers 377

170 ° C for ¾35 min and postcured at 150 ° C for 1 h followed Water protons adsorbed to the silica surface appear in the
by 8 h at 250 ° C. The cured rubber was sliced into small 3–5 ppm range. Water physisorbed to the silica has a 1 H
square pieces (¾1–2 mm in length) with a razor blade and chemical shift of 3.5 ppm. A liquidlike 1 H chemical shift for
packed into the zirconia rotors. The puttylike uncured DC745 water of 4.9 ppm appears in humidified fumed silica. Water
was placed directly into the rotor. exchange between the liquid and physisorbed environments
All NMR spectra were acquired on a Varian Inova is observed at an intermediate shift of 4.1 ppm.6
spectrometer and a wide bore 9.4 Tesla Oxford magnet with In theory, the proton-decoupled 29 Si SPE spectral intensi-
operational frequencies for 1 H and 29 Si of 399.80 MHz and ties are quantifiable. However, 29 Si quartz sites are noto-
79.426 MHz respectively. Varian Chemagnetics T3 double riously known for their long T1 values, and intensities
resonance probes, 4.0 mm and 7.5 mm, were used. Magic- are underestimated. Thus, one can only approximate the
angle spinning with an automated spin controller was amounts of the DC745 components based on signal intensity
employed at speeds between 2.0 and 5.0 kHz for 29 Si spectra. and make relative qualitative comparisons between simi-
The 1 H spectrum was acquired at 15 kHz, spinning with the lar materials. The 29 Si SPE spectra of uncured and cured
high-speed-spinning-assisted WaHuHa pulse sequence.27 DC745 are shown in Fig. 3, where some interesting differ-
A 90° pulse width of 4.0 and 5.5 µs for 1 H and 29 Si ences are noted. The broadness of the
20-ppm D signal in
was employed respectively. For the 29 Si spectra, relaxation the uncured sample obscures the Dphenyl signal at
32 ppm.
delays of 2–120 s were used. The signal was detected The different silicon Q sites are more pronounced in the
under 27 kHz of continuous wave decoupling, 0.05 ms cured spectrum. The spinning sidebands, a manifestation of
acquisition time, and a dwell time that produced a 250-ppm magic-angle spinning, in the uncured sample do not appear
spectral window. For cp experiments, the duration of the above the noise of the spectrum. However, sidebands have
Hartmann–Hahn match was arrayed from 0.1–20 ms for been observed in spectra with slower spinning speeds (not
variable 1 H/29 Si contact-time data. A spin-lock field of 47 kHz shown). The differences in the spinning sideband intensity
was used during the Hartmann–Hahn match condition. The may indicate a change in the chemical shift anisotropy after
2D HetCor experiments28 incorporated frequency-switched curing.
Lee–Goldburg29 homonuclear decoupling during the f1
period. A mixing time of 10 ms was used during cp in the Cross-polarization
2D HetCor experiment. All spectra were acquired at room Typically, the use of cp, the heteronuclear transfer of
temperature. The spectra were referenced externally to liquid magnetization from an abundant to a dilute spin system, is to
TMS. The spectra were processed with the Varian VNMRJ achieve an increase in signal-to-noise. Here, cp is employed
software. The peak intensities from variable contact-time as a means to edit or filter the 29 Si spectrum by decreasing
data were analyzed in a curve-fitting tool of Matlab.30 signal that corresponds to the mobile polymer. The filtering
process exploits the angular and distance dependence of the
dipolar interaction, which serves as the transfer mechanism
RESULTS AND DISCUSSION
during cp. This is evident in the spatial factors of the
1D spectra of bulk material heteronuclear Hamiltonian,
The single-pulse excitation (SPE) 29 Si and multiple-pulse
  
high-speed spinning 1 H spectra of DC745 (Fig. 2) reveal 1 o h H Si
ωijHSi D
3 cos2 ij
1 1
the mobile polymer as the dominant component in the 2 82 r3ij
elastomer. The overpowering polymer signals, 0 ppm and

20.5 ppm in the 1 H and 29 Si spectra respectively, render the where the usual definitions are assumed.32 The interaction is
minor, but informative, inorganic silicate and phenyl-/vinyl- scaled by the inverse cubic distance between spins i and j, r3ij ,
substituted polymer signal almost useless. NMR peaks which confines the spatial transfer of magnetization to the
can be assigned to their chemical sites based on known nearest spin pair. Owing to the averaging of the angular term,
literature values.5,6,15,31 The phenyl-substituted siloxane molecular motion reduces or in the presence of fast isotropic
polymer (Dphenyl ) appears at
32 ppm, while the inorganic motion can even eliminate the dipolar coupling. Above
fumed silicate shows up as broad peaks in the
80 to the glass-transition temperature (
120 ° C), PDMS exhibits

115 ppm region of the 29 Si spectrum. Geminal silanols, a relatively significant motion. However, the complete
single silanol, and siloxane, designated by Q2 , Q3 , and Q4 , averaging of the dipolar interactions is prevented by the
have 29 Si shifts of
89,
99, and
109 ppm respectively. polymer chain entanglements, cross-links, and adhesion to
The sharp peak at
110 arises from crystalline quartz. The the silicate filler and quartz. Residual dipolar couplings
hydrogen-bonded silanol sites of fumed silicate and aromatic above the glass-transition temperature have been studied in
protons of the phenyl rings account for the downfield DC745.33
1
H peaks. A broad chemical shift range of 1–8 ppm has In the absence of homonuclear decoupling, spin diffusion,
been reported in silanol protons in fumed silica. This wide the homonuclear transfer of magnetization throughout the
chemical shift range is due to the varying degree of hydrogen abundant spin network, can transpire during cp. This
bonding that the silanols can be involved with. An increase indirectly spreads the magnetization beyond the nearest
in hydrogen bonding results in a decreased shielding. The spin pair. Spin diffusion can span distances of 20–200 nm
center of the hydrogen-bonded silanol shift is around 5 ppm in solids, depending on the proton spin density and the
and isolated silanol protons have a chemical shift of 2.0 ppm.6 strength of homonuclear dipolar coupling.32 Given the weak

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
378 R. Iuliucci, C. Taylor and W. K. Hollis

(a)

(b)

Figure 2. Single-pulse excitation 1 H (a) and 29 Si (b) spectra of DC745. The polymer signal dominates both spectra. The blown-up
spectrum reveals features in the baseline.

homonuclear dipolar couplings found in DC745, smaller difference being the larger amount of rigid polymer signal
length scales are presumed here. in the cured spectrum. The broadness of the Dphenyl and
The cp spectra enhance the appearance of the subtle the inorganic silicate peaks become apparent. This chemical
peaks corresponding to the smaller amounts of inorganic shift dispersion can be attributed to a variation in local
and substituted polymer components (Fig. 4). Both the structure. The PDMS signal remains relatively narrow,
cured and uncured spectra are similar, with the primary suggesting either a homogenous local structure and/or

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
 1 H/29 Si cross-polarization NMR experiments of silica-reinforced polydimethylsiloxane elastomers 379

(a) (b)

Figure 3. Single-pulse excitation 29 Si spectra of uncured (a) and cured (b) DC745. Three silicon polymer signals are observed:
PDMS (D), phenyl mono-substituted (Dphenyl ), and vinyl mono-substituted (Dvinyl ) polymer. There exist four types of inorganic 29 Si
nuclei: tetrahedral silica where all four peripheral oxygen are bonded to another silicate (Q4 ), or where one or two peripheral silicates
are replaced by a hydroxyl group (Q3 or Q2 , respectively). The Q4 can either be crystalline (sharp resonance) or disordered (broad
resonance).

(a)

(b)

Figure 4. The cp/mas 29 Si spectra of uncured (a) and cured (b) DC745. Two inorganic silica sites Q3 /Q4 are readily observed. Note
the increased intensity of the polymer signal in the cured spectrum.

residual motion. Two inorganic silicate sites (Q3 and Q4 ) to sites that are at or near the surface of the fumed
are evident. Some indication of the presence of Q2 is seen silicate, where, through space, dipolar couplings would
with a small shoulder at ¾85 ppm. The lack of chemically exist. Unlike the Q4 in fumed silicate, the narrow quartz
bonded protons to the Q4 silicon would challenge cp to this peak observed in the SPE spectra is not cross-polarizable and
site. It is suspected that this observed cp signal corresponds disappears.

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
380 R. Iuliucci, C. Taylor and W. K. Hollis

The chemical shifts of the cured and uncured DC745 (a)

are nearly identical. Although the physical change with
curing is self-evident as the puttylike material transforms
into a rubbery material, there is no NMR evidence of
significant chemical changes upon curing. This suggests
that the surface-binding mechanism is not chemical and
constitutes hydrogen bonding and van der Waal forces. It is
believed that the siloxane polymer chemically bound to the
silicate surface would result in a 29 Si chemical shift in the
50
to
70 ppm region. Aminopropylalkoxysilanes deposited on
silicate surfaces resulted in shifts of
49 to
54 ppm for
monodentate bonding and
57 to
60 ppm for bidentate.34
Tridentate bonding showed shifts of
66 to
68 ppm. DC745
subjected to high temperatures for extensive periods of time
produces a new resonance in the 29 Si spectrum in the
65 20 18 16 14 12 10 8.0 6.0 5.0 4.0 3.0 2.5 2.0 1.5 1.0 0.5 0.1
to
70 ppm region.35 Under those conditions, however, the
chemical changes are obvious. The material undergoes a (b)
complete degradation of the polymer where the elasticity is
lost and the rubber decomposes into a chalklike solid.

Variable contact-time cross-polarization spectra

While the signal-enhancing feature of cp is advantageous
in itself, here we wish to characterize the materials based
on their cp by observing the cp dynamics. Under cp, the
magnetization (M) of the 29 Si dilute spin builds up with time
according to the following expression:36

 
1
1

THSi t
t
 THSi  T1
M T1
Mt D
1
e e 2
THSi
1

T1 18 16 14 12 10 8.0 6.0 5.0 4.0 3.0 2.0 1.0 0.5 0.1

where THSi and T1 are the proton–silicon cross-relaxation Figure 5. An array of cp/mas 29 Si spectra for uncured (a) and
time and the longitudinal relaxation of the protons in the cured (b) DC745 where the contact time is varied from 0.1 ms
rotating frame respectively. This expression is valid only in (right) to 20 ms (left).
the fast cp regime where THSi << T1 and the decay of the
signal due to the longitudinal relaxation of the silicones in the
Table 1. Cross-polarization parameters DC745
rotating frame is ignored. The three parameters of Eqn (2),
M1 , THSi , and T1 , can be determined by fitting the peak T1 (ms) THSi (ms) M1 a.u.
intensities in a series of spectra where the cp time is varied.
Cured
The variable contact-time spectral array for the cured and
D 13 š 3 13 š 3 361 š 43
uncured DC745 are shown in Fig. 5. The exponential fits for
Dphenyl 15 š 2 1.5 š 0.3 73 š 6
the variable contact experiment are shown in Fig. 6 and the
Q3 14 š 4 2.8 š 0.7 66 š 10
corresponding cp parameters are reported in Table 1.
Q4 13 š 3 4.2 š 0.9 100 š 16
Changes to the relaxation dynamics in the two DC745
samples are clearly evident from the spectral intensity Uncured
differences. In the cured sample, the buildup of D is faster D 10 000 77 700
and a clear maximum point is observed. Above 8 ms, all of Dphenyl 63 š 40 1.3 š 0.3 30 š 3
the peaks including the polymer begin to decay. The buildup Q3 10 000 1.6 17
curve of the uncured D takes place at a slower rate and a Q4 10 000 1.7 18
maximum transfer is never observed within the time frame
probed. The decay of signal in the uncured Q sites is less
pronounced. magnetization transfer. Without observing a maximum, M1
The extraction of three parameters from nonlinear curve for D is underestimated. The lack of a measurable decay
fitting can be daunting. The curve fitting becomes further in the signal at longer contact periods prevents a realistic
challenged with spectra lacking desirable signal-to-noise. measurement of T1 . The nonlinear analysis reports the
The reliability of the parameters reported in Table 1 is upper-bound limit for this parameter in the cases of D,
therefore questionable. A mathematically unique solution Q3 , and Q4 .
cannot be determined with confidence for D, Q3 , and Q4 Other considerations bring into question the cp values
sites in the uncured sample. The PDMS signal shows a linear reported here. A fast cp regime cannot be determined

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
 1 H/29 Si cross-polarization NMR experiments of silica-reinforced polydimethylsiloxane elastomers 381

(a)

(b)

Figure 6. Buildup curves for the peak intensities of the cp/mas 29 Si spectra of uncured and cured DC745. The curves are fit for the
cp parameters according to expression 2. Plot (a) contains the polymer peaks, while the peaks corresponding to the Q sites are
shown in plot (b).

from the appearance of the transfer curve alone and the only a monoexponential polarization transfer, which is too
slow cp regime, THSi > T1 , has been reported for silicate simple a model to explain a composite material. A single
systems.37 – 39 It should be emphasized that the cured D spectral resonance can be described by multiple species with
solution finds THSi D T1 , which violates the fast cp regime separate transfer dynamics. This situation is not uncommon
assumption. Complimentary information is necessary, such for Q3 sites and typically two separate species are reported.6
as the independent measurement of T1 , to determine the With the presence of PDMS, it would not be unrealistic to
type of cp regime. Secondly, the above analysis considers expect more than two Q3 species in DC745.

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
382 R. Iuliucci, C. Taylor and W. K. Hollis

2D HetCor experiments of the degree to which the polymer can be polarized. But this
To help alleviate the issue of a single resonance represent- quadrant does not specify if the ability to polarize is due to
ing multiple species, we resort to 2D NMR. All neighboring entanglements, cross-linking, or the silica interface.
protons participate in cp, creating multiple pathways for Pronounced changes to the HetCor spectrum are
magnetization transfer. Given favorable resolution, the 1 H observed with sample curing. There is significant inten-
chemical shift of the different polarization sources can be dis- sity loss to the hydroxyl-silicate region that suggests that the
tinguished in a 2D experiment. When cp drives the exchange curing process dehydrates the material. The 5.7 ppm peak
mechanism, the 2D 1 H/29 Si HetCor spectrum (Fig. 7) cor- in the uncured sample disappears. Two mechanisms could
relates the 1 H/29 Si chemical shifts where dipolar couplings be proposed: the direct removal of adsorbed water or indi-
exist. The projection along either dimension is identical to a rectly from a condensation reaction between the neighboring
1D cp/mas spectrum. A 1D cp/mas 1 H spectrum differs from silanol site. Temperatures used in the curing and postcur-
a 1D 1 H WaHuHa high-speed MAS spectrum in that only the ing of DC745 could drive adsorbed water from the silica
proton signal participating in cp is observed. Note the subtle surface. The proton peak at 5.7 ppm could correspond to a
differences in the 1 H projections to that shown in Fig. 7(a). silanol hydrogen bonded to water, and it is this water that is
The 1 H projection clearly shows the increased polymer sig- removed during the curing/postcure period. However, even
nal with curing. The uncured 1 H projections show the strong temperatures of 190 ° C can allow siloxane bonds to form
presence of a unique peak at 5.7 ppm. On the basis of the pro- between two neighboring surface silanols.40 A stiffening of
ton shifts of fumed silica, this peak is attributed to hydrogen- the polymer also takes place as suggested by the growth
bonded silanols. It should be emphasized that a lack of signal of polarizable polymer signal and the enhancement of the
in the 3–5 ppm proton shift range does not exclude the pres- spinning sideband. The formation of cross-links during cur-
ence of additional water near the silica surface. Through the ing would result in an increase of cross-polarizable PDMS.
hydrogen bonding with the silanol groups, silica can have However, cross-linking cannot explain this signal increase
many layers of adsorbed water. However, these outer layers alone since the polymer’s spinning sideband also appears
have liquidlike properties and are not cross-polarizable. A with the 29 Si silicate shift. This suggests that the polymer is
spinning sideband appears at 5 ppm in the 1 H dimension in close proximity with the filler.
of the cured spectrum. With Lee–Goldburg homonuclear
decoupling being applied during the 1 H evolution, the Lar-
mor precession of the protons becomes scaled by a factor
Future HetCor studies
of 1.73. Adjusting the 1 H chemical shift axis for this scaling
In order to extract quantifiable information from the 2D spec-
artificially places the 3.5 kHz spinning sideband at 5.0 ppm.
tra, the polarization dynamics must be studied. A series of
The advantage of the 2D spectrum, however, is the fact
2D HetCor spectra where the contact time is varied would
that the correlated spectrum gives more information by
provide buildup curves similar to the 1D data. This 3D data
providing the proton source of transferred magnetization
during cp. The multiple cp pathways of DC745 are evident set (1 H chemical shift – 29 Si chemical shift – contact time)
in the HetCor spectrum. For example, the phenyl-substituted would distinguish the cp characteristics of the different pro-
siloxane has a 29 Si shift of
30 ppm. Two peaks are seen along ton sites. In principal, different cp parameters would be
the
30 ppm 29 Si axis. The peak that has a 1 H shift at 0 ppm observed for the 29 Si because of the different magnetization
corresponds to magnetization transferred by methyl protons, sources. Alternatively, one could exploit the separation of the
while the down field peaks stem from aromatic protons. In proton chemical shifts. An array of 1D cp experiments with
a similar fashion, the inorganic silicate along the
85 to selective 1 H excitation would provide similar information

115 ppm 29 Si axis has three potential proton polarization to the 2D. The polarization dynamics would be assessed in
sources – hydroxyl groups of the Q2 /Q3 sites, water on the conjunction with an independent measurement of T1 . Spin
silica surface, and methyl protons of the polymer. Because diffusion, homonuclear proton magnetization transfer, takes
of the chemical shift separation in the two dimensions, four place during cp. This spin diffusion extends the length scale
distinct regions of peaks appear in the 2D HetCor spectrum. monitored by polarization by means of the proton–proton
The top left quadrant is the 29 Si-detected silicate that is cross- network. The signal in the upper-right quadrant that corre-
polarized by either water or the Q2 /Q3 hydroxyl groups. sponds to the surface interface must be at or near the surface
Owing to the low occurrence of the phenyl rings, significant due to the distance constraint of the dipolar interaction.
contributions of detected silicate signal from aromatic However, the act of spin diffusion extends the probed region
protons is unlikely. The top right quadrant is 29 Si-detected beyond the vicinity of the polymer–silicate interface. The sig-
silicate cross-polarized from the protons in the methyl or nal buildup according to the proton–silicon cross-relaxation
vinyl group of the polymer. It is in this region that the time would not reflect the directly coupled 1 H and 29 Si spins.
signal confidently corresponds to the surface interface since Homonuclear proton decoupling, i.e. Lee–Goldburg, would
the transferred proton source can only come from polymer quench the proton–proton spin diffusion. Cross-polarization
nuclei. The bottom left quadrant is the 29 Si-detected polymer under these conditions41 would be limited to directly coupled
silicone that has been cross-polarized by either the aromatic proton–silicon spins. Without homonuclear decoupling, one
protons or by hydroxyl protons on the silicate surface. The cannot assume with confidence that only directly coupled
last quadrant, bottom right, is the silicone polymer cross- proton–silicon interactions are observed. At this point in
polarized by methyl protons. This quadrant gives a measure time, the modified pulse sequence has not been implemented.

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
 1 H/29 Si cross-polarization NMR experiments of silica-reinforced polydimethylsiloxane elastomers 383

(a)

(b)

Figure 7. The 2D HetCor 1 H/29 Si spectra for uncured (a) and cured (b) DC745. The 1 H chemical shift runs horizontally, while the 29 Si
shift is vertical. On the basis of 1 H/29 Si chemical shift differences, the peaks can be assigned to the different material components.

CONCLUSIONS this interaction by its cp dynamics. In addition, the use of

cp enhances the spectral appearance of a composite material
Little is known about the microscopic details relating to the that contains regions with varying degrees of motion. Here,
shelf life and age profile of reinforced siloxane elastomers. the techniques of cp are explored in a cured and uncured
If NMR signals corresponding to nuclei at or near the poly- polysiloxane reinforced with fumed silica and DC745 that
mer–filler interface can be isolated, one may characterize serves as an example of siloxane elastomer. Although no 29 Si

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384
384 R. Iuliucci, C. Taylor and W. K. Hollis

chemical shift changes are observed between the two sam- 2. Maciel GE, Sindorf DW. J. Am. Chem. Soc. 1980; 102: 7607.
ples, their transfer profiles are dramatically different. This 3. Sindorf DW, Maciel GE. J. Phys. Chem. 1982; 86: 5208.
4. Sindorf DW, Maciel GE. J. Am. Chem. Soc. 1983; 105: 3767.
indicates a variation in the cp dynamics due to chemical
5. Chuang IS, Maciel GE. J. Am. Chem. Soc. 1996; 118: 401.
changes that take place with curing. 6. Liu CC, Maciel GE. J. Am. Chem. Soc. 1996; 118: 5103.
The polarization parameters may prove sensitive to 7. Litvinov VM, Spiesis HW. Makromol. Chem. 1991; 192: 3005.
chemical changes in siloxane materials associated with aging. 8. Legrand AP, Lecomte N, Vidal A, Haidar B, Papirer E. J. Appl.
However, care must be taken with the extraction of these Polym. Sci. 1992; 46: 2223.
9. Ito M, Nakamura T, Tanaka K. J. Appl. Polym. Sci. 1985; 30: 3493.
parameters before conclusive remarks can be made with
10. Cosgrove T, Turner MJ, Thomas DR. Polymer 1997; 38: 3885.
confidence. The type of cp regime, fast or slow, describing 11. Huijgen TP, Gaur HA, Weeding TL, Jenneskens LW, Schu-
the transfer must be identified by complementary informa- urs HEC, Huysmans WGB, Veeman WS. Macromolecules 1990;
tion. This can be supplied by separate T1 measurements and 23: 3063.
by the TORQUE experiment.37 Secondly, the transfer pro- 12. Pfleiderer B, Albert K, Bayer E, van de Ven L, de Haan J,
Cramers C. J. Phys. Chem. 1990; 94: 4189.
file should represent only NMR signals stemming from the
13. Gambogi JE, Blum FD. Macromolecules 1992; 25: 4526.
polymer–filler interface. The influence of multiple pathways 14. Comotti A, Simonutti R, Sozzani P. Chem. Mater. 1999; 8: 2341.
for magnetization transfer to each 29 Si site can complicate 15. Simonutti R, Comotti A, Negroni F, Sozzani P. Chem. Mater.
the physical interpretation of the cp parameters. This com- 1999; 11: 822.
plication can be eliminated by going to 2D NMR. The added 16. Simonutti R, Comotti A, Bracco S, Sozzani P. Chem. Mater. 2001;
1 13: 771.
H chemical shift dimension allows one to assign regions
17. Osaheni JA, Truby KE, Silvi N. Macromol. Symp. 2001; 169: 261.
of the 2D spectrum to different components of the material. 18. Maxwell R, Chien A, Balazs B, LeMay J. ACS Polym. Mater. Sci.
In this regard, one can isolate NMR signals that correspond Eng. 2000; 82: 183.
to the silicate surface and the polymer–silicate filler inter- 19. Chahal RS, St. Pierre LE. Macromolecules 1969; 2: 193.
face. A series of 2D HetCor spectra where the contact time 20. Hanson DE. J. Chem. Phys. 2000; 113: 7656.
21. Gee RH, Maxwell RS, Balazs B. Polymer 2004; 45: 3885.
is varied in increments would display the cp dynamics of
29
22. Polmanteer KE. Rubber Chem. Technol. 1981; 54: 1051.
Si signal due to its proton source. The implementation of 23. Litvinov VM, Zhadanov AA. Polym. Chem. USSR 1988; 30: 1000.
Lee–Goldburg homonuclear decoupling during cp would 24. Wagner MP. Rubber Chem. Technol. 1976; 49: 703.
quench spin diffusion. This would restrict magnetization 25. Bachmann JH, Sellers JW, Wagner MP, Wolf RF. Rubber Chem.
transfer to directly coupled 1 H/29 Si spins. Under such con- Technol. 1959; 32: 1286.
26. Iler RK. The Chemistry of Silica. Wiley: New York, 1979.
ditions, internuclear distances may be referred from the
27. Hafner S, Spiess HW. Solid State NMR 1997; 8: 17.
observed proton–silicon cross-relaxation time. These dis- 28. Caravtti P, Braunschweiler L, Ernst RR. Chem. Phys. Lett. 1983;
tances may be used to refine atomistic models of the surface 100: 305.
interaction.20 29. Bielecki A, Kolbert AC, Levitt MH. Chem. Phys. Lett. 1989; 155:
Qualitatively, the 2D HetCor spectra display changes to 341.
30. Matlab Release 12.1. The MathWorks: Natick, 2001.
DC745 due to the curing process where notable changes to the
31. Mayer BX, Zoellner P, Kaehlig H. J. Chromatogr., A 1999; 848:
signal intensity are observed. These changes are consistent, 251.
with the curing process dehydrating the silicate surface 32. Schmidt-Rohr K, Spiess HW. Multidimensional Solid-STate NMR
as well as increasing the polymer’s rigidness. 2D 1 H/29 Si and Polymers. Academic Press: New York, 1994.
HetCor spectra on artificially aged DC745 may reveal other 33. Maxwell R, Balazs B. J. Phys. Chem. 2002; 116: 23, 10 492.
34. Ek S, Iiskola EI, Niinistoe L, Vaittinen J, Pakkanen TT, Root A.
changes. Further quantitative information may be provided
J. Phys. Chem. 2004; 108: 11 454.
by exploring the polarization dynamics in the 2D 1 H/29 Si 35. Iuliucci R. 1 H/29 Si Cross-polarization NMR spectrum of
HetCor or by 1D 1 H selective excitation cp experiments. thermally decomposed DC745 acquired at the University of
Utah, August 2000; unpublished results.
Acknowledgements 36. Mehring M. Principles of High resolution NMR in Solids. Springer:
The NMR spectrometer was purchased through the acquisition Berlin, 1983.
of funds from the National Science Foundation’s Major Research 37. Klur I, Jacquinot JF, Brunet F, Charpentier T, Virlet J, Scheider C,
Instrumentation Program (Award No. CHE-0216075). Tekely P. J. Phys. Chem. B 2000; 104: 10 162.
38. Fyfe CA, Brouwer DH, Tekely P. J. Phys. Chem. A 2005; 109: 6187.
39. Gardiennet C, Marica F, Fyfe C, Tekely P. J. Chem. Phys. 2005;
REFERENCES 122: 054705.
1. Bovey FA, Mirau PA. NMR of Polymers. Academic Press: San 40. Zhuravlev LT. Pure Appl. Chem. 1986; 61: 1969.
Diego, 1996. 41. Fu R, Hu J, Cross TA. J. Magn. Reson. 2004; 168: 8.

Published in 2006 by John Wiley & Sons, Ltd. Magn. Reson. Chem. 2006; 44: 375–384