Results and Discussion
The present study was carried out as a follow up of our previous
publications presenting the 1H LF-NMR 2D
T1-T2 graphical maps for butter,
rapeseed oil, soybean oil and linseed oil (Resende et al., 2019a,b). In
these articles, we showed that the composition of the
saturated/unsaturated fatty acids affects the oil’s (TAG) chemical and
morphological energy relaxation time domains, which could be correlated
to oxidative propensity. Furthermore, we reported that using accurateL 1/L 2 norm regularization
parameters for prima dual convex optimization (PDCO) solver for1H LF-NMR relaxation data processing (Campisi et al.,
2018; 2019), could reconstruct the multi-exponential accurate 1D
T2 and 2D T1-T2 time
domains of TAG and fatty acids segmental motions. This allowed the
assignment of each segment’s (e.g., glycerol, aliphatic chain, double
bonds and tail segments) mobility in terms of T1 and
T2 during the sample’s thermal autoxidation (Resende et
al., 2019b). The 1H LF-NMR 2D
T1-T2 relaxation method was shown to be
an efficient informative tool to characterize and monitor PUFA
oxidation. However, as was also stated by Hwang et al. (2017) the
relative long time required for 2D T1-T2operation is limiting its application as an on-line and also at-line in
industrial production processes. In the present study, we aim to develop
a rapid facile application based on selective NMR relaxometric
assessment of the aliphatic tail’s segmental mobility of omega-3
PUFA-rich LSO, to monitor the oil’s chemical and structural changes
during autoxidation.
Considering the fact with current data that in cases of relatively small
molecules such as oils, T1 is almost equal to
T2 (Berman et al., 2013a) and the length of time
required for T2 (spin-spin) determination is very short,
and it is much shorter than the time required for T1(spin-lattice). Thus by focusing on T2 determination,
shorter times of material characterization are possible. Indeed the 2D
T1-T2 is providing additional important
information, especially in the final oxidation phase of
termination/polymerization of LSO, characterized as a ”bending effect”
in which T2 is significantly decreased and
T1 remains constant (Resende et al., 2019a,b). However,
the fact that single T2 test provides somewhat less
chemical and structural information than 2D
T1-T2 it can be compensate by
multi-exponential selective assessment of the transverse relaxation time
of the LSO tail as shown in Fig. 1a,b&c, wherein increasing temperature
of the LSO sample together with air/O2 pumping and even
with only temperature increase without air pumping (Berman et al., 2015,
Meiri et al., 2015), shows H abstraction from the bisallylic carbons in
omega-3 PUFAs, and a subsequent structural rearrangement yielding a
conjugated diene can be observed, as also reported by Hwang (2015),
using 1H high filed NMR. Following this change a
decomposition of the PUFA chain results in separation of alpha, beta
unsaturated aldehydes from the original PUFA chain. This reaction is
well described and documented (Vieira et al., 2017; Gorkum and Bouwman,
2005). Small fractions of low molecular weight aldehydes such as
malonaldehyde (MDA) are volatized and large fractions are nonvolatile
including aldehydes such as 4-hydroxy-trans-2-nonenal (HNE);
4-hydroxy-trans-2-hexanal (HHE) remains in the oxidized oil sample and
interacts with other oxidation chain reaction products to form highly
crosslinked polymer products (Gorkum and Bouwman, 2005). In Fig. 1a, a
brief decomposition scheme is given for LSO omega-3 PUFA, consisting of
55% linolenic acid (18:3). The chemical structure of LSO omega-3
linolenic acid and omega-9 oleic acid tail segments are marked by a red
circle, and the next bisallylic segments of PUFA chains can be seen. The
most common alpha, beta unsaturated aldehydes decomposed/released by
products, MDA, HHE and HNE (MW= 72.06, 114.1, 156.22, respectively) have
been identified in oxidized LSO samples by GC-MS analyses and are shown
in Fig 1a.
Our group developed in the recent years a special multi-exponential
computing reconstruction data processing program, based on a prima dual
convex optimization (PDCO) solver for 1H LF-NMR
relaxometry signal inverse Laplace transformation (ILT) (Berman et al.,
2013a,b; 2015; Meiri et al., 2015, Wiesman et al., 2018; Campisi et al.,
2018; 2019). This novel reconstruction system has been reported as an
efficient tool to distinguish between different molecular ensembles in
complex systems with differential segmental motion of molecular
components, and/or different morphologies (Berman et al., 2015; Meiri et
al., 2016; Resende et al., 2019a,b; Campisi et al., 2018; 2019). This
system’s capability to differentiate the different T2relaxation time components of LSO (glycerol, double bonds, aliphatic
chains, and tails) is demonstrated for LSO (Fig 1b). It could be readily
observed from the time domain (TD) peaks, that the tail segment with the
highest T2 values (818 ms) and is the most mobile among
all the LSO structural segments. In this regards it should be noted that
the T2 relaxation time value of the tail includes all of
the different fatty acid chains in the LSO, wherein due to its high
content, the T2 relaxation time of the PUFA tails is the
most mobile, with the highest values and dominates this TD peak. Using
this selective segments TD system, we recently developed and
demonstrated a method for selective assessment of the LSO tail
T2 relaxation time (Resende et al., 2019a,b), which
could determine the TD signal of tail’s segmental mobility during
present LSO oxidation experiments at 25, 40, 60, 80, 100,
120oC for a time period up to 168 h (Fig. 1c). The
data shows that the tail segment’s T2 transverse
relaxation time is initially ~750 ms in all LSO tested
samples. At 60oC T2 relaxation times
values peak (2300 ms) after 24 h and then rapidly decline until 168 h
(200 ms). At 40oC tail T2 peaks (2400
ms) after 48 h, decline until 96 h (550 ms) and then increase (1500 ms)
up to 168 h. In the control LSO sample at 25oC the
tail T2 transverse relaxation time moderately increases
and peaks after 72-96 h (1200 ms) and then returns to its original
mobility (~750 ms) after 120 h subsequently until 168 h
it moderately increases (1300 ms). Though not shown in Fig. 1c (but will
be further addressed later) the trend and slope of the best straight
line average of tail T2 of LSO at 25oC
is moderately increasing over the 168 h of oxidation. A similar trend is
shown for LSO tail T2 at 40oC. Both of
these LSO samples (25 and 40oC) are marked as Group A.
At 60oC however, the trend and slope of the LSO tail
T2 energy relaxation time is decreasing. This data of
LSO tail T2 relaxation times suggests that in the later
stages of each of the samples oxidized at 25oC and
40oC a significantly more mobile tail segment is
appearing. In the case of 60oC, however the tails’
decomposition (in term of alpha, beta unsaturated aldehyde) products are
more rapidly formed than in the other two lower temperatures of 40 and
25oC. This may be rationalized that in the case of
60oC the very mobile TD peak after 24 h is associated
with released tail-aldehyde and the later continuously decreased
mobility is possibly due to a crosslinking with another fatty acid chain
of the oxidized LSO, forming a viscous gel-like polymeric product. It
should be noted that such temperature effect on increase of fatty acids
T2 values was already reported and described in details
(Meiri et al., 2015).
In the case of 40oC oxidation it is postulated that
the relatively low heat energy cannot induce polymerization reactions
resulting only an initial rapid increase in the tail’s
T2, that is explained by H abstraction due to
introduction of heat/energy enough for decomposition of H bonds,
followed by a continuous moderate structural rearrangement in the tail
segment until the end of the experiment at 168 h (as supported by the
study of the changes of viscosity and self-diffusion that will be
presented and discussed later). In the control case of
25oC oxidation the kinetic rate of all the oxidation
reactions are slower due to too low heat energy such that the reaction
kinetics does not achieve the polymerization crosslinking phase in the
frame time of the experiment. In the cases of higher oxidation
temperatures (80, 100, 120oC) marked in Fig. 1c as
Group B, the oxidation reaction kinetics is high and the release of the
decomposition aldehyde products occur already in the first few hours
(until 9 h). This is followed by a constant and continuous decline of
the alkyl tail’s T2 mobility wherein the final product
is a polymeric viscous gel. In the case of the highest temperature of
120oC, the termination phase ends after 96 h and the
T2 value is 100 ms. In the cases of 80 and
100oC the tail T2 declines up to 168 h
to values of 120 ms. Thus the termination phase via polymerization
crosslinking forming highly viscous gel materials is characterized by
low alkyl tail T2 values, representing low degrees of
mobility. Therefore, it is suggested that the rate of T2change to lower values as a function of time can be correlated with the
kinetics of the aging reactions of crosslinking polymerization. It
should be noted that in control cases that air/O2 was
not supplied or when N2 was used instead of air, to the
LSO heated samples no increase of viscosity and no decrease of tail
T2 values was observed even in high temperature
administration (see supplemental materials 3). These results suggest
that at the early stage of LSO autoxidation (initiation) the temperature
increase is dominating the process and in more advanced stage
(propagation and termination) the combination of temperature and
O2 supply is controlling the process and changes of tail
T2 can be explained accordingly.
The important molecular parameters in understanding the rates of the
initiation, propagation and terminations phases, as monitored by
T2 changes are the degree of olefin functionality and
the ratio of the olefin components with the different degrees of
functionality that control the onset of crosslinking polymerization and
gel structure formation, and the subsequent lowering of
T2 values. The linseed oil is rich in monomeric units
with multifunctional double bonds (linolenic acid (18:3), and linoleic
acid (18:2)). The number of monomer groups, in this case the alkyl
chains of the triglycerides that must react to reach the gelation point,p , can be correlated to the average degree of reactive groups per
monomer units in the mixture (fav) byp =1/fav. In the present study of linseed oil
oxidation, this gel point value due to polymerization crosslinking is
low because of the high content of multifunction 18:3 and 18:2. This is
seen by the rapid drop in linseed oils T2 tail values
because of higher crosslinking occurring at high values of reactivity of
the higher temperatures compared to the lower temperatures of
autoxidation (Figure 1c). This may also be readily correlated with
viscosity changes as shown n Figure 3. At lower temperatures, the
T2 values rapidly increase due to changes in secondary
interactions possibly because of alkyl chain hydrolysis without
oxidative crosslinking effects. Thus in addition to covalent
crosslinking affects T2 may be also influenced cohesive
interactions as quantified by cohesive energy density of non-covalent
interaction (e.g. polar, ionic and van der Waals forces). The strength
of these secondary interactions are strongly influenced by temperature
and have a corresponding effect on T2 values (Berman et
al., 2015; Meiri et al., 2015).