Kinetics and pathways for an algal phospholipid (1,2-dioleoyl-sn-glycero-3- phosphocholine) in high-temperature (175-350 °c) water

Article abstract of DOI:10.1039/c2gc35639b

We examined the behavior of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) in high-temperature water at 175, 200, 225, and 350 °C. DOPC hydrolyzed to give oleic acid and a number of phosphorus-containing products. The hydrolysis was catalyzed by oleic and phosphoric acids, which were also reaction products. DOPC formed 1-acyl and 2-acyl lyso-phosphatidylcholine (LPC) along with oleic acid as primary products. LPC subsequently formed other phosphorus-containing intermediates, which finally led to phosphoric acid as the ultimate P-containing product. At 350 °C, phosphoric acid and oleic acid were the only products observed. We observed an ester of oleic acid and glycerol (9-octadecenoic-2,3- dihydroxypropyl ester), which likely formed via the hydrolysis of LPC. A reaction network is proposed to explain the formation of the observed products. A quantitative kinetics model based on the proposed pathways was consistent with the experimental data.

Full text of DOI:10.1039/c2gc35639b

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Green Chemistry
Cite this: Green Chem., 2012, 14, 2856
www.rsc.org/greenchem
PAPER
Kinetics and pathways for an algal phospholipid (1,2-dioleoyl-sn-glycero-
3-phosphocholine) in high-temperature (175–350 °C) water
Shujauddin Changi,a Adam J. Matzgerb and Phillip E. Savage*a
Received 26th April 2012, Accepted 10th July 2012
DOI: 10.1039/c2gc35639b
We examined the behavior of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) in high-temperature
water at 175, 200, 225, and 350 °C. DOPC hydrolyzed to give oleic acid and a number of phosphorus-
containing products. The hydrolysis was catalyzed by oleic and phosphoric acids, which were also
reaction products. DOPC formed 1-acyl and 2-acyl lyso-phosphatidylcholine (LPC) along with oleic acid
as primary products. LPC subsequently formed other phosphorus-containing intermediates, which finally
led to phosphoric acid as the ultimate P-containing product. At 350 °C, phosphoric acid and oleic acid
were the only products observed. We observed an ester of oleic acid and glycerol (9-octadecenoic-
2,3-dihydroxypropyl ester), which likely formed via the hydrolysis of LPC. A reaction network is
proposed to explain the formation of the observed products. A quantitative kinetics model based on
the proposed pathways was consistent with the experimental data.
these conditions. Hydrothermal processing is well suited for wet
1
Introduction
biomass because it obviates the need to dry the feedstock, an
energy-intensive pretreatment step that lowers the overall
efficiency of the process. Of course, energy inputs are also
required for hydrothermal processing, but the thermal energy can
be more easily recovered via heat integration because it will be
available at a higher temperature. Moreover, hydrothermal pro-
cessing has the added advantage of requiring a thermal energy
input that is lower than the heat of vaporization of water, as
required in drying.
Research on fuels from biomass is actively being conducted to
help meet the increasing energy demand of the world’s popu-
lation. Of the different types of biomass sources, algae are prom-
ising because of their high photosynthetic efficiency (∼20 times
that of terrestrial plants),1 their ability to grow in brackish and
sea water, and their lipid content per area of land cultivated
exceeding that of oil seed crops such as soy, sunflower, or palm.2
Of course, there are also challenges associated with algae culti-
vation for biofuel production, and these include water supply
and management and sustainable use of nutrients.
Hydrothermal processing of algal biomass has been widely
reported in the past. It can be used to convert wet algal biomass
to crude bio-oils,3 carbonized solids,4,5 and fuel gases.6 Hydro-
thermal carbonization (HTC) provides a route to algal biodiesel
that avoids algae drying and the use of organic solvents for lipid
extraction. Hydrothermal liquefaction (HTL) of microalgae is a
potentially promising route to obtain high-energy-density, fungi-
ble advanced biofuels. Both processes treat algal biomass in an
aqueous medium at elevated temperatures (175–350 °C) and
pressures (0.9–17 MPa). Water has distinctly different properties
at high temperatures and pressures than at ambient conditions.
For example, water at high temperature and high pressure has a
lower dielectric constant, higher dissociation constant, and
increased solubility of organic compounds.7 Thus, biomacro-
molecules in algae are more susceptible to hydrolytic attack under
Algae biomass is a complex mixture of biomacromolecules
such as polysaccharides, lipids, proteins, and chlorophyll. The
study of model compounds containing the same functional
groups as the corresponding biomacromolecules can provide
useful insight and understanding into the reaction products, path-
ways, kinetics, and mechanisms of hydrothermal treatment of
actual biomass. We recently reported on the hydrothermal reac-
tions of ethyl oleate,8,9 phytol,10 and phenylalanine11 as model
compounds representing ester linkages in triglycerides, a chloro-
phyll moiety, and amino acids, respectively. In this article, we
use 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) as a
model compound to elucidate the reactions of algal phospho-
lipids during hydrothermal carbonization and liquefaction.
Phospholipids are an important class of lipids and can consti-
tute 10–50 wt% of the total lipids in algae.12 They contain a gly-
cerol backbone with a diglyceride, a phosphate group, and an
organic group. Since phospholipids contain a hydrophobic
portion comprised of non-polar n-alkyl chains and a hydrophilic
portion that contains heteroatoms such as phosphorus and nitro-
gen, the products obtained from their reaction in high-tempera-
ture water (HTW) can partition not only into the organic phase
(e.g., hydrochar, crude bio-oil) but also the aqueous phase
during HTC or HTL of algae. A step following HTC or HTL of
aChemical Engineering Department, University of Michigan, Ann Arbor,
MI 48109, USA. E-mail: psavage@umich.edu; Fax: +734-763-0459;
Tel: +734-764-3386
bChemistry Department, University of Michigan, Ann Arbor, MI 48109,
USA
2856 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
hydrolysis of GPC to form glycerophosphoric acid (GPA) as the
end product. GPA did not hydrolyze further under their con-
ditions. The initial hydrolysis of PC gave an isomeric mixture of
1-acyl LPC and 2-acyl LPC. However, no rate constants were
reported for the individual reaction steps of their network.
Fig. 1 Structure of DOPC.
Koning and McMullan studied the hydrolysis of DOPC using
a 2 N HCl solution at 120 °C.21 They too proposed a reaction
network similar to Kensil and Dennis,16 according to which the
first steps involve removal of two fatty acids from the PC to
form GPC. Additionally, they also suggest that GPC is further
hydrolyzed via a cyclic ortho triester to form choline and cyclic
glycerophosphate. Cyclic glycerophosphate was eventually con-
verted to an equilibrium mixture of 85% α-glycerophosphate and
15% of β-glycerophosphate in the acidic medium. The authors
algae is recycling the aqueous phase to grow more biomass.
Ideally, all of the phosphorus in the biomass would partition into
the aqueous phase after reaction to facilitate it’s recycling and
reuse as a nutrient. Clearly, understanding the kinetics and path-
ways for formation of the different P-containing compounds
from phospholipids would inform the design and optimization of
HTC and HTL processes. With this motivation, we studied the
reactions of a phospholipid in HTW.
obtained 80
5% of the initial phosphorus in the combined
Bailey and Northcote found that phosphatidylcholine (PC)
was the major phospholipid present in the green alga Hydrodict-
yon africanum (around 40% of the cell weight).13 Moreover,
oleic acid is known to be an abundant fatty acid in green
algae.3,14 We chose DOPC as a model phospholipid because it
contains oleate moieties and a choline group, each attached to
the glycerol backbone (see Fig. 1).
To the best of our knowledge, phospholipid behavior in HTW
has not been reported previously. However, research has been
conducted at much lower temperatures for phospholipids in lipo-
somes and bilayers. Zuidam and Crommelin studied the kinetics
of hydrolysis of phospholipid in liposomes comprising egg phos-
phatidylcholine (EPC), dimyristoyl phosphatidylcholine, distear-
oyl phosphatidylcholine, and dipalmitoyl phosphatidylcholine/
cholesterol (DPPC/CHOL) at a pH of 4.0 and different tempera-
tures (maximum 80 °C).15 They modeled the disappearance of
the phospholipid with pseudo-first-order kinetics. They have also
shown that changes in fatty acid chains and size do not influence
the hydrolysis rate constants significantly.
Kensil and Dennis studied the alkaline hydrolysis (pH 12.7)
of EPC and DPPC in liposomes and in mixed micelles between
15–60 °C.16 Free fatty acids and sn-glycerol-3-phosphorylcho-
line (GPC) were the main products. These authors proposed that
GPC was formed though a lyso-phosphatidylcholine (LPC) inter-
mediate. The first step of their proposed scheme involves the
attack of a hydroxide ion at either the syn-1 or syn-2 positions of
the PC to produce 2-acyl LPC and 1-acyl LPC isomers, respect-
ively. The authors found equal hydrolysis rates at both positions
over the first half-life of the hydrolysis process. The next step of
the reaction scheme involves the hydrolysis of either of the LPCs
(1-acyl and 2-acyl) to remove a second fatty acid moiety and to
form GPC. The authors reported the kinetics to be first order for
both the phosphatidylcholine and the hydroxide ions. They did
not observe further hydrolysis of GPC and mentioned that this
reaction was slow under their conditions. Hanahan have also
mentioned in their review that further hydrolysis of GPC is quite
slow compared to acyl hydrolysis.17
α- and β-glycerophosphates. Maruo and Benson have shown that
the rate of hydrolysis of cyclic glycerophosphate to α- and β-gly-
cerophosphate is rapid in 0.1 N HCl and 24 °C, with 50% con-
version being achieved in less than 10 min under these
conditions.22 Koning and McMullan also quantified small
amounts of phosphorylcholine (15 5% of total phosphorus).21
They proposed that phosphorylcholine could form via the
hydrolysis of the phosphate group in GPC, a reaction parallel to
the formation of cyclic glycerophosphate from GPC. Lastly
5
3% of inorganic phosphorus compound was quantified and
hypothesized to originate via the hydrolysis of glycerophosphate
and phosphorylcholine. However, no kinetics was reported for
the individual steps of the reaction network.
It should be emphasized that although the above studies are
done using liposomes and at low temperatures, they give useful
information about the products that are formed during the
hydrolysis of phosphatidylcholine. However, the reaction chem-
istry and mechanism of phosphatidylcholine is not well under-
stood at higher temperatures (greater than 120 °C). A complete
carbon or phosphorus balance remains elusive in the literature.
Moreover, all of the above studies report an overall first-order
rate constant for disappearance of PC. No one has yet explored
the reaction at higher concentrations of PC, wherein some other
kinetic phenomenon (e.g., autocatalysis) may become promi-
nent.8 Although, researchers have proposed different reaction
networks for hydrolysis of phospholipids, none of the reports
mention the rate constants for all of the individual steps of the
network. This article fills these gaps in the literature by reporting
on DOPC reactions from 175–350 °C in HTW and providing a
detailed product distribution and kinetics for the complete
DOPC hydrothermal reaction network. These results have impli-
cations for understanding the behavior of phospholipids during
the HTC and HTL of algae.
2
Materials and experimental methods
Grit and coworkers studied the hydrolysis of different phos-
phatidylcholines (PC) in aqueous liposomal dispersions under
acidic and alkaline conditions and reported pseudo-first order
kinetics.18–20 They demonstrated that the rate of hydrolysis was
acid- and base-catalyzed with a minimum rate occurring at a pH
of 6.5 between 40–70 °C. They proposed a reaction scheme
similar to Kensil and Dennis,16 except that they observed further
All experiments were carried out in batch reactors assembled from
3/8 – inch 316 stainless steel Swagelok tube fittings (one port con-
nector and two caps). Each reactor had an internal volume of
approximately 1.5 mL. Prior to initial use of these reactors, they
were conditioned by filling them with water and heating in a
fluidized sand bath at 300 °C for 60 min to remove any residual
materials and expose the reactor walls to the HTWenvironment.
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2857

View Article Online
DOPC and LPC were purchased in high purity (>99%) from
Avanti Polar Lipids Inc. Ethyl oleate, oleic acid, and a high
purity TraceCERT® phosphorus standard of 1000 4 mg L−1
phosphorus in water was purchased from Sigma Aldrich. Metha-
nol and acetic acid (HPLC grade) were purchased from Fisher
Scientific in high purity (≥99%) and used as received. Water was
distilled prior to use.
Reactors (capacity 1.5 mL) were loaded at room temperature
with approximately 75 mg of DOPC. The exact weight of DOPC
added to the reactor was calculated as the difference in the
weight of the reactor before and after loading. The amount of
water loaded in the reactors was such that the aqueous fluid
phase occupied 95% of the reactor volume at the reaction con-
ditions, based on the density of pure water at the same conditions
(i.e., 1.27, 1.23, 1.18, and 0.92 mL of water was loaded at 175,
200, 225, and 350 °C). The pressure at reaction conditions can
be estimated from the saturated steam tables as 0.9, 1.6, 2.5, and
17.0 MPa at 175, 200, 225, and 350 °C, respectively. All con-
centrations were based on the fluid phase volume being 95% of
(referred to as OPE here onwards). Since, a pure standard of
OPE was not available, we generated a calibration curve with
different amounts of another similar ester (ethyl oleate) to quan-
tify OPE. The glycerol peak was not well resolved, especially at
high temperatures and long batch holding times, and therefore
not quantified. To determine whether diglycerides were among
the reaction products, we used a high temperature ASTM 6584
column installed in an Agilent GC-FID 7890 equipment and the
procedure outlined previously by Levine et al.4
We identified and quantified a number of the phosphorus-con-
taining compounds generated during the reaction of DOPC by
using 31P-NMR at 25 °C on a Varian MR-400 NMR spec-
trometer at 161.7 MHz. The inverse gated decoupling technique
was used to suppress the Nuclear Overhauser Effect (NOE). An
observation frequency of 1619.2 Hz was used for acquisition.
A pulse angle of 90° and a relaxation delay of 10 s were utilized.
Typically, we obtained 128 scans before Fourier transformation
was carried out. Agilent’s VnmrJ 3 software qEstimate tool
enabled us to estimate the absolute concentration (in terms of
phosphorus equivalent) for any sample. This capability is based
on the linearity of the digital receiver and does not require the
use of reference signals or the addition of reference standard
compounds to the NMR sample. A one-point external calibration
was performed for the probe before using the quantification tool.
This calibration was set up using a phosphorus standard of
1000 mg L−1 or 32.26 mM equivalent of phosphorus (obtained
from Sigma Aldrich) and the methodology outlined for calibra-
ting the probe.23 We tested the validity of this calibration by
running phosphorus standards of known concentrations, which
gave concentrations within 5% of the actual concentration.
Having established confidence in the external calibration, we
quantified DOPC and other phosphorus-containing compounds
using this technique. Next, we identified the products corres-
ponding to different peaks in the NMR spectrum by matching
their chemical shifts with those of pure compounds that were
expected as products from the hydrolysis of DOPC. To confirm
these identities, we added small amounts of these pure com-
pounds to a sample obtained from reaction at 175 °C and 150 min.
No new peaks appeared in the NMR spectrum. Rather, the areas of
existing peaks increased, thereby confirming the product identities.
Product molar yields were calculated by dividing the number
of moles of the product formed by the number of moles of
DOPC initially loaded into the reactor. Experiments at 175 °C
and 30 min, 200 °C and 45 min, and 225 °C and 45 min were
carried out in triplicate to assess reproducibility.
the reactor volume. The initial concentration of DOPC (CDOPC, ini
)
in all experiments was about 0.07 mol L−1. The loaded reactors
were sealed and placed in a preheated, fluidized sand bath
(Techne SBL-2D) and maintained at the desired temperature
using a temperature controller (Techne TC-8D). The sand bath
was isothermal within 1 °C. The reactor heat-up time has been
measured previously to be 2–3 min,8 which is short relative to
the typical batch holding times used in this study. Upon reaching
the desired batch holding time, the reactors were removed from
the sand bath and quenched by immersing them in a cold-water
bath. The reactors were opened and the contents were recovered
using methanol.
A control experiment at room temperature was used to deter-
mine the recovery of DOPC. A reactor was loaded with 20 mg
of DOPC and 1.23 mL of water. The sealed reactor was kept at
room temperature for 30 min. The post-reaction work-up pro-
cedure used for the experiments at the elevated temperatures was
then performed. This experiment gave an average DOPC recov-
ery of 98%, thereby verifying the suitability of the methods used
for quantifying the amount of DOPC.
We analyzed oleic acid (hydrolysis product of DOPC) with
HPLC and a UV detector using the method outlined previously.8
We used an Agilent Technologies model 6890N GC equipped
with an autosampler, autoinjector, and mass spectrometric detec-
tor to identify some of the products from the HTW treatment of
DOPC. A Wiley mass spectral library was used for compound
identification by matching the mass spectra of observed chroma-
togram peaks with those in the library. An Agilent Technologies
model 6890 GC equipped with a FID was used to quantify the
products. The reaction products were separated on a HP-5MS
fused silica, non-polar capillary column (50 m length × 0.20 mm
inner diameter × 0.33 μm film thickness). We used an inlet temp-
erature of 300 °C, a split ratio of 5 : 1, and an injection volume
of 1.0 μL. The temperature program involved an initial oven
temperature of 50 °C followed by heating to 300 °C at a rate of
25 °C min−1 (isothermal for 10 min), giving a total run time of
20 min. Helium served as the carrier gas (1 mL min−1). As will
be discussed in the next section, the major products that we
identified using the GC were glycerol and 9-octadecenoic-2,3-
dihydroxypropyl ester, an ester of oleic acid and glycerol
3
Results and discussion
In this section we report the product distribution, elucidate the
reaction pathways, and develop a corresponding quantitative
-phenomenological kinetics model that is consistent with the
experimental data.
3.1 Reaction products
Fig. 2 shows the NMR spectrum of a sample from reaction of
DOPC at 200 °C and 60 min along with the identities of several
products. We have positive identification for peaks 1–4, but we
2858 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 2 31P NMR spectrum of products from DOPC reaction at 200 °C and 60 min.
were not able to identify with certainty the compounds corre-
sponding to peaks 5–7 due to lack of authentic compounds as
standards. As will be discussed later in this section, using the
yields of these products and knowledge of likely pathways from
the literature, we tentatively identified peaks 5, 6, and 7.
obtained, indicating that this fatty acid was another end product
of the hydrolysis of DOPC.
An ester of oleic acid and glycerol was another non-
phosphorus-containing product that was quantified. The yield of
the 9-octadecenoic-2,3-dihydroxypropyl ester (OPE) increased
steadily with time at 175 and 200 °C. At 225 °C, however,
the yield of OPE first increased from 0.04 at 15 min to 0.13
at 45 min and then decreased to 0.03 at 90 min, indicating
secondary decomposition of OPE. No OPE was detected at
350 °C.
Table 1 presents the product molar yields and DOPC conver-
sion at different times and temperatures. The mean carbon
balance was 90
15%. We expect the carbon balance to be
below 100% since we have not quantified some of the carbon-
containing products, such as glycerol and choline. The mean
phosphorus balance was 95
phosphorus-containing compounds were identified and
quantified.
15%, indicating that all of the
Of the three tentatively identified NMR peaks (peak 5, 6, and
7 in Fig. 2), the molar yield of peak 5 is the highest at all con-
ditions. The yields of the products corresponding to peaks 5 and
6 responded to time and temperature very much like the yield of
OPE. These yields increased with an increase in the batch
holding time at 175 and 200 °C, but at 225 °C they reached a
maximum and then decreased to zero. Furthermore, the yield of
peak 5 at the shortest time at any temperature is similar to the
yield of OPE formed under similar conditions. This observation
of peak 5 being formed nearly in a molar equivalence to OPE
and the formation of OPE requiring the co-formation of a P-con-
taining co-product, suggests that peak 5 could possibly be this
co-product. We speculate that peak 5 corresponds to phosphoryl-
choline, which is the P-containing compound that would be
formed along with OPE, as will be described in the next section.
Phosphorylcholine has been observed previously during the
acidic hydrolysis of DOPC.21 The yield of peak 5 does not
remain equal to that of OPE at higher temperatures or longer
batch holding times. Rather, its yield exceeds that of OPE at the
more severe conditions, even before the yield of OPE begins to
decrease. This behavior indicates that if peak 5 is indeed phos-
phorylcholine, then as the batch holding time increases at 175
and 200 °C, there is also another pathway forming phosphoryl-
choline. As will be shown in the next section, phosphorylcholine
can indeed be formed through a parallel reaction involving
hydrolysis of the LPC isomers. Furthermore, at a higher temp-
erature of 225 °C, the yield of peak 5 is lower than OPE
1-Acyl LPC was the major phosphorus-containing product at
short batch holding times. Smaller amounts of 2-acyl LPC also
formed. The yields of 1-acyl LPC at the two lower temperatures
initially increased with increasing batch holding time, reached a
maximum value, and then decreased. This behavior suggests that
1-acyl LPC decomposes to secondary products under the more
severe reaction conditions. The yields of 2-acyl LPC were low
(<0.05) at all conditions, but they followed a trend similar to that
of 1-acyl LPC. That is, the yield of 2-acyl LPC initially
increased and then subsequently decreased at longer batch
holding times and higher temperatures.
Phosphoric acid was not observed at the mildest reaction con-
ditions, but once formed, its yield rapidly increased. It became
the major phosphorus-containing product at higher temperature
or longer batch holding time. Yields of essentially unity were
obtained at the most severe conditions, indicating that phospho-
ric acid is the ultimate phosphorus-containing product from the
hydrolysis of DOPC.
Oleic acid, the fatty acid obtained via the hydrolysis of
DOPC, was always the product formed in highest yield. The
yield of oleic acid increased steadily with time at all tempera-
tures. The molar yields of oleic acid may exceed unity because
one mole of DOPC can produce two moles of oleic acid. At
350 °C, this stoichiometric yield of oleic acid of two was
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2859

View Article Online
indicating that if this peak corresponds to phosphorylcholine,
then the decomposition of phosphorylcholine is faster than that
of OPE. Note that we could not obtain a pure standard of phos-
phorylcholine to confirm its identity.
The other phosphorus products that could appear as peak 6
and 7 could be cyclic glycerophosphates or α- and β-glycero-
phosphates, or glycerophosphoric acid, as has been suggested
previously.21 Once again these compounds were not available in
their pure form, which prohibited their positive identification
using NMR. For the unidentified peak 7, its yield increased and
then decreased at all temperatures.
The conversion of DOPC was below 0.35 at the mildest con-
ditions examined (i.e., the two shortest reaction times at the two
lowest temperatures). Near complete conversion was observed at
225 °C (>60 min) and at all batch holding times at 350 °C.
Moreover, a plot of DOPC conversion against the batch holding
time displays a sigmoidal behavior (Fig. 3), i.e. the conversion
increases slowly at shorter batch holding time and then rapidly
increases. This behavior is associated with an autocatalytic
system and is similar to the behavior observed previously for
hydrolysis of ethyl oleate in HTW8,9 and hydrolysis of triglycer-
ides in sunflower oil.24
In the system under consideration, DOPC hydrolysis can
likely be catalyzed by oleic acid and by phosphoric acid, as
these compounds form during the course of the reaction. To test
for catalysis by these acids, we carried out hydrolysis of DOPC
at 200 °C and 15 min with different amounts of added oleic acid
(equimolar) and phosphoric acid (one and a half times molar
concentration of DOPC). Table 2 shows that both the conversion
of DOPC and the yield of oleic acid increased in the presence of
added acids. Therefore, hydrolysis of DOPC is catalyzed by the
oleic acid and phosphoric acid produced during the reaction.
3.2 Reaction pathways
The previous section showed that hydrothermal treatment of
DOPC produced a number of products. In this section, we eluci-
date more fully the reaction pathways for DOPC that lead to
these products.
On the basis of the product yields discussed in the previous
section, the main primary path for DOPC in HTW is hydrolytic
cleavage of a fatty acid moiety linked to the glycerol backbone
to form either 1- or 2-acyl LPC (see reactions I and II). This path
has been mentioned previously in a number of studies on
hydrolysis of phospholipids in liposomal dispersions at low
temperature.16,20,21 1-Acyl LPC is the major isomer observed in
the present work, which is consistent with the results of Grit and
coworkers.18–20 Ester hydrolysis in HTW most commonly
follows an Aac2 mechanism wherein protonation of the ester is
followed by a nucleophilic attack of water.25 1-Acyl LPC is
expected to be the major product, since it forms rapidly from 2-
acyl LPC via acyl migration and is the more stable of the two
isomeric forms.20,26 Hydrolysis of the ester linkages would
release oleic acid, which could subsequently catalyze the further
hydrolysis of DOPC. Reactions I and II provide a graphical rep-
resentation of the primary paths for hydrolysis of DOPC.
Note that no path is included for hydrolytic cleavage of the
phosphorus-containing moiety. Two possible reactions could
2860 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
occur here depending on which phosphorus–oxygen bond was
cleaved. Cleavage of the choline group would form a 18 : 1
phosphatidic acid (see reaction III), but this product was not
observed. The chemical shift of a pure 18 : 1 phosphatidic acid
standard did not match any peak in the NMR spectrum shown
in Fig. 1. Another possibility is cleavage to form a diglyceride
(see reaction IV). We used high-temperature GC but did not
detect the diglyceride that would necessarily form in such a
path. The absence of the products shown in reactions III and IV
indicates that these paths are negligible under the conditions
of our investigation. Thus, we exclude these paths from further
consideration.
Following the formation of the LPC isomers in paths I and II,
hydrolysis of the second fatty acid can take place to produce
sn-glycero-3-phosphorylcholine (GPC), as has been suggested in
Table 2 Conversion of DOPC and yield of oleic acid with and without
added acid at 200 °C and 15 min
Acid
DOPC conversion
Oleic acid yield
None
Oleic
Phosphoric
0.10
0.50
1.00
0.13
0.43
1.31
Fig. 3 Temporal variation of DOPC conversion.
ðIIIÞ
ðIVÞ
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2861

View Article Online
ðVÞ
ðVIÞ
ðVIIÞ
ðVIIIÞ
previous work at lower temperatures, see reaction V.16 Since this
step also involves hydrolysis of an ester bond, it could also be
catalyzed by the fatty acid produced.
In parallel with the hydrolytic cleavage of the oleate moiety,
LPC can also undergo hydrolysis at the phosphate bond in two
different ways, similar to reactions III and IV. One possibility is
that the 1-acyl LPC, being the major isomer, would form 9-octa-
decenoic-2,3-dihydroxypropyl ester (OPE) (reaction VI), consist-
ent with our observation of this compound as one of the reaction
products. Phosphorylcholine would also be released as a co-
product of this reaction. Alternatively, 1-acyl LPC could liberate
choline and form 18 : 1 lyso-phosphatidic acid (see reaction VII).
However, none of the peaks in the NMR spectra corresponded to
a pure standard of 18 : 1 lyso-phosphatidic acid. Therefore, reac-
tion VII seems unlikely under these conditions and has been
excluded from further consideration.
OPE, once formed in reaction VI, can hydrolyze further to lib-
erate glycerol and oleic acid as shown in reaction (VIII). Once
again, OPE hydrolysis can be catalyzed by oleic acid and phos-
phoric acid.
2862 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
ðIXÞ
ðXÞ
ðXIÞ
ðXIIÞ
ðXIIIÞ
GPC, formed in path V, could react at either of the two P–O
bonds, giving rise to two parallel reactions.17,21 Protonation of
the oxygen between the phosphorus atom and the choline
moiety can occur via a cyclic ortho triester to form cyclic glycer-
ophosphate and choline (reaction IX). In a parallel reaction, pro-
tonation of oxygen between the glycerol carbon and phosphorus
can occur to release glycerol and phosphorylcholine (reaction
X). Koning and McMullan mention that under acidic conditions,
reaction IX is favored over reaction X, i.e. more cyclic glycero-
phosphate is formed than phosphorylcholine.21
Cyclic glycerophosphate can be converted to an equilibrium
mixture of α- and β-glycerophosphate or glycerophosphoric acid
(GPA), depending on the pH, as shown in reaction XI. Usually
the α-GPA is the major product.17,22
Glycerophosphate can further hydrolyze to release phosphoric
acid and glycerol, both of which were observed as products in
this investigation (reaction XII).
Lastly, phosphoric acid could also be formed via the hydro-
lysis of phosphorylcholine, see reaction XIII. This reaction also
liberates a choline molecule.
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2863

View Article Online
Fig. 4 Streamlined reaction network for DOPC in HTW.
dC3
dt
This section has summarized the different reaction pathways
likely to be operative for the reaction of DOPC in HTW.
We offer these pathways on the basis of the literature and
the experimental results reported in the previous section.
Other phospholipids could potentially undergo similar chemistry
in HTW.
¼ k3C2 ꢀ k4C3 ꢀ k8C3C4 þ k9C2C4 ꢀ k13C3C6
þ k14C2C6
ð3Þ
ð4Þ
dC4
dt
¼ k1C1 þ k2C2 þ k4C3 þ k6C1C4 þ k7C2C4
þ k8C3C4 þ k11C1C6 þ k12C2C6 þ k13C3C6
3.3 Kinetics model
dC5
dt
¼ k2C2 þ k3C2 ꢀ k5C5 þ k7C2C4 þ k9C2C4
ꢀ k10C5C4 þ k12C2C6 þ k14C2C6 ꢀ k15C5C6
In this section, we use the reaction pathways just discussed as
the basis for a quantitative kinetics model. Fig. 4, a simplified
reaction network for DOPC in HTW, accounts for the reactions
outlined in the previous section. We have streamlined the reac-
tion network by lumping the 1- and 2-acyl LPC isomers together
and denoting them as LPC. Additionally, we lumped together all
of the phosphorus-containing products of LPC since we did not
quantify all of these products and some of their identities are
tentative.
ð5Þ
ð6Þ
dC6
¼ k5C5 þ k10C5C4 þ k15C5C6
dt
C is the concentration and the subscripts 1–6 represent DOPC,
LPC, OPE, oleic acid, lumped P-containing compounds, and
phosphoric acid, respectively. We used fmincon in MATLAB to
solve simultaneously the set of ordinary differential equations
above and perform parameter estimation to obtain the values of
the rate constants at a given temperature.27 The objective func-
tion for minimization was the summation of squared relative
error (SSRE) between the calculated and experimental product
concentrations at a given reaction temperature. This quantity was
calculated using eqn (7), where j is the number of discrete reac-
tion times at a particular temperature and i is the number of
components.
This streamlined reaction network contains five different
hydrolytic pathways. On the basis of our previous work with
ethyl oleate hydrolysis,8,9 wherein, hydrolysis of ester
linkages was found to be catalyzed by acid, we allow each
of the hydrolytic paths to be catalyzed by both oleic acid and
phosphoric acid. Thus, this network can be represented by a
set of 15 individual reactions as shown on the following
pages. The first group of five reactions in this set is
uncatalyzed, the next set of five reactions is catalyzed by oleic
acid, and the last set of five reactions is catalyzed by phosphoric
acid.
We assume that all reactions are irreversible (since water was
present in large excess) and first-order in each reactant. We do
not include the concentration of water in the rate equations since
it did not change appreciably during the reaction. We obtain
numerical values for the rate constants for the individual reaction
paths by fitting the experimental product concentrations to the
model equations (eqn (1) –(6)), which apply to this reaction
network and the constant-volume batch reactors used in the
experiments.
ꢀ
ꢁ
2
6
X X
Cj;i;model ꢀ Cj;i;exp
SSRE ¼
ð7Þ
ðCj;i;model þ Cj;i;expÞ=2
j
i¼1
The rate constants at each temperature were then fit to the line-
arized form of the Arrhenius equation to get Ai and Ea along
i
with the associated standard errors for each rate constant.
Table 3 gives the Arrhenius parameters for the DOPC reaction
network and the rates for the different reaction steps at 200 °C and
60 min. Reactions 1, 6, and 11 all represent the same pathway
(hydrolysis of DOPC), but steps 6 and 11 are catalyzed. The rate
of the uncatalyzed path is nearly two and a half times less than
that for the oleic acid catalyzed path and almost equal to that for
the phosphoric acid catalyzed path. Reactions 2, 7, and 12 also
represent the same pathway (hydrolysis of LPC), with steps 7 and
12 being catalyzed. For this path, the uncatalyzed step is orders of
magnitude slower than the catalyzed paths. The same is found to
be true when comparing the rates of reactions 4, 8, and 13, which
dC1
¼ ꢀk1C1 ꢀ k6C1C4 ꢀ k11C1C6
ð1Þ
ð2Þ
dt
dC2
dt
¼ k1C1 ꢀ k2C2 ꢀ k3C2 þ k6C1C4 ꢀ k7C2C4
ꢀ k9C2C4 þ k11C1C6 ꢀ k12C2C6 ꢀ k14C2C6
2864 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
describe OPE hydrolysis. Since the rates of uncatalyzed paths 2
and 4 are negligible at conditions of Table 3 and more generally at
all times at 200 °C, we removed these paths from the model and
fit the data again using just thirteen rate constants. Removing
these two parameters did not change the other rate constants.
Fig. 5 shows the parity plot for the experimental and calculated
concentrations of the different chemical species. The dotted dia-
gonal line represents a perfect fit of the model to the experimental
data, and the dotted lines above and below represent 20% error.
Fig. 5a shows that the model generally fits the experimental data
for the major products to within 20% and also predicts the con-
centrations of the major products at 350 °C well. Moreover, the
data are scattered on both sides of the diagonal, indicating the
absence of systematic error. Fig. 5b shows that there is more
scatter for the minor products, but even here the model often fits
the data to within the experimental uncertainty.
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2865

View Article Online
3.4 Implications for hydrothermal processing of algae
distribute themselves into the organic phase, whereas phos-
phoric acid would go into the aqueous phase. Thus, phosphorus
recycling after hydrothermal processing of algae would be facili-
tated by processing at conditions that produce phosphoric
acid as the sole phosphorus-containing product. This study
provides a reaction network and kinetics of different pathways
that can be useful for identifying these conditions. For example,
the model predicts that at an isothermal temperature of 300 °C
a reaction time of at least 8 min is required to convert 99%
of the phosphorous in DOPC into phosphoric acid.
The results from this study of DOPC reactions in HTW have
several implications for hydrothermal liquefaction of algae. The
production of fatty acid from phospholipids being autocatalyzed
has implications for process design and optimization for hydro-
thermal biofuel production process. Autocatalytic reactions
exhibit a maximum rate at some intermediate conversion.
Designing to operate at this maximum rate would minimize the
total reactor volume required.
In an ideal process, all of the phosphorus atoms would reside
in compounds that partition into the aqueous, rather than
organic, phase after hydrothermal processing. This outcome
would provide a phosphorus-free fuel product and a phosphorus-
rich aqueous stream that can be used to grow more algal
biomass. Many of the phosphorus-containing compounds
reported here (α-, β-GPA, phosphorylcholine) would likely
Conclusion
Under the conditions studied, DOPC hydrolyzed to form
oleic acid, several phosphorus-containing compounds, glycerol,
and choline. We also detected an ester of oleic acid and glycerol
(OPE), which had not been documented previously. The conver-
sion for DOPC showed a sigmoidal trend with increasing batch
holding time, indicating catalysis by oleic and phosphoric acid,
which was also confirmed experimentally.
Table 3 Arrhenius parameters for DOPC reaction network
Rate 200 °C,
The primary paths from DOPC form 1-acyl LPC and 2-acyl
LPC, which subsequently hydrolyze via parallel reactions to
form either OPE or α-GPA + β-GPA + phosphorylcholine, via
GPC. Hydrolysis of both DOPC and LPC releases oleic acid.
Phosphoric acid was the sole phosphorus-containing end
product. A phenomenological kinetics model based on these
pathways accurately describes the trends in the data and typically
estimates the species concentrations to within their experimental
errors.
The results from this study of DOPC in HTW have impli-
cations to algae processing. For instance, the existence of auto-
catalysis influences process design decisions, and the reaction
network can be used to identify processing conditions that facili-
tate recycling phosphorus in the aqueous phase for additional
biomass growth.
Ea
60 min
Index no. log A
(kJ mol−1
)
Units of ki
(mol L−1 min−1
)
1
6.4 0.5
77
5
6
min−1
1.84E-4
2.83E-8
1.24E-4
2.64E-10
5.45E-4
4.75E-4
8.33E-4
2.31E-5
3.45E-5
1.92E-4
1.15E-4
1.10E-4
1.37E-7
4.75E-5
1.29E-4
2
3
7.5 0.7 119
min−1
5.9 2.9
72 27
13.7 4.4 192 40
11.9 3.1 122 28
42 16
71
18.3 4.6 182 42
10.8 4.8 111 43
64 10
47 14
16.5 3.9 151 36
14.9 7.2 162 65
min−1
4
min−1
5
6
7
min−1
4.1 1.7
7.8 0.4
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
min−1 mol−1
L
L
L
L
L
L
L
L
L
L
4
8
9
10
11
12
13
14
15
6.2 1.1
4.7 1.5
3.5 0.4
7.6 0.3
37
72
3
3
Fig. 5 Comparison of experimental and calculated concentrations at 175 °C, 200 °C, 225 °C, and 350 °C for (a) DOPC
ric acid and (b) LPC OPE lumped P-compounds.
oleic acid
phospho-
2866 | Green Chem., 2012, 14, 2856–2867
This journal is © The Royal Society of Chemistry 2012

View Article Online
12 E. W. Becker, Microalgae: Biotechnology and Microbiology, Cambridge
University Press, Cambridge, 1994.
Acknowledgements
13 D. S. Bailey and D. H. Northcote, Phospholipid composition of the
plasma membrane of the green alga, Hydrodictyon africanum, Biochem.
J., 1976, 156, 295–300.
14 K. Fahl and G. Kattner, Lipid content and fatty acid composition of algal
communities in sea-ice and water from the Weddell Sea (Antarctica),
Polar Biol., 1993, 13, 405–409.
We gratefully acknowledge Mr Eugenio Alvardo and Mr Chris
Kojiro for training in NMR and assistance in quantifying phos-
phorus-containing compounds. We also acknowledge financial
support from the National Science Foundation (EFRI 0937992).
15 N. J. Zuidam and D. J. A. Crommelin, Chemical hydrolysis of phospholi-
pids, J. Pharm. Sci., 1995, 84(9), 1113–1119.
16 C. R. Kensil and E. A. Dennis, Alkaline hydrolysis of phospholipids in
model membranes and the dependence on their state of aggregation, Bio-
chemistry, 1981, 20, 6079–6085.
17 D. J. Hanahan, Lipide Chemistry, John Wiley & Sons, Inc., New York,
1960.
18 M. Grit, J. H. de Smidt, A. Struijke and D. J. A. Crommelin, Hydrolysis
of phosphatidylcholine in aqueous liposome dispersions, Int. J. Pharm.,
1989, 50, 1–6.
19 M. Grit and D. J. A. Crommelin, The effect of aging on the physical
stability of liposome dispersions, Chem. Phys. Lipids, 1992, 62, 113–122.
20 M. Grit and D. J. A. Crommelin, Chemical stability of liposomes:
implications for their physical stability, Chem. Phys. Lipids, 1993, 64,
3–18.
21 A. J. De Koning and K. B. McMullan, Hydrolysis of phospholipids with
hydrochloric acid, Biochim. Biophys. Acta, Lipids Lipid Metab., 1965,
106, 519–526.
References
1 J. A. V. Costa and M. G. de Morais, The role of biochemical engineering
in the production of biofuels from microalgae, Bioresour. Technol., 2011,
102, 2–9.
2 T. M. Mata, A. A. Martins and N. S. Caetano, Microalgae for biodiesel
production and other applications: A review, Renewable Sustainable
Energy Rev., 2010, 14, 217–232.
3 T. M. Brown, P. Duan and P. E. Savage, Hydrothermal liquefaction and
gasification of Nannochloropsis sp., Energy Fuels, 2010, 24(6), 3639–3646.
4 R. B. Levine, T. Pinnarat and P. E. Savage, Biodiesel production from
wet algal biomass through in situ lipid hydrolysis and supercritical trans-
esterification, Energy Fuels, 2010, 24, 5235–5243.
5 S. M. Heilmann, H. T. Davis, L. R. Jader, P. A. Lefebvre,
M. J. Sadowsky, F. J. Schendel, M. G. von Keitz and K. J. Valentas,
Hydrothermal carbonization of microalgae, Biomass Bioenergy, 2010,
34(6), 875–882.
6 Q. Guan and P. E. Savage, Gasification of Nannochloropsis sp. in super-
critical water, J. Supercrit. Fluids, 2012, 61, 139–145.
7 P. E. Savage, Organic chemical reactions in supercritical water, Chem.
Rev., 1999, 99, 603–621.
22 B. Maruo and A. A. Benson, Cyclic glycerophosphate formation from the
glycerolphosphatides, J. Biol. Chem., 1959, 234(2), 254–256.
23 http://www.umich.edu/~chemnmr/docs.html, Date last accessed 12th
April, 2012.
8 S. Changi, T. Pinnarat and P. E. Savage, Modeling hydrolysis and esterifi-
cation kinetics for biofuel processes, Ind. Eng. Chem. Res., 2011, 50(6),
3206–3211.
24 R. Alenezi, G. A. Leeke, R. C. D. Santos and A. R. Khan, Hydrolysis
kinetics of sunflower oil under subcritical water conditions, Chem. Eng.
Res. Des., 2009, 87, 867–873.
9 S. Changi, T. Pinnarat and P. E. Savage, Mechanistic modeling of
hydrolysis and esterification kinetics for biofuel processes, Ind. Eng.
Chem. Res., 2011, 50(22), 12471–12478.
25 J. N. E. Day and C. K. Ingold, Mechanism and kinetics of carboxylic
ester hydrolysis and carboxyl esterification, Trans. Faraday Soc., 1941,
37, 686–705.
10 S. Changi, T. M. Brown and P. E. Savage, Reaction kinetics and pathways
for phytol in high- temperature water, Chem. Eng. J., 2012, 189, 336–345.
11 S. Changi, M. Zhu and P. E. Savage, Hydrothermal reaction kinetics and
pathways of phenylalanine alone and in binary mixtures, ChemSusChem,
2012, DOI: 10.1002/cssc.201200146.
26 A. Plückthun and E. A. Dennis, Acyl and phosphoryl migration in lyso-
phospholipids: importance in phospholipid synthesis and phospholipase
specificity, Biochemistry, 1982, 21, 1743–1750.
27 http://www.mathworks.com/help/toolbox/optim/ug/fmincon.html.
last accessed, 3rd September 2011.
Date
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 2856–2867 | 2867