Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve the cold-flow characteristics of biodiesel

Article abstract of DOI:10.1016/j.cattod.2014.02.041

Triglyceride autoxidation using a homogeneous Co/Mn/Zr/bromide catalyst in acetic acid (93%) of low grade tallow, canola oil or soy bean oil in a batch reactor at 150 °C for 2 h, produced lower molecular weight products relative to the fatty acids of the starting triglycerides. For the autoxidation of tallow the main products after esterification were monoesters Me(CH ₂)_mC(O)OMe (m = 5-12) and diesters MeOC(O)(CH ₂)_nC(O)OMe, (n = 7-12). Oxidation of the saturated fatty acids in triglycerides was confirmed and modelled using methyl palmitate. Post-treatment esterification of tallow autoxidation products to produce biodiesel (BD) esters resulted in improved cold temperature properties by a mean of 13.0 °C, i.e. a mean cloud point (CP) 1.0 °C (cf. unmodified tallow biodiesel: CP 14 °C).

Full text of DOI:10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
Catalysis Today xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Metal/bromide autoxidation of triglycerides for the preparation of
FAMES to improve the cold-flow characteristics of biodiesel
Peter Phung, William N. Rowlands, Appadurai Thiyakesan, Paul Benndorf,
Anthony F. Masters, Thomas Maschmeyer∗
Laboratory for Advanced Catalysis for Sustainability, School of Chemistry, University of Sydney, Sydney, NSW 2006, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
Triglyceride autoxidation using a homogeneous Co/Mn/Zr/bromide catalyst in acetic acid (93%) of low
grade tallow, canola oil or soy bean oil in a batch reactor at 150 ◦C for 2 h, produced lower molecular
weight products relative to the fatty acids of the starting triglycerides. For the autoxidation of tal-
low the main products after esterification were monoesters Me(CH2)mC(O)OMe (m = 5–12) and diesters
MeOC(O)(CH2)nC(O)OMe, (n = 7–12). Oxidation of the saturated fatty acids in triglycerides was confirmed
and modelled using methyl palmitate. Post-treatment esterification of tallow autoxidation products to
produce biodiesel (BD) esters resulted in improved cold temperature properties by a mean of 13.0 ◦C, i.e.
a mean cloud point (CP) 1.0 ◦C (cf. unmodified tallow biodiesel: CP 14 ◦C).
Received 13 September 2013
Received in revised form 20 February 2014
Accepted 22 February 2014
Available online xxx
Keywords:
Autoxidation
Co/Mn/Zr/bromide
Biodiesel
© 2014 Elsevier B.V. All rights reserved.
Cloud point
Tallow
FAME
oils (e.g. soy- or canola-based)i and yield BDs with a CP of around
1. Introduction
14 ◦C—Falling well outside the useable range [2,10,11]. Cold flow
The production of biodiesel (BD) by transesterification of tri-
glycerides with alcohols to generate fatty acid alkyl esters [1–5],
i.e. fatty acid methyl esters (FAMES) in the case of methanol, is
an established route in the area of renewable fuels. Typically, the
feedstock triglycerides are found in food crops (first-generation
biofuels). However, as an increasing plant oil demand for first-
generation biofuels leads to land use issues (rainforest vs. oil palm
plantations, fuel vs. food dilemma, etc.), it is prudent to investigate
existing waste oils as a source from which to produce BD, partic-
ularly the use of low-grade waste, e.g. tallow and related residues,
which are high in saturated fats [6–9].
Low grade tallow products, used vegetable oils and grease trap
wastes provide an ideal feedstock for a sustainable, local produc-
tion of a renewable fuel from a low-value source. However, a
significant issue is the difficulty in obtaining BD with a low cloud
point (CP), i.e. the on-set of crystallisation, when derived from such
low grade sources, as triglycerides from these sources generally
exhibit a greater degree of saturation and oxidation than plant
properties (freezing-, cloud- and gel-points) of BD are of crucial
importance for its successful application. Given the feedstock con-
straints mentioned above, it is of considerable interest to produce
low melting fatty acid esters from low grade sources [10–15].
In this contribution we set out to develop a new methodology
to obtain low CPs for tallow-derived BDs, based on an autoxidation
approach reported by Partenheimer [16], in which catalyst mix-
tures of cobalt (II) and manganese (II) acetates and zirconium (IV)
chloride and sodium bromide in aqueous acetic acid oxidised a vari-
ety of hydrocarbon polymers in 70 bar of air at 150–180 ◦C, yielding
predominantly dicarboxylic acids. Partenheimer has also shown
that this catalyst system can oxidise lignin model compounds, and
lignin samples [17,18].
Thus our concept is to produce mono- and diacids with shorter
chain-lengths than those of the fatty acids of the triglyceride feed-
stock, and to subsequently esterify these shorter chain-length acids
to yield the corresponding mono- and diesters. These should now
have a comparatively lower CP than that obtainable via a straight
triglyceride transesterification of the unmodified feedstock. Our
i
∗
Corresponding author. Tel.: +61 2 9351 2581/+61 4 1606 0824;
fax: +61 293513329.
Canola is a trademarked quality description of low erucic acid ((Z)-docos-13-
enoic acid) rapeseed variants. The erucic acid content of Canola oils for food use is
generally less that 1%.
E-mail address: th.maschmeyer@chem.usyd.edu.au (T. Maschmeyer).
http://dx.doi.org/10.1016/j.cattod.2014.02.041
0920-5861/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
2
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
approach has several potential advantages in that air is used as
an oxidant and the catalysts and solvents employed are relatively
non-hazardous and inexpensive.
Accordingly, we report here the batch autoxidation of the
triglyceride BD feedstocks tallow, canola oil and soy bean oil with
air in the presence of metal/bromide catalysts, and an assessment
of the modified cold flow properties of the FAMES derived from this
process.
(ꢀ = 205 nm), using a Phenomenex Luna 5 ␮m C18(2) 100 Å col-
umn (250 × 4.6 mm, 5 ␮m). The mobile phase was a mixture of (A)
methanol, and (B) 2-propanol/hexane (5:4) (HPLC grade), with the
following operating method: (1) an initial mixture consisting of
A:B = 100:0 was ramped to A:B = 50:50 for 15 min; (2) then ramped
to A:B = 0:100 for 5 min and was maintained at A:B = 0:100 for
5 min; (3) the mixture was then allowed to equilibrate to the ini-
tial ratio A:B = 100:0 within 5 min and then run for a further 5 min
before the next analysis was performed. Sample injections were of
20 ␮L, and a total flow rate of 1 mL/min was employed.
2. Experimental
2.1. Materials
2.3. Oxidation reactions
Cobalt (II) acetate, manganese (II) acetate, zirconium (IV)
chloride, potassium bromide, methyl palmitate, methyl stearate,
anhydrous sodium sulfate, sodium chloride, silica gel, n-hexane,
sulfuric acid, sodium hydrogen carbonate, diethyl ether, and glacial
acetic acid (Sigma-Aldrich) were used as received. HPLC grade n-
hexane, 2-propanol and methanol (Labscan) were filtered through
a PTFE filter prior to use. Domestic, food grade tallow (Allowrie
beef dripping), soy bean and canola oil (Crisco) were used as pur-
chased. Dimethyl butandioate, dimethyl pentandioate, dimethyl
hexandioate and dimethyl octandioate were obtained from Fluka.
Methyl butanoate, methyl pentanoate, methyl hexanoate, methyl
heptanoate, methyl octanoate, methyl nonanoate, methyl hexa-
decanoate, methyl octadecanoate, dimethyl heptandioate and
dimethyl nonandioate were purchased from Sigma-Aldrich. All
esters were used as received.
All substrates were oxidised using an Anton Parr 600 mL capac-
ity reactor (Method 1). Some experiments with tallow were also
conducted with a modified procedure referred to as Method 2 (Parr
Instruments Co. 250 mL capacity reactor). Where not specified,
reactant amounts were scaled accordingly to the vessel size.
In a typical experiment (Method 1) the catalyst mixture was
prepared by dissolving cobalt (II) acetate (1.11 g, 4.47 mmol), man-
ganese (II) acetate (1.10 g, 4.47 mmol), zirconium (IV) chloride
(33 mg, 0.14 mmol) and potassium bromide (1.06 g, 8.94 mmol) in
glacial acetic acid (150.0 g, 2.50 mol) and distilled water (12.0 g,
0.66 mol). This mixture was transferred into a stainless steel high-
pressure reactor (Parr, 600 mL capacity, Hastalloy C) to which the
triglyceride fat or oil (8.0 g, approximately 9 mmol, composition
dependent) or the methyl hexadecanoate (8.0 g, 29.6 mmol) were
added. The reactor was then sealed, pre-pressurised to 40 bar (or
50 bar, Method 2) with compressed air and then heated to 150 ◦C
while being stirred by means of an integral PTFE blade mixer
rotating at 500 rpm. Once the temperature reached 150 ◦C, the pres-
sure was adjusted to 70 bar with additional air (if required). The
time required to reach 150 ◦C was approximately 30 min. The sys-
tem was maintained at the reaction temperature for 2 h and then
allowed to cool to room temperature (typically around 30 min) and
depressurised.
2.2. Instrumentation
A Shimadzu Gas Chromatograph (GC-17A) equipped with a FID
detector and a polyethylene glycol column (BP21, 30 m length,
0.25 mm ID) was used for the quantification of reaction products.
Initially the column was kept at 80 ◦C for 2 min and heated to
240 ◦C over 8 min and maintained at 240 ◦C for another 10 min. The
ester products were calibrated from prepared standards contain-
ing known amounts of esters with a known amount of mesitylene
as an external standard. Calibration curves were drawn as (concen-
tration of esters)/(concentration of external standard) on the X-axis
and (area of the ester peak)/(area of external standard peak) on the
Y-axis. We have assumed that the FID response per carbon atom
is the same for those FAMES for which standards were unavail-
able commercially, i.e. methyl decanoate, methyl undecanoate,
ethyl dodecanoate, methyl tridecanoate, methyl tetradecanoate,
methyl pentadecanoate, dimethyl decandioate, dimethyl undecan-
dioate, dimethyl dodecandioate, dimethyl tridecandioate, dimethyl
tetradecandioate, dimethyl pentadecandioate and dimethyl hexa-
decandioate.
2.4. Work-up—Methyl hexadecanoate
Refer to Fig. 1. The liquid products typically appeared as a
dark brown mono-phasic solution. Initially, volatile components
were isolated from the product liquid under reduced pressure on
a Schlenk line (Step 1) and analysed by GC/MS. This product iso-
lation was achieved using two cold traps arranged in series: trap
one was at −78 ◦C (dry ice and acetone) and trap two was at
−196 ◦C (liquid nitrogen). The resultant product liquid was vis-
cous with the formation of some precipitates. Acetic acid (20 mL)
was added to achieve a homogenous solution. Subsequently, it was
dried using silica gel (10 g) and anhydrous Na2SO4 (20–30 g) (Step
2). The drying agents were removed by filtration. The residue was
extracted further with diethyl ether and n-hexane (20 mL each).
Esterification (2–6 h) of the combined organic fractions proceeded
smoothly in excess methanol with concentrated sulfuric acid as
catalyst (3 mol%, calculated based on the conservative assumption
that all of the substrate was acetic acid). The esterified mixture
was neutralised by the addition of solid NaHCO3, and the mixture
subsequently filtered (Step 3). To remove any methyl acetate (orig-
inating from the esterification of acetic acid), the filtered mixture
was subjected to reduced pressure on a rotary evaporator (Step 4:
final volume approx. 15 mL). To purify the esters (Step 5), crude
esters were washed with brine (3 × 10 mL), stirred overnight with
anhydrous Na2SO4 to remove water, filtered and the dried product
evacuated for 90 min on a Schlenk line before conducting cloud and
gel point tests.
A Shimadzu GC/MS (QP 2010 using a fused silica Rtx-5Sil MS
column, 30 m length, 0.25 mm ID) was used to identify the reaction
products. Initially the column was kept at 50 ◦C for 3 min and then
heated at a rate of 15 ◦C/min to 300 ◦C and kept at 300 ◦C for 5 min.
Volatile acids were measured with and without silylation. Com-
pounds were identified using the NIST database and by comparison
to the standards prepared.
Differential scanning calorimetry (DSC) was performed using
a TA Instruments modulated differential scanning calorimeter
(MDSC) 2920 equipped with the Liquid Nitrogen Cooling Acces-
sory (LNCA). Samples were measured at a cooling rate of 1 ◦C/min
from 40 to −30 ◦C.
High Performance Liquid Chromatography (HPLC) analyses
were used to determine the lack or presence of triglycerides
and diglycerides. The analyses were carried out with a Shimadzu
Prominence HPLC system with a SPD-M20A diode array detector
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
3
Fig. 1. General purification procedure.
2.5. Work-up—Tallow oxidation products
which was progressively lowered (CPs determined in the ASTM
method are also based on visual observations of the forming wax
crystals) [21]. Due to this constraint, in order to report conservative
values only, at all times the highest temperature at which the sam-
ple appeared to turn cloudy—‘cloud point’ or ‘CP’—was recorded.
Several of the experiments were also analysed by differential scan-
ning calorimetry (DSC) to supplement the visual measurements.
The work-up protocol for tallow oxidation was based on that
used for methyl hexadecanoate oxidation with minor differences.
Prior to the esterification of acids (Step 2), the drop-wise addition of
concentrated sulfuric acid was used to adjust the pH to <2.5 (from
pH 4) and any precipitated solids were removed by filtration. The
precipitate was not analysed as it was assumed to consist of sulfated
autoxidation catalyst. In the purification of crude esters (Step 5), a
saturated NaHCO3 solution (5 mL) wash preceded the brine wash
(10 mL). This was repeated three times.
3. Results and discussion
3.1. Work-up rationale
2.6. Cloud- and gel-point measurements
Minor variations in the work-up procedures were used for
improved manipulation and workability of the samples, although
these are not expected to affect the results. For example, in the
esterification stage (Step 2), to ensure the pH was low enough
for esterification, the pH was reduced beforehand. This produced
a precipitate, which was presumed to be sulfated autoxidation
catalyst. As the focus was to achieve improved BD with low freez-
ing points, these solids were not analysed further. To determine
Since insufficient volumes of products were produced in this
study to determine cloud points (CPs) by a standard method i.e.
ASTM D2500–11 (CP test of the American Society for Testing and
Materials which requires 40 mL) [19,20], the approximate CP was
determined by periodic visual observation of samples contained
in glass vials immersed in a refrigerated bath, the temperature of
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
4
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
were too volatile for BD use. Hence, the samples were evacuated for
90 min on a Schlenk line before conducting any cloud and gel point
tests. Qualitative analysis by GC/MS demonstrated that this proce-
dure was sufficient to completely remove methyl acetate and all,
but a trace amount, of methyl pentanoate (b.p. 126.5 ◦C) from the
samples. The final product was a room temperature-mobile liquid.
3.2. Methyl palmitate oxidation
To verify that saturated fatty acid chains can be oxidised via
Partenheimer’s autoxidation method and to better understand
the reactions involved, we examined the oxidation of the satu-
rated fatty acid ester, methyl hexadecanoate (methyl palmitate),
H3C(CH2)14C(O)OCH3. Due to the wide variety of products and
subsequent unfeasibility of full quantification, the aim was to
provide a qualitative description of the products. A partial GC/MS
is shown in Fig. 2. Autoxidation of methyl hexadecanoate, followed
by esterification of the products with methanol yielded, as main
products (by GC/MS and GC/FID) diesters MeOC(O)(CH2)mC(O)OMe
(m = 5–12). Smaller quantities of monoesters Me(CH2)nC(O)OMe
where n = 7–12 and diesters where m = 4 and 13 were also formed.
Traces of monoesters where n = 4–6 and 13 and diesters where m = 2
and 3 were also observed. Propanoic (the most abundant), butanoic
and pentanoic acids were detected among the most volatile com-
ponents of the post reaction liquids. The GC/MS chromatogram of
Fig. 2 also shows that not all of the methyl hexadecanoate (n = 14)
was autoxidised, but conversion to all oxidation products was 93%
as quantified by GC-FID from the acetic acid solution removed from
the reactor using mesitylene as an external standard.
Fig. 2. Partial GC/MS chromatograms of ester mixture from autoxidation of
methyl hexadecanoate showing monoesters Me(CH2)nC(O)OMe and diesters
MeOC(O)(CH2)mC(O)OMe. Other less abundant products were detected outside the
elution range shown.
the time required for complete esterification, reactions were con-
ducted with various reaction times (2–6 h). GC analyses indicated
the absence of identifiable carboxylic acids, and that virtually com-
plete esterification with methanol occurred within 2 h under the
conditions used. Measurements of samples from one of the reac-
tions by HPLC (UV detection at 205 nm) showed the absence of
triglyceride or diglyceride fragments.
In the purification stage (Step 5), to rule out a contribution
to the cloud/gel point depression by any or all of, e.g. acetic acid
and methanol, the product was washed with saturated aqueous
NaHCO3 solution.
To obtain meaningful measurements of cloud and gel points,
it was necessary to ensure the removal of shorter chain esters that
Similarly to the results reported for polyethylene autoxidation
[16] the products of methyl palmitate oxidation are dominated by
Fig. 3. Partenheimer’s proposed mechanism of diacid formation from polyethylene [16].
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
5
stabilised near the centre of the chain, leading to the observed
product distributions.
*
O
O
3.3. Oxidation of canola, soy oil, and tallow triglycerides
methyl palmitate
The triglycerides of tallow, soy and canola oils represent a
significant increase in substrate complexity over that of methyl
palmitate. In addition to the glycerol backbone, they possess vary-
ing degrees of unsaturation. Table S1 lists the products identified in
the ester mixtures derived from the autoxidation and esterification
of these materials, carried out under the same conditions as used for
methyl hexadecanoate. The most abundant compounds were quan-
tified by GC/FID analysis, however the complexity of the product
mixture meant that complete quantification was not possible.
The yields in Table S1 are shown as mass concentration (g/g) of
substrate and molar concentration (mol/mol) of fatty acid, making
the simplifying assumptions that (3 mol fatty acids = 1 mol triglyce-
rides) and that the triglyceride molecular weight is that of triolein,
i.e. 885 g/mol. The complexity of the oxidation reactions and of the
substrates used precludes any detailed commentary on the reac-
tion mechanisms taking place, therefore this discussion is confined
to general features of the reactions studied, but the reaction prod-
ucts observed are consistent with the mechanistic discussions of
the methyl palmitate reactions (vide supra) and the presence of
additional paths involving the unsaturated moieties.
OH
HO
OH
O
O
O
n = 7
m = 5
*
Fig. 4. Proposed product formation route of saturated fatty acid ester chains.
denotes a possible site for hydrogen abstraction and subsequent autoxidation.
diacids (here, detected as their diesters). The distribution of chain
lengths, however, is significantly different in the present case. The
most abundant products in the distribution of diacids (m = 7–9)
correspond to compounds substantially higher in mass than
those found to dominate in polyethylene autoxidation (m = 2–4),
notwithstanding that the latter oxidation was carried out at higher
pressures and with a different oxygen:substrate ratio.
For simplicity, it is assumed that the fats and oils are composed
entirely of triglycerides with a fatty acid distribution as shown in
Table S2, consistent with accepted literature [2].
Partenheimer suggested that the mechanism of polyethylene
oxidation (Fig. 3) was similar to that involved in the thermal autox-
idation of hexadecane [22]. Such mechanisms involve initial radical
formation either by hydrogen atom abstraction from a methy-
lene or methyne group or carboxyl radical formation [23]. The
tertiary C–H bonds of the methyne groups formed during the poly-
merisation of ethylene [24], are favoured as the initial points of
attack for oxidation mechanisms involving hydrogen abstraction,
because the resulting radicals are more stable than those formed
from methylene groups. Partenheimer noted that although the
C–H bonds at the branching points formed during ethylene poly-
merisation were most favoured for initial attack during oxidation,
the overall high yields of short chain diacids (HOOC(CH2)mCOOH,
where m = 2–4) indicated that the linear backbone was also oxi-
dised in his experiments. A possible mechanism for the oxidation
of the linear backbone was proposed by Partenheimer, involving
a cyclic transition state that originates from a dihydroperoxide
via a ‘backbiting’ process. This involved reversible intramolecular
hydrogen atom abstraction of the initially formed peroxy radical.
Presumably cyclic transition states, where m = 2–4 were energeti-
cally favoured; leading, in the case of polyethylene oxidation, to the
observed distribution of products dominated by dicarboxylic acids
HOOC(CH2)mCOOH. A similar mechanism, also involving a cyclic
transition state, but starting with a carboxyl radical, has been sug-
gested to explain the distribution of products from the photo- and
thermo-degradation of LDPE [23].
The difference in the product distributions between the oxi-
dations of polyethylene and methyl palmitate suggests the
dihydroperoxide ‘backbiting’ mechanism (Fig. 3) implicated as a
possible reaction path in the polyethylene autoxidation is not the
dominant mechanism under the conditions of our experiments. It
seems likely in the absence of methyne C–H groups, the preferred
sites for initial hydrogen abstraction are the methylene groups
toward the centre of the chain, (and that the resulting acid may
undergo further attack in a similar manner) with the overall effect
of producing a series of diacids (Fig. 4). This is consistent with the
observations that the higher abundance acids and diacids observed
are m and n = 6–9 as opposed to chain-end oxidation events (e.g.
m, n = 4 or 12). The rationale is that the formed radical is most
The dominant products as g/g substrate from the
autoxidation–esterification of canola oil are dimethyl nonandioate
(m = 7) at 0.216 g/g and methyl nonanoate (n = 7) at 0.118 g/g.
This suggests that C=C cleavage is a significant reaction pathway.
These compounds are expected derivatives of the products of such
oxidative cleavage from oleic acid (64.40 wt% by composition from
Table S2). Since HPLC measurements were showing the absence
of glycerides, it is further inferred that under the acidic reaction
conditions the ester bonds of the glycerides are completely hydrol-
ysed. Interestingly, only trace amounts of glycerol were detected
among the products across the triglyceride sources, suggesting
that glycerol had also been oxidised.
The autoxidation products for soy oil are mainly a result of
alkene cleavage reactions. In addition to oleic acid (22.80 wt%)
products, linoleic acid (53.70 wt%) yields dimethyl nonandioate
(m = 7), methyl hexanoate (n = 4) and dimethyl malonate (m = 1)
due to the position of the C=C bond (at 9 and 12 positions). As soy
oil mostly consists of linoleic groups, the most abundant product
quantified was methyl hexanoate (n = 4) at 0.205 g/g followed by
dimethyl nonandioate (m = 7) 0.118 g/g which is consistent with
C=C cleavage products.
Alkene cleavage has been reported for metal/bromide autoxida-
tion reactions. Cleavage of 1-octene to heptanoic acid and formic
acid, along with the formation of 5-acetylpentanoic acid, was
reported by Jason and Fields, using a Co/Mn/Br catalyst in acetic
acid at 170 ◦C with an air flow at 28 bar pressure [25]. More com-
monly, however, allylic oxidation or acetoxylation of the carbon
atom(s) in an ␣-position to the double bond has been reported
[26]. The metal/bromide autoxidation of hydrocarbon compounds
is generally regarded as operating by means of a free radical chain
mechanism [26]. Alkene cleavage is also well known in ozonoly-
sis. This is typically achieved by bubbling through a methanolic
solution of the alkene at low temperatures (−78 ◦C) with ozone.
Following a reductive work-up, cleaved products result amongst
others in aldehydes, ketones, alcohols or carboxylic acids [27].
In the case of the autoxidation of tallow, besides alkene cleav-
age products due to the substantial oleic acid composition (42.40%),
products from the autoxidation of unsaturated palmitic (23.30%)
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
Table 2
6
Cloud point (CP) and differential scanning calorimetry (DSC) measurements of indi-
vidual tallow autoxidation experiments..
Experiment
Method
CP (◦C)
Tco (◦C)
1
2
1
1
1
2
2
1
5
3
0
4
3
3
4a
5a
−2
−2.5
−2b
−12.5
Values are recorded to the nearest 0.5 ◦C. Tco refers to the crystallisation onset tem-
perature (◦C)
a
Autoxidation step was conducted for 4 h instead of 2 h.
b
A conservative value, the sample exhibited liquid properties towards −12.5 ◦C.
Fig.
5. Partial
GC/MS
chromatograms
of
ester
mixture
from
case. Mixing samples of the esters obtained after autoxidation in
the appropriate proportions as quantified in Table S1, allowed an
estimate cloud point for autoxidised canola and soy BD (Table 1, col-
umn 2). Both representative samples of soy oil and canola indicated
lowered CPs of −13 ◦C, which again are substantial reductions rela-
tive to the original esters (canola: 0 3 ◦C, soy oil: 1 2 ◦C) [28,29].
Potentially, other triglyceride substrates, not limited to soy and
autoxidation–esterification of tallow showing monoesters Me(CH2)nC(O)OMe
and diesters MeOC(O)(CH2)mC(O)OMe. Other less abundant products were
detected outside the elution range shown.
and stearic (19.40%) acids are expected. The GC/MS (Fig. 5) clearly
shows peaks for dimethyl nonandioate (m = 7) and the shorter
diesters (m = 5 and 6) similar to the product distribution of the
autoxidation reaction of soy and canola oil. Additionally consid-
erable amounts of the monoesters of palmitic (n = 14) and stearic
(n = 16) acid can be detected, suggesting that the autoxidation
reaction of saturated fatty acids is slower than that observed for
unsaturated fatty acids. The remaining contributions stem from
the oxidation of the saturated fatty acid chains (e.g. m = 14) in a
manner that resembles the autoxidation of methyl hexadecanoate.
For example, dimethyl hexadecandioate is likely to have originated
from the oxidation of stearic acid.
3.4. Low temperature properties
The low temperature properties of the autoxidised tallow, soy
and canola BD and a comparison with regular tallow BD are shown
in Table 1.
The cloud point of the tallow sample prepared from triglyceride
substrates by transesterification using methanol and potassium
hydroxide as catalyst is consistent with typical values reported in
the literature [10]. A promising 1 ◦C CP mean value was achieved
through the autoxidation method. However, some variation in the
CP and yields of tallow autoxidation–esterification products was
observed. In some instances this was attributable to deliberate vari-
ations in the autoxidation conditions (e.g. method 1 or 2). In other
instances the differences are believed to originate from the nature
of the reaction: as the autoxidation is radical mediated, there is
inherent variability in the product distribution. Table 2 indicates
the CP values of each individual tallow autoxidation experiment.
For those experiments conducted using method 1, the spread of
CPs is large (5 to 1 ◦C), although this still represents a very signif-
icant improvement over unmodified tallow BD (cf. 14 ◦C) in each
Table 1
Summary of cloud point (CP) and differential scanning calorimetry (DSC)
measurements.
Cloud pointa (◦C)
Tco (◦C)
Autoxidised soy BD
Autoxidised canola BD
Tallow BD
−13.0b
−13.0b
14.0
n/a
n/a
12.5
−1.5
Autoxidised tallow BD
1.0
Values are recorded to the nearest 0.5 ◦C. Tco refers to the crystallisation onset tem-
perature (◦C).
a
Values of CP are rounded to the nearest 0.5 ◦C.
Estimated from a simulated mixture of esters according to Table S1.
Fig. 6. Cooling DSC curves for (a) tallow BD (Tco = 14 ◦C) and (b)
autoxidation–esterification tallow BD (Tco = 5 ◦C).
b
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041

G Model
CATTOD-8934; No. of Pages7
ARTICLE IN PRESS
P. Phung et al. / Catalysis Today xxx (2014) xxx–xxx
7
canola oil, could be used to produce BD with improved CPs over
their unmodified esters.
The autoxidised tallow, canola and soy bean oil BDs all
have reduced CPs or Tco relative to BD from the unmodified
oils as determined by visual observation and DSC measure-
ments. The CP values corresponded reasonably well with those of
3.5. Differential scanning calorimetry (DSC)
Tco
.
Even though the basis of the ASTM CP determination is also
visual determination, differential scanning calorimetry (DSC) was
conducted to augment and add further consistency to our obser-
vations. Cooling DSC curves for tallow BD and autoxidation tallow
BD are shown in Fig. 6. The curves were recorded at a cooling rate
of 1 ◦C/min. The crystallisation onset temperature Tco (Table 2) was
recorded to the nearest 0.5 ◦C determined by the intersection of
a tangent to the baseline and a tangent to the slope at the high
temperature end of the endothermic peak. Under the conditions
employed in this study, as a general rule, Tco corresponds reason-
ably well to the approximate CP (Table 2). As a matter of interest,
two experiments were autoxidised for twice the duration (4 h vs.
2 h) and produced lower CP and Tco (experiments 4 and 5 in Table 2).
Further optimisation of reaction parameters (temperature, pres-
sure, reaction time, etc.) is required but in principle, potentially
other non-food and cost-effective (e.g. trap greases) feedstocks with
a large saturated fat composition (and thus, inherently high CP),
may be autoxidised in this manner. Depending on local fuel spec-
ifications, these autoxidised products may find use as an outright
fuel or in blends to meet specific cold temperature property tar-
gets.
Acknowledgements
We acknowledge support from the Australian Research Coun-
cil. PP acknowledges receipt of a Henry Bertie and Florence Mabel
Gritton Research Scholarship. PB thanks the DFG for a research
fellowship.
3.6. Viability and practicality
The ester mixture ultimately obtained from the tallow
autoxidation–esterification is potentially useful as a fuel in its own
right, however, by virtue of its composition it might not meet cur-
rent BD standards in Australia or elsewhere. The main reasons
for this would be that the shorter chain esters and possibly the
shorter chain diesters would fall outside the flash point criteria
specified for BD, so that the overall flash point of the BD is likely
to be lower than permitted by the standards (minimum 120 ◦C)
[30]. For this reason a likely application of the ester mix would be
blending with petroleum diesel to give mixtures with lower cloud
point than would result from blending with unmodified tallow BD.
Alternatively, the ester mixtures could be blended with, e.g. unmo-
dified tallow BD to produce a BD that meets relevant BD standards,
but possesses a lower cloud point. In these scenarios it may prove
cost-effective to treat the acids produced during the oxidation by
esterification with ethanol instead of methanol. Methanol is usu-
ally the alcohol of choice for BD production, because of its lower
cost, but additional cloud point depression is likely to be obtained
for ethyl esters [28,31]. The use of ethanol may also lower the
volatility of the ester mixture with a subsequent increase in flash
point. However, these are commercial considerations directed by
the regulatory framework of the time.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.cattod.2014.
02.041.
References
[1] B. Freedman, R.O. Butterfield, E.H. Pryde, J. Am. Oil Chem. Soc. 63 (1986) 1375.
[2] F.R. Ma, M.A. Hanna, Bioresour. Technol. 70 (1999) 1.
[3] L.C. Meher, D.V. Sagar, S.N. Naik, Renewable Sustainable Energy Rev. 10 (2006)
248.
[4] B.L. Salvi, N.L. Panwar, Renewable Sustainable Energy Rev. 16 (2012)
3680.
[5] A.E. Atabani, A.S. Silitonga, I.A. Badruddin, T.M.I. Mahlia, H.H. Masjuki, S.
Mekhilef, Renewable Sustainable Energy Rev. 16 (2012) 2070.
[6] S.N. Naik, V.V. Goud, P.K. Rout, A.K. Dalai, Renewable Sustainable Energy Rev.
14 (2010) 578.
[7] M. Ghazavi, M. Fallahipanah, H. Jeshvaghani, Int. J. Recycl. Org. Waste Agric. 2
(2013).
[8] A. Ajanovic, Energy 36 (2011) 2070.
[9] M. Gürü, B.D. Artukog˘lu, A. Keskin, A. Koca, Energy Convers. Manage. 50 (2009)
498.
[10] H. Imahara, E. Minami, S. Saka, Fuel 85 (2006) 1666.
[11] M. Mittelbach, C. Remschmidt, Biodiesel: The Comprehensive Handbook, Mit-
telbach, Graz, 2005.
[12] L.D. Bergelson, M.M. Shemyakin, Angew. Chem. Int. Ed. 3 (1964) 250.
[13] M. Ledoux, P. Juaneda, J.L. Sebedio, Eur. J. Lipid Sci. Technol. 109 (2007) 891.
[14] E.G. Giakoumis, Renewable Energy 50 (2013) 858.
[15] S.K. Hoekman, A. Broch, C. Robbins, E. Ceniceros, M. Natarajan, Renewable
Sustainable Energy Rev. 16 (2012) 143.
4. Conclusions
Homogeneous Co/Mn/Zr/bromide catalysed autoxidations in
aqueous acetic acid at ca. 150 ◦C and 70 bar of air were con-
ducted in a batch reactor using methyl palmitate, low grade
tallow, canola oil and soy bean oil as substrates. It was found
that under these reaction conditions the glyceride ester linkages
will get hydrolysed and an oxidative cleavage of the fatty acid
chains occurs. Preferably, the oxidative cleavage takes place at
the double bonds of unsaturated fatty acids, but cleavage of sat-
urated fatty acid chains can be also detected, as demonstrated
for methyl palmitate as “model” substrate. After esterification
of the autoxidised reaction mixture, the products formed were
mainly (methyl-) diesters and monoesters with varying chain
lengths: MeOC(O)(CH2)mC(O)OMe where m = 5–12 and monoesters
Me(CH2)nC(O)OMe) where n = 7–12 as noted by GC/FID. GC/MS and
GC/FID analyses of methyl palmitate indicated that for saturated
fatty acids the product distribution is best rationalised via a radical
mechanism, involving initial hydrogen abstraction form the centre
of the chains.
[16] W. Partenheimer, Catal. Today 81 (2003) 117.
[17] W. Partenheimer, Adv. Synth. Catal. 346 (2004) 1495.
[18] W. Partenheimer, Adv. Synth. Catal. 351 (2009) 456.
[19] Y. Sasaki, T. Endo, N. Tanaka, H. Inoue, Green Chem. 11 (2009) 27.
[20] ASTM, Standard test method for cloud point of petroleum products ASTM
Standard D2500-11, D 2500, ASTM, 2011.
[21] A. Hammami, J. Ratulowski, J.A.P. Coutinho, Pet. Sci. Technol. 21 (2003) 345.
[22] R.K. Jensen, S. Korcek, M. Zinbo, J. Am. Chem. Soc. 114 (1992)
7742.
[23] S. Karlsson, M. Hakkarainen, A.C. Albertsson, Macromolecules 30 (1997) 7721.
[24] E.F. McCord, W.H. Shaw, R.A. Hutchinson, Macromolecules 30 (1997) 246.
[25] E.F. Jason, E.K. Fields, US Patent No. 3,076,842, Standard Oil Co., 1963, 3,076,
842.
[26] W. Partenheimer, Catal. Today 23 (1995) 69.
[27] M.B. Rubin, Helv. Chim. Acta 86 (2003) 930.
[28] X. Lang, A.K. Dalai, N.N. Bakhshi, M.J. Reaney, P.B. Hertz, Bioresour. Technol. 80
(2001) 53.
[29] B.R. Moser, Energy Fuels 22 (2008) 4301.
[30] P.R. Garrett, Fuel standard (biodiesel) amendment determination 2009 (No. 1),
in: E. Department of Sustainability, Water, Population and Communities (Ed.),
F2009L00693, 2009.
[31] H. Joshi, B.R. Moser, J. Toler, T. Walker, Biomass Bioenergy 34 (2010) 14.
Please cite this article in press as: P. Phung, et al., Metal/bromide autoxidation of triglycerides for the preparation of FAMES to improve
the cold-flow characteristics of biodiesel, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.02.041