Nitroxyls for Scorch Suppression, Cure Control,
and Functionalization in Free-Radical
Crosslinking of Polyethylene
Bharat Indu Chaudhary,1 Lamy Chopin,2 John Klier3
1
The Dow Chemical Company, 1 Riverview Drive, Somerset, New Jersey 08873
2
The Dow Chemical Company, 171 River Road, Piscataway, New Jersey 08854
3
The Dow Chemical Company, 2030 Dow Center, Midland, Michigan 48674
This paper describes the use of 2,2,6,6-tetramethylpi-
peridinyl-oxy (TEMPO) derivatives for scorch suppres-
sion, cure control, and functionalization in peroxide
crosslinking of polyethylene. When 4-hydroxy 2,2,6,6-tet-
ramethylpiperidin-1-oxyl was used for scorch suppres-
sion, there was often a loss in ultimate degree of cross-
linking. In contrast, with bis(1-oxyl-2,2,6,6-tetramethylpi-
peridine-4-yl)sebacate, both scorch suppression and
ultimate degree of crosslinking were enhanced. A model
study in hexadecane showed that TEMPO radicals ter-
minate with carbon-centered radicals formed as a con-
sequence of peroxide homolysis and propagation steps.
This termination occurs preferentially over peroxide-initi-
ated crosslinking and results in TEMPO-grafted polymer.
In addition to polymer radical formation, several addi-
tional reaction pathways are available following thermal
homolysis of the peroxide, including unimolecular disas-
sociation of the peroxy radical to yield a methyl radical
and a ketone, and proton extraction from one of several
substrates by the peroxy radical to yield an alkyl radical
and an alcohol. This study reveals that the reaction rate
is limited by the rate of peroxide homolysis, and pro-
ceeds to statistical products with little or no preference
for any specific species. The implication is that choice
of peroxide is a dominant controlling factor over
whether the TEMPO derivatives are ultimately grafted to
the polymer or are bound to small alkyl radicals. POLYM.
ENG. SCI., 47:50–61, 2007. ª 2006 Society of Plastics Engineers
INTRODUCTION
Crosslinking of polyolefins is done for a variety of appli-
cations in order to achieve one or more of the following
performance attributes: increased resistance to high temper-
Correspondence to: Bharat Indu Chaudhary; e-mail: bichaudhary@dow.com
DOI 10.1002/pen.20664
Published online in Wiley InterScience (www.interscience.wiley.com).
V
C 2006 Society of Plastics Engineers
ature deformation, increased flow resistance, good impact
properties (toughness), abrasion resistance, environmental
stress crack resistance, solvent (chemical) resistance, and
good tensile properties. A commonly used method for
crosslinking of olefinic polymers is the generation of poly-
mer radicals through the use of organic peroxides that sub-
sequently combine to form carbon–carbon bonds. Organic
peroxides decompose by a first-order reaction, which can
be controlled by raising the temperature [1]. In the case of
polyethylene, depending on the time and temperature, the
gel content and molecular weight between crosslinks (both
measures of degree of crosslinking) generally reach a pla-
teau with peroxide concentration, after which further
increases in peroxide concentration only lead to more deg-
radation products that have detrimental effects on proper-
ties [2]. Crosslinked polyethylene is believed to comprise
crystalline regions that are interconnected via linear or
crosslinked tie molecules embedded in an amorphous cross-
linked matrix [2]. The effect of polyethylene molecular
structure (including vinyl unsaturation) on free-radical
crosslinking of polyethylene has been extensively studied,
for both branched and linear polyethylenes, as well as
blends of these polyethylenes [3–8].
It is generally necessary to mix compositions containing
peroxide at conditions where significant decomposition of
peroxide does not occur, and subsequently crosslink by
heating to elevated temperatures in a mold, autoclave, or
continuous vulcanization tube. To minimize ‘‘scorch’’ (pre-
mature crosslinking) during mixing (e.g., extrusion), mix-
ing is conventionally done at temperatures above the melt-
ing point of the resin, but sufficiently low so that the
selected peroxide decomposes very slowly. Scorch may
also be retarded through the use of additives such as hin-
dered phenols [9]. Because of the often large size of the
polyethylene parts, it then takes a considerable amount of
time to increase the temperature postextrusion and decom-
pose all the peroxide, in order to complete crosslinking.
POLYMER ENGINEERING AND SCIENCE—-2007
The rate and degree of crosslinking can be enhanced
through the use of cure boosters such as a methyl styrene
dimer [10].
This paper describes a novel approach of using nitroxyl ra-
dicals such as 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO)
and its derivatives to retard scorch and control cure in per-
oxide crosslinking of polyethylene. In addition, because of
the unique mechanism of action, these nitroxyl derivatives
also serve to directly functionalize the polymer. Model
compound work was conducted to understand the mecha-
nism and showed that the reaction proceeds via peroxide
decomposition followed by hydrogen abstraction to pro-
duce the polymer radical, which is then ‘‘trapped’’ via re-
action with the nitroxyl radical to yield a polymer bound
alkoxyamine. The methyl radical often produced from
industrial peroxide (e.g., dicumyl peroxide (DCP)) decom-
position also reacts with the nitroxyl radical. The results
show that when the 4-hydroxy TEMPO (h-TEMPO) de-
rivative is used for scorch suppression in peroxide cross-
linking of polyethylene, there may be a loss in ultimate
degree of crosslinking. However, stable organic free radi-
cals (SFRs) comprising more than one nitroxyl not only
suppress scorch, but also enhance the ultimate degree of
crosslinking (vide infra). Furthermore, TEMPO derivatives
having functional groups (such as hydroxyl) may be used
to introduce desired attributes into the polymer. While
TEMPO has been extensively studied as an initiator for
living free radical polymerizations, the use of TEMPO to
control free radical vulcanization and functionalization has
generally not been well studied.
EXPERIMENTAL MATERIALS
AND PROCEDURES
Resins
Two different grades of low-density polyethylene
(LDPE) were used in this investigation, referred to as
LDPE-1 and LDPE-2. Both were free of additives, to start
with, and various properties of the resins were measured.
Melt index was measured in accordance with ASTM D
1238 (1908C: I2 with 2.16 kg load). Density was deter-
mined according to ASTM D 792.
Number-average (Mn), weight-average (Mw), and z-
average (Mz) molecular weights were determined by gel
permeation chromatography (GPC). The chromatographic
system consisted of a Waters 1508C high temperature chro-
matograph equipped with a PolymerChar (Valencia, Spain)
IR4 infra-red detector. Data collection was performed using
Viscotek TriSEC software version 3 and a 4-channel Visco-
tek (Houston, TX) Data Manager DM400. The carousel
compartment and the column compartment were both oper-
ated at 1508C. The columns used were four Shodex 13-mm
HT-806M mixed-bed columns and one 15-cm HT-803 col-
umn. The solvent used for elution and sample preparation
DOI 10.1002/pen
was 1,2,4-trichlorobenzene containing 200 ppm of butylated
hydroxytoluene. The samples were prepared at a 2 mg/ml
and were dissolved for 4 h at 1608C while stirring gently.
The injection volume was 200 ml and the flow rate was 1.0
ml/min. Calibration of the GPC column set was performed
with 21 narrow polystyrene standards purchased from Poly-
mer Laboratories. The polystyrene standard peak molecular
weights were converted to polyethylene molecular weights
using the following equation, as described by Williams and
Ward [11]:
Mpolyethylene ¼ A
ðMpolystyrene ÞB :
Where M is the molecular weight, A has a value of 0.41, and
B is equal to 1.0.
Thermal transitions were determined by differential
scanning calorimetry (DSC), using TA Instruments Model
2920. The sample size was 3 mg and the following proce-
dure was used: the sample was placed in a crimped alumi-
num pan, equilibrated at 308C (for about a minute), and
heated to 1508C (at a rate of 108C/min), kept at 1508C for 2
min to ensure complete melting, cooled at 108C/min to
08C, kept at this temperature for 3 min for DSC stabiliza-
tion, and heated to 1808C at a rate of 108C/min. The peak
crystallization temperature (Tc) was obtained from the cool-
ing curve. The peak melting temperature (Tm) was obtained
from the melting curve (second heat).
The amounts of unsaturations (vinyls/1000 carbons,
vinylidenes/1000 carbons, and trans/1000 carbons) were
determined by infra-red spectroscopy. The films were
scanned on a Nicolet Nexus 470 FTIR using Omnic 6.1A
software. The collection parameters were as follows:
number of scans ¼ 64, resolution ¼ 4, Apodization ¼
Happ-Genzel, Zero filling ¼ none. Air was used as the
background. The different unsaturated structures in poly-
ethylene are as follows:
.
The methyls/1000 carbons were also measured by
infra-red spectroscopy in accordance with ASTM D 2238,
which in turn was used to compute the amount of tertiary
hydrogens present. Each time a methyl group forms a
branch on the backbone, a tertiary hydrogen is also pre-
sent as shown below:
POLYMER ENGINEERING AND SCIENCE—-2007 51

FIG. 1. Chemical structure of DCP.
The tertiary hydrogens per 1000 carbons was calcu-
lated as shown below, taking into account the methyl end
groups:
Tertiary Hydrogens/1000 C ¼ Methyls/1000 C þ
Vinyls/1000 C  23,000/Mn
Free-Radical (Peroxide) Crosslinking of Polyethylene
Peroxide and optional additives (different types of
TEMPO derivatives and antioxidant ranging in concentra-
tion from 0.25 to 2.0 wt%) were blended with polyethyl-
ene using a Brabender mixing bowl. The polymers and
optional additives (except peroxide) were mixed for 3 min
at 1258C, followed by additional 4 min mixing with
peroxide at that temperature, in the mixing bowl. The tem-
perature was low enough to avoid significant decomposi-
tion of the peroxide, while still melting the polymer. The
free-radical generators studied were DiCup R DCP and
2,5-bis(tert-butylperoxy)-2,5-dimethyl-3-hexyne [Luperox
130 (L130)]. The structures of the free-radical generators
are shown in Figs. 1 and 2. The antioxidant used was
Irganox 1081 FF [6,60 -di-tert-butyl-2,20 -thiodi-p-cresol;
CAS number 90-66-4], a product of Ciba. The nitroxyls
evaluated were 4-hydroxy 2,2,6,6-tetramethylpiperidin-1-
oxyl (h-TEMPO) and bis(1-oxyl-2,2,6,6-tetramethylpiperi-
dine-4-yl)sebacate [Trade name: PROSTAB (Trademark
of Ciba) 5415; CAS number 2516-92-9]. In this paper,
PROSTAB 5415 has been referred to as ‘‘bis-TEMPO’’.
Crosslinking kinetics of the blends were evaluated using
a Monsanto Moving Die Rheometer (Model MDR 2000)
from Alpha-Technologies, set at 100 cycles per minute
and an arc of 0.58. The torque data correlated to the
degree of crosslinking as a function of cure time. The
FIG. 2. Chemical structure of 2,5-bis(tert-butylperoxy)-2,5-dimethyl-3-hexyne (L130).
52 POLYMER ENGINEERING AND SCIENCE—-2007
minimum torque is a measurement of the viscosity of the
uncured compound at molten state. This measurement can
show the difference in viscosity between two samples.
Maximum torque is a measurement of the shear modulus
or stiffness of material after full crosslinking or
cure. Change in torque or DT is the difference between
the minimum torque and the maximum torque. The DT
provides a good idea of degree of crosslinking. For this
study, the temperature in the MDR chamber was set at
temperatures of 140 or 1828C. About 6 g of sample were
placed on the disk (between Mylar or Teflon films), and
the test was started and programmed to stop after certain
lengths of time. After the test was stopped, the cross-
linked product was removed. Some of the crosslinked
products from the MDR experiments were submitted for
decalin extractables, which is a method used for meas-
uring the amount of gels in crosslinked polyethylene. This
method, ASTM D 2765, Method A, determines the per-
centage of a sample that is soluble in decalin (boiling
point ¼ 1908C) over 6 h. The material which is insoluble,
and therefore does not filter through a 120 mesh screen,
is the crosslinked portion of the material (and is reported
as the gel content).
Samples containing LDPE-2, various loadings of peroxide,
and hydroxy TEMPO were evaluated, both before and after
running in MDR, by NMR spectroscopy, both in solution
and in solid-state experiments. The solid-state NMR experi-
ments were carried out on a Varian/Chemagnetics CMX
Infinity-200 spectrometer using a Chemagnetics 7.5-mm
HX magic-angle spinning (MAS) probe. The 1H and 13C
resonance frequencies were at 200.07 and 50.31 MHz,
respectively. The polymer samples were cut into small
pieces before they were packed into MAS rotors. The
MAS rates were set at 6.0 kHz. The 13C single-pulse MAS
spectra of the samples were acquired with a recycle delay
of 1.4 s, 58,000 scans, and proton composite-pulse decou-
pling during the 410 ms data acquisition period [12]. The
13
C NMR chemical shifts were referenced to TMS using
glycine as the secondary external standard.
Solution NMR was performed at 400 MHz on poly-
mers as 10% solutions in tetrachloroethane-d2/orthodi-
chlorobenzene with chromium acetylacetonate added as a
relaxation agent. Polymers were run with the probe at
1158C. For the partially crosslinked resins, spectra were
acquired on 0.3 g polymer dissolved in TCE/ODCB (with
relaxation agent) and data was collected at 1158C for
10,000 scans with a 6 s recycle delay.
DOI 10.1002/pen
 TABLE 1. Melt index, density, and thermal transitions of polyethylenes.
Resin Melt index, I2 (dg/min) Density (g/cc)
LDPE-1 2.1 0.9184
LDPE-2 2.4 0.9200
Model System Investigation
A model system of TEMPO and DCP in the presence
of hexadecane as a substrate was studied by NMR. Hexa-
decane is an easily mixed liquid at room temperature and
is suitable for spectroscopic analysis. It also has both pri-
mary and secondary targets for radical attack so it was
hoped that these might indicate whether there is a prefer-
ence for one site over another in this reaction. NMR solu-
tion experiments were performed on a Varian INOVA
400 MHz spectrometer. Generally, experiments on the
model system of peroxide in hexadecane were performed
using a capillary containing D2O to provide a lock signal
so as to avoid any issue with radical reaction with the
solvent. NMR for the model work was performed at room
temperature, where the reaction kinetics are very slow
relative to the observation time.
Heating for the model experiments was performed out-
side of the magnet using an aluminum block milled to
hold NMR tubes heated on a hot plate. The temperature
of a solution was measured using a thermocouple and
found to take 5 min to reach 1408C in this apparatus. Lag
time for heating was simply corrected by subtracting
5 min from the time of each heating.
Samples of the pure components and mixtures were
examined by 1H and 13C NMR spectroscopy before and
after heating. Because of the fact that the unpaired electron
in TEMPO causes rapid magnetic relaxation of nearby
spins, the TEMPO radicals themselves are unobservable
and also lead to broadening in the observed lines. This
makes 1H NMR less useful than 13C NMR where the in-
herent spectral dispersion is greater. After the radicals
have reacted, it is possible to detect any products.
RESULTS AND DISCUSSION
Resin Characterization
The measured melt index, density, and thermal transi-
tions of the polyethylene resins are given in Table 1.
Table 2 contains data on the various types of unsaturation
TABLE 2. Unsaturation and tertiary hydrogen of polyethylene resins.
Resin Trans/1000 C Vinyl/1000 C
LDPE-1 0.044 0.179
LDPE-2 0.032 0.270
DOI 10.1002/pen
Tm (8C) Tc (8C) Heat of fusion (J/g) Heat of crystallization (J/g)
110.0 97.6 117.1 104.4
110.2 96.6 120.1 107.6
and tertiary hydrogens in the polyethylenes. The type of
short chain branching in LDPE (i.e. level of methyl,
butyl, pentyl, ../1000 C) was not determined. The mole-
cular weights are given in Table 3. The differences in
molecular weights and unsaturations of the two polymers
were particularly significant.
Effects of Nitroxyls and Hindered Phenol on Peroxide
Crosslinking of LDPE
The results of the crosslinking experiments are
presented in Figs. 3–6. It is well known that hindered
phenols interfere with free-radical crosslinking of polyeth-
ylene resulting in decreased rate and degree of crosslink-
ing. This has been demonstrated herein through the use of
Irganox 1081 FF (6,60 -di-tert-butyl-2,20 -thiodi-p-cresol),
as shown in Figs. 3 and 4. It is interesting to compare the
effect of h-TEMPO with that of Irganox 1081 FF (albeit
at slightly different concentrations). There was no particu-
lar reason for using 0.28 wt% of Irganox 1081 FF, instead
of 0.25 wt% concentration of h-TEMPO. Decreasing the
amount of Irganox 1081 FF to 0.25 wt% would not have
greatly affected the results. Both additives had remarkably
similar detrimental effects on the rate and degree of cross-
linking of LDPE-1 at 1828C (see Fig. 4). However, there
were noticeable differences in kinetics of scorch suppres-
sion at 1408C (see Fig. 3). Unlike the hindered phenol,
hydroxy TEMPO yielded a flat induction time at the
beginning. This unusual induction period is presumed to
correspond to rapid trapping of carbon-centered polymer
and methyl radicals generated from peroxide decomposi-
tion, leading to partial grafting of the TEMPO derivative.
That is, h-TEMPO essentially quenched crosslinking of
polyethylene in the first few minutes at 1408C, whereas
the hindered phenol only retarded the rate of crosslinking
in that period of time. The net effect was relatively less
scorch within the first 20 min at 1408C with h-TEMPO.
However, over a longer period of time, Irganox 1081 FF
was a more effective scorch retardant than h-TEMPO.
Similar trends were observed with LDPE-2.
Vinylidene/1000 C Methyls/1000 C Tertiary hydrogen/1000 C
0.051 20.9 19.7
0.015 19.8 18.9
POLYMER ENGINEERING AND SCIENCE—-2007 53
 TABLE 3. Molecular weights of polyethylene resins.
Resin Mn Mw Mz
LDPE-1 16,490 76,990 217,600
LDPE-2 19,230 67,990 198,500
The effect of bis-TEMPO on scorch and cure kinetics
of LDPE-2 can be seen in Figs. 5 and 6, and similar obser-
vations were made with LDPE-1. The addition of 0.25 wt%
bis-TEMPO resulted in increased scorch at 1408C and
increased cure at 1828C, in contrast to the use of
h-TEMPO. However, increasing the amount of bis-
TEMPO from 0.25 to 1 wt% resulted in decreased rate
and degree of crosslinking at both 140 and 1828C. The
gel contents of the samples crosslinked at 1828C were as
follows, and generally increased with higher ultimate
torque: 78 wt% gels without bis-TEMPO; 84 wt% gels
with 0.25 wt% bis-TEMPO; 80 wt% gels with 0.5 wt%
bis-TEMPO; and 74 wt% gels with 1 wt% bis-TEMPO. It
is hypothesized that one end of bis-TEMPO molecules
first trapped polymer radicals (as well as methyl radicals
from peroxide decomposition) followed by radical-trapping
with the pendant nitroxyls. At low concentrations of bis-
TEMPO, the pendant nitroxyls were used efficiently,
resulting in increased crosslinking of the polymer. At
increased concentrations of bis-TEMPO, the mobility of
the polymer-bound nitroxyl was restricted such that radical-
trapping was increasingly due to the ‘‘free’’ bis-TEMPO,
with the pendant nitroxyls being used less efficiently,
thereby resulting in decreased crosslinking (compared
with 0.25 wt% bis-TEMPO). Be that as it may, the use of
1 wt% bis-TEMPO still yielded the desired combination
of fast and high degree of crosslinking at 1828C, and
increased scorch protection at 1408C (unlike h-TEMPO
and Irganox 1081 FF). It is possible that the amount of
peroxide could be decreased in combination with 1 wt%
bis-TEMPO, in order to achieve a fixed degree of cross-
linking, which in turn would further enhance scorch pro-
tection at 1408C. Thus, SFRs comprising more than one
nitroxyl act not only as scorch-retardants in the initial
FIG. 3. Scorch suppression of LDPE-1 with h-TEMPO and Irganox
1081 FF at 1408C (with 1.7 wt% DCP).
54 POLYMER ENGINEERING AND SCIENCE—-2007
Mw/Mn
4.7
3.5
FIG. 4. Cure suppression of LDPE-1 with h-TEMPO and Irganox 1081
FF at 1828C (with 1.7 wt% DCP).
stages of cure, but also as cure-boosters for the ultimate
degree of crosslinking.
Samples based on LDPE-2, with various loadings of
DCP and 2% of hydroxy TEMPO, were evaluated by
NMR both before and after processing in the MDR. The
analyses were carried out both in solution and in solid
state. The data showed evidence of grafting of the
TEMPO derivative (Figs. 7 and 8). The primary indica-
tion for grafting in these systems is the existence of a
methine in the region around d ¼ 82 ppm. Since it may
be possible to have oxygen–carbon bonds formed in the
presence of the peroxide but without TEMPO, controls
involving only peroxide were also examined. The pres-
ence of an oxygenated carbon that is only found in the
presence of TEMPO is a strong evidence that TEMPO
grafting is occurring in the polymers.
Figure 7 shows a solution NMR spectrum of a sample
of polyethylene that had been reacted with h-TEMPO and
DCP in a MDR for 15 min at 2008C. Under these condi-
tions, the polymer had not reached its gel point and
was still soluble. The expanded spectrum showed signals
associated with both grafted and methylated h-TEMPO.
The oxygenated carbon associated with TEMPO grafting
was observed at 82.6 ppm, along with signals associated
with methylated TEMPO. In the polymer, these signals
were approximately equal in intensity.
FIG. 5. Effect of PROSTAB 5415 bis-TEMPO on scorch characteris-
tics of LDPE-2 at 1408C (with 1.7 wt% DCP).
DOI 10.1002/pen

FIG. 6. Effect of PROSTAB 5415 bis-TEMPO on crosslinking of
LDPE-2 at 1828C (with 1.7 wt% DCP).
Solid-state NMR was also useful in observing the pres-
ence of grafting in polymers after they had been substan-
tially crosslinked. The best conditions for observing graft-
ing were at 1308C, where there was considerable motion
in the polymer chains to reduce the line broadening expe-
rienced in solid-state NMR. However, caution must be
used in interpreting the results under these conditions as
well: at 1308C, the reaction of TEMPO and a peroxide
can be driven further, making quantitative assessments of
polymer samples difficult. Figure 8 shows two samples of
the same resin blended with h-TEMPO and examined at
room temperature and 1308C by NMR. The samples were
observed for 22 h to achieve adequate signal. The top
traces are solid state (MAS) NMR spectra of LDPE-2
resin blended with 1.5 wt% DCP and 0.5 wt% hydroxy
TEMPO, observed at room temperature and 1308C. The
bottom traces are MAS NMR spectra of the same poly-
mer blend processed by MDR at 1408C for 20 min, again
observed by NMR at room temperature and 1308C. Note
the increased signal due to grafted hydroxy-TEMPO in
the bottom traces. The room temperature signals were
13
FIG. 7. C NMR spectrum of LDPE-2 containing 2% h-TEMPO and 1.5% DCP, processed at 2008C for
15 min.
DOI 10.1002/pen
broad and there was no clear evidence of TEMPO graft-
ing. At elevated temperatures, grafting was observed in
both samples. However, the first sample was not treated
at elevated temperatures prior to examination by NMR, so
the conclusion was that the grafting must have occurred
during the NMR experiment.
The data presented thus far substantiate that TEMPO
derivatives do indeed graft to carbon-centered radicals ini-
tiated by decomposition of DCP. Thus, TEMPO deriva-
tives can not only be used to retard scorch, but also to
functionalize polymers (using hydroxy TEMPO, amino
TEMPO, and other possible derivatives) and to crosslink
polymers with SFRs comprising more than one nitroxyl.
However, a critical requirement for functionalization and
crosslinking with TEMPO derivatives would be to maxi-
mize the grafting, versus any other trapping reactions
between TEMPO and carbon-centered radicals (e.g. methyl
radicals) arising spontaneously from peroxide decomposi-
tion. DCP produces significant amounts of methyl radical
on decomposition. Thus, one approach to maximize the
amount of TEMPO grafted would conceivably be the use
of peroxides that generate fewer methyl radials than DCP,
such as 2,5-bis(tert-butylperoxy)-2,5-dimethyl-3-hexyne
[Luperox 130 (L130)]. To study this, comparative samples
were prepared by mixing 2 wt% h-TEMPO and varying
loadings of DCP or L130 with LDPE-2, followed by cure
in a MDR at 2008C for 15 min. The maximum torques
measured in the MDR are presented in Fig. 9. The MDR
torque data indicate that about 1.0 wt% L130 or 1.5 wt%
DCP would be optimal for grafting 2 wt% hydroxy
TEMPO to LDPE-2, without large increases in viscosity
(in order to retain good extrudability or processability).
The samples were subsequently analyzed by solution state
NMR to determine the grafting efficiency and the amounts
POLYMER ENGINEERING AND SCIENCE—-2007 55
 FIG. 8. Solid-state (MAS) NMR spectra of LDPE-2 resin blended with 1.5 wt% DCP and 0.5 wt% hydroxy
TEMPO, observed at room temperature and 1308C (top traces), and the same polymer blend processed by
MDR at 1408C for 20 min, again observed by NMR at room temperature and 1308C (bottom traces).

of h-TEMPO that formed adducts with the methyl radical Model Compound Investigations
generated from peroxide decomposition (see Fig. 10). Note
To understand the mechanisms of scorch suppression,
that ‘‘methylation’’ refers to the formation of the NOCH3
cure control, and functionalization with nitroxyls in per-
derivative by quenching of methyl radical with h-TEMPO.
oxide crosslinking of polyethylene, a model system of
The results confirmed that L130 was significantly more
TEMPO and DCP was studied in the presence of hexade-
efficient than DCP at grafting hydroxy TEMPO to LDPE,
cane as a substrate. The 13C experiments conducted in
and yielded much less of the methylated TEMPO. It is
this study showed that the heating of DCP in hexadecane
interesting to note that the total amount of h-TEMPO
led to the reaction scheme of Fig. 11. The DCP first
accounted for by NMR was 1.4 wt% with DCP and
decomposed thermally to dicumyloxy radicals that further
1.3 wt% with L130, significantly less than the 2 wt%
decomposed to methyl radical plus phenylisopropanol. It
added to the composition. It is reasonable to suppose that
the remainder of the h-TEMPO would have volatilized
out of the melt pool, and that the relatively faster decom-
posing DCP would have resulted in fewer losses because
of volatility than L130. It is interesting to note that when
hydroxy TEMPO was melt-blended with polyethylene,
the color of the blend turned light brown because of the
orange color of hydroxy TEMPO. However, as the tem-
perature was raised to decompose the peroxide, the
brown or orange color disappeared. Since the h-TEMPO
radical is highly colored, the loss of color in the sample
indicates that the h-TEMPO was substantially reacted or
lost through volatilization rather than remaining in its FIG. 9. Grafting of 4-hydroxy TEMPO to LDPE in MDR at 2008C for
radical form. 15 min.

56 POLYMER ENGINEERING AND SCIENCE—-2007 DOI 10.1002/pen

 FIG. 10. NMR results showing improved grafting of 4-hydroxy TEMPO to LDPE with L130.

is known that several variables can affect the kinetics
(and by extension the initiation temperature) of peroxide
decomposition, including the structure of peroxide,
solvent matrix, contaminants such as alcohols, olefins,
and others, the presence of alkaline substances, and free
radical sources [13, 14]. Effects can be both enhancing
and retarding, depending on the conditions. The reaction
is initiated by decomposition of the peroxide, which leads
to hydrogen abstraction from the polymer, followed by
bimolecular reaction of two radicals to form a crosslink
in the polymer. In situ ESR studies of peroxide reactions
in polymers have detected only alkyl radicals but
no oxygen centered radicals, indicating that the abstrac-
tion of protons from the polymer substrate is quite rapid
[15–17]. Crosslinking reactions are comparatively slow,
presumably because the diffusion of polymer radicals is
constrained, but eventually the polymer network develops.
Figure 12 and the associated Table 4 show the expected
products from a reaction involving TEMPO, DCP, and
hexadecane (C16) with their predicted and observed 13C
chemical shifts. Identification of the peaks was accom-
plished using control experiments, distortionless enhance-
ment by polarization transfer (DEPT) experiments, com-
parison of peak intensities, and comparison to predicted

FIG. 11. Crosslinking of polyolefins by initiation, propagation, and termination of DCP (major pathways).

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2007 57

 FIG. 12. Major products from the reaction of TEMPO with hexadecane in the presence of DCP. Numbered
carbons correspond to the assignments in Table 4.

shifts. In the case of some carbons, multiple peaks were tions on the hydrocarbon substrate or to C16 versus
observed indicating the production of several related methyl radicals. The carbon labeled 2 and 6, for instance,
structures, such as adducts of TEMPO to different loca- shows five resolved peaks of differing intensities, corre-

TABLE 4. Characteristic shifts of compounds in the TEMPO-DCP-hexadecane system.

Carbon number Associated structure Attached protons ACD shift predictiona (ppm) Observed shift (ppm)b

8 TEMPO-C16 1 78.2 6 3 82.6 (82.3–83.8)c

2, 6 TEMPO-C16 0 59.6 6 1.5 59.6–50.6c
3, 5 TEMPO-C16 2 40.5 6 1 40.3–41.2c
4 TEMPO-C16 2 17.1 6 1 17.9–18.7c
19, 22 TEMPO-C16 3 20.1 6 2 20.4?d
20, 21 TEMPO-C16 3 20.6 6 2 26.8?d
35 TEMPO-C16 (methyl attack) 2 72.9 6 2 78.8?d
58 Methylated h-TEMPO, etc. 1 60.9 6 2 63.0–63.5c
62 (75) Methylated h-TEMPO (TEMPO) 3 61.8 65.7
76, 79 Methylated TEMPO 3 31 6 4 26.8?d
77, 78 Methylated TEMPO 3 21.5 6 1.5 20.4?d
86, 89 DCP 0 82.0 6 3 82.0
90, 91, 92, 93 DCP 3 26.5 6 0.5 27.5
106 Phenylisopropanol 0 72.2 72.5
108, 109 Phenylisopropanol 3 31.5 32.4
116 Acetophenone 0 197.3 195.6
118 Acetophenone 3 26.0 26.4
127, 134 Hexadecane 3 14 6 1 14.8
126, 133 Hexadecane 2 22.7 23.6

a
Advanced Chemistry Development, Inc. ACD/CNMR Predictor 7.00 release (version 7.05).
b
Referenced to DCP at 82.0 ppm.
c
Multiple related structures are observed.
d
? Indicates uncertainties in the NMR assignments.

58 POLYMER ENGINEERING AND SCIENCE—-2007 DOI 10.1002/pen

 13
FIG. 13. C NMR spectrum of 15% DCP and 3% TEMPO in hexadecane after 25 min at 1408C.
sponding to the methyl adduct and at least three distinct
attachments to the C16 substrate. At least one characteris-
tic carbon for each of the compounds has been identified
in the 13C spectrum. Many of the products have also been
observed by mass spectroscopy and liquid chromatogra-
phy of the reaction solutions and polymers reacted in the
presence of TEMPO (vide infra).
The 13C NMR spectra obtained after heating of DCP
and TEMPO confirmed that one of the primary sinks for
TEMPO is indeed adduction to the methyl radical gener-
ated after thermal decomposition of DCP. This was
consistent with the observations from studies in polyethyl-
ene reported earlier in this paper and has profound impli-
cations for the control of where TEMPO or its derivatives
end up in a given formulation. For instance, to drive
toward TEMPO modification of the polymer, it is desira-
ble to use a peroxide that does not form small alkyl radi-
cals after decomposition.
The ultimate fate of the reactants can be observed in
13
C spectra taken at high extent of reaction (see Fig. 13).
For a 15 wt% DCP solution, 60–70% of the starting DCP
was observed to react after 25 min of heating at 1408C.
At this point in the reaction, the ratio of phenyl isopropa-
nol to acetophenone was 1.1:1. That is, the DCP had
decomposed to approximately equal amounts of oxy and
methyl radicals. The extent of reaction in terms of
TEMPO at this point was found from the NMR spectrum
to be 70%. Of the reacted TEMPO, the ratio of TEMPO
adducted to hexadecane to TEMPO adducted to methyls
was 1.3:1. So, most of the methyl radicals were scavenged
by TEMPO immediately upon formation. A small fraction
of the methyl radicals must have abstracted protons from
the alkane substrate, which were then quenched by
TEMPO, to account for the increase in C16-TEMPO rela-
tive to methyl TEMPO. Considering that the mole ratio of
carbons available at the substrate to TEMPO is 300, the
fact that TEMPO reacted with methyl radicals at all
DOI 10.1002/pen
strongly indicates that the quenching reaction was diffu-
sion controlled and essentially statistical. As soon as
TEMPO encountered a carbon centered radical, it was
trapped or quenched.
Further evidence of the nature of this reaction was ob-
tained by examining the TEMPO adducts of hexadecane.
For instance, the resonances near d ¼ 82 ppm, corre-
sponding to carbons on the alkyl substrate bound to
TEMPO, had an intensity pattern (within the limits of ex-
perimental error) of 4:1:1:1 (see Fig. 14). DEPT-90 NMR
experiments, which show only carbons with a single pro-
ton, revealed that there was a methine of equal intensity
to the 82.3 and 83.8 ppm peaks under the DCP peak at
82 ppm (data not shown). This was consistent with a
statistical distribution of TEMPO on the chain, where
NMR at 400 MHz only resolves differences in carbons
closer than d to the methyl ends. With a total of 16
carbons per chain, eight carbons were unresolved, with
symmetry giving a total of five observable resonances
(the methyl-attack resonance was not near d ¼ 82 ppm)
of intensity 8:2:2:2:2. Figure 15 shows the reaction
scheme of the TEMPO, DCP, and hexadecane system.
Stability of the Alkoxyamine
An important feature for functionalization as well as
crosslinking of polymers with nitroxyls is the stability of
the graft, since elimination of the alkoxyamine is known
to occur under certain conditions. The thermal stability of
bonds formed between the nitroxyl and carbon-centered
radicals has been studied extensively in model systems
[18–23]. The results presented in these studies indicate
that primary and secondary alkoxyamines should be robust
enough to withstand processing and testing conditions at
elevated temperatures for many hours to several days or
longer, such that the integrity of grafts to primary and sec-
ondary polymeryl carbons will be maintained. In contrast,
POLYMER ENGINEERING AND SCIENCE—-2007 59
 13
FIG. 14. Expansion of C NMR spectrum of 15% DCP/3% TEMPO/C16 solution after 25 min at 1408C.

grafts to tertiary carbons are predicted to be unstable with
respect to alkene elimination and this reaction will occur
relatively rapidly under testing/processing conditions. Thus,
the fraction of grafts that occur to tertiary positions will be
important in determining the thermal stability of the
alkoxyamine. The primary/secondary/tertiary selectivity
will be strongly dependent on the relative abundance of
each type of CH bond (statistics) as well as the choice
of peroxide used. For polymers with low branching, stable
TEMPO grafts should still be possible. Since tertiary

FIG. 15. Reactions of TEMPO in the presence of DCP and an alkane substrate.

60 POLYMER ENGINEERING AND SCIENCE—-2007 DOI 10.1002/pen

CH bonds are the weakest, increasing ‘‘selectivity’’ will
enhance the fraction of tertiary-centered grafts. Using an
unselective peroxide (e.g., a fast-acting one that acts
mostly through oxy radicals) and polymers, which are stat-
istically biased toward primary and secondary hydrogen
abstraction, may be beneficial.
CONCLUSIONS
The reaction of carbon-centered radicals with nitroxides
such as TEMPO and its derivatives has been demon-
strated to be a novel means for scorch suppression and
cure control in peroxide crosslinking of polyethylene, and
can also enable functionalization of the polymer. A model
study in hexadecane showed that the mechanism involves
the selectivity of TEMPO for radical termination with
carbon-centered radicals formed after peroxide homolysis
and propagation steps before peroxide-initiated crosslink-
ing. The buildup of radical concentration is controlled
through the introduction of the nitroxyls, which are able
to trap the early forming radicals from the peroxide. The-
oretically, alkyl radical concentrations are kept low by the
TEMPO and propagation is stopped before crosslinking
can occur. Once the TEMPO is consumed, crosslinking
proceeds as usual. By varying the concentrations and
types of TEMPO derivatives and peroxides, the balance
of scorch and cure control can be varied.
When 4-hydroxy TEMPO was used for scorch suppres-
sion in peroxide crosslinking of polyethylene, there was
often a loss in ultimate degree of crosslinking. However,
when bis-TEMPO was used, some degree of scorch sup-
pression was still obtained, but as the pendant nitroxyls
grafted onto other polymer chains, the ultimate degree of
crosslinking was enhanced.
The findings of this study will be applicable to olefinic
polymers in general and with a variety of nitroxyls, with
any form of free radical generators (including radiation).
It is important to emphasize that the rate constants for
reactions of carbon-centered radicals with nitroxides will
depend strongly on the structures of the nitroxide and the
carbon-centered radical [24]. Furthermore, the architecture
of the olefinic polymer will strongly influence the results.
For instance, in the case of propylene polymers, which
are susceptible to chain scission with free-radical chemis-
tries, quenching with nitroxyls followed by alkene elimi-
nation may still be effective at preventing chain scission
at tertiary sites under free radical conditions.
ACKNOWLEDGMENTS
Thanks are extended to Tom Peterson for his analysis
of the literature on alkoxyamine stability and to Jincheng
DOI 10.1002/pen
Xiong for conducting the solid-state NMR measure-
ments.
REFERENCES
1. M. Dorn, Adv. Polym. Technol., 5, 87 (1985).
2. J.V. Gulmine and L. Akcelrud, J. Appl. Polym. Sci., 94, 222
(2004).
3. T. Bremner, A. Rudin, and S. Haridoss, Polym. Eng. Sci.,
32, 939 (1992).
4. M. Lazár, A. Kleinová, A. Fiedlerová, J. Janigová, and
E. Borsig J. Polym. Sci. Part A: Polym. Chem., 42, 675
(2004).
5. G. Wu, Y. Katsumura, H. Kudoh, Y. Morita, and T. Segu-
chi, J. Polym. Sci. Part A: Polym. Chem., 37, 1541 (1999).
6. D. Abraham, K.E. George, and D.J. Francis, Die Ange-
wandte Makromolekulare Chemie, 200, 15 (1992).
7. D. Abraham, K.E. George, and D.J. Francis, J. Appl. Polym.
Sci., 67, 789(1998).
8. D. Abraham, K.E. George, and D.J. Francis, J. Appl. Polym.
Sci., 81, 2545 (2001).
9. B. Gustafsson, J.-O. Bostrom, and R.C. Dammert, Die
Angewandte Makromolekulare Chemie, 261/262, 93
(1998).
10. T. Yamazaki and T. Seguchi, J. Polym. Sci. Part A: Polym.
Chem., 39, 2151 (2001).
11. T. Williams and I.M. Ward, J. Polym. Sci. Part B: Polym.
Lett., 6, 621 (1968).
12. A.E. Bennett, C.M. Rienstra, M. Auger, K.V. Lakshmi, and
R.G. Griffin, J. Chem. Phys., 103, 6951 (1995).
13. M.S. Kharasch, A. Fono, and W. Nudenberg, J. Org. Chem.,
16, 113 (1951).
14. O.L. Mageli and J.R. Kolczynski, Ind. Eng. Chem., 58, 25
(1966).
15. W. Zhou and S. Zhu, Macromolecules, 31, 4335 (1998).
16. T. Yamazaki and T. Seguchi, J. Polym. Sci. Part A: Polym.
Chem., 35, 2431 (1997).
17. W. Zhou and S. Zhu, Ind. Eng. Chem. Res., 36, 1130 (1997).
18. C. Wetter, K. Jantos, K. Woithe, and A. Studer, Org. Lett.,
5, 2899 (2003).
19. S. Marque, H. Fischer, E. Baier, and A. Studer, J. Org.
Chem., 66, 1146 (2001).
20. M.V. Ciriano, H.-G. Korth, W.B. van Scheppingen, and
P. Mulder, J. Am. Chem. Soc., 121, 6375 (1999).
21. M.E. Scott, J.S. Parent, J. Dupont, and R.A. Whitney, Ind.
Eng. Chem. Res., 42, 3662 (2003).
22. P. Marsal, M. Roche, P. Tordo, and P. de Sainte Claireqq,
J. Phys. Chem., 103, 2899 (1999).
23. E. Minaux, ‘‘Controlled Radical Polymerization at Pressures
up to 2000 Bar’’, dissertation, Georg August Universität,
Göttingen, Germany, 150, 153 (2001).
24. J. Sobek, R. Martschke, and H. Fischer, J. Am. Chem. Soc.,
123, 2849 (2001).
POLYMER ENGINEERING AND SCIENCE—-2007 61