Feruloylated Products from Coconut Oil and Shea Butter

J Am Oil Chem Soc

DOI 10.1007/s11746-017-2953-7

ORIGINAL PAPER

Feruloylated Products from Coconut Oil and Shea Butter

David L. Compton¹· John R. Goodell²· Mark A. Berhow³· James A. Kenar³·

Steven C. Cermak⁴· Kervin O. Evans¹

Received: 21 October 2016 / Revised: 12 January 2017 / Accepted: 12 January 2017

Abstract The synthesis of feruloylated coconut oil and

feruloylated shea butter were demonstrated in 0.5-L scale,

shaken, batch reactions. Ethyl ferulate and the vegetable

oil/fat were combined in a 1.0:1.3 mol ratio in the pres-

ence of Candida antartica lipase B immobilized on an

acrylic resin (Novozym 435) at 60 °C. The transesteri-

ﬁcation of ethyl ferulate with coconut oil and shea butter

reached equilibrium conversions, after 22 days, of 63 and

70%, respectively, with the shea butter transesteriﬁcations

producing a white precipitate not observed in the coconut

oil transesteriﬁcations. The faster transesteriﬁcation rates,

equilibrium conversions and white precipitate were shown

to result from di- and monoacylglycerols (DAG and MAG)

present in the shea butter. The transesteriﬁcation of ethyl

ferulate and coconut oil was also tested in a continuous,

enzymatic, packed-bed bioreactor using Novozym 435 at

60 °C to produce feruloylated coconut oil at rates of 0.5–

0.9 kg/day over 4.5 months. The feruloylated coconut oil

acylglycerol species were identiﬁed by LC–MS analysis of

transesteriﬁcation reactions of ethyl ferulate with medium

chain triacylglycerol (TAG) standards, C8–C14. The feru-

loylated vegetable oils possessed an ultra violet (UV)

absorbing λ_max328 nm, making them good UVAII absorb-

ers, as deﬁned by the U.S. Food and Drug Administration.

The feruloylated coconut oil possessed a 17.5% higher

absorption capacity than feruloylated shea butter on a per

weight basis. All the feruloylated vegetable oils possessed

rapid antioxidant capacity (50% reduction of initial radical

concentration <5 min) at the concentrations tested, 0.5–

2.5 mM. Feruloylated coconut oil possessed chemical and

physical characteristics that suggested it would be fungible

for feruloylated soybean oil in current retail formulations.

Mention of trade names or commercial products in this

publication is solely for the purpose of providing speciﬁc

information and does not imply recommendation or endorsement

by the United States Department of Agriculture (USDA). USDA

is an equal opportunity provider and employer.

Keywords Biotechnology · Biocatalysis · Biobased

products · Fats and oils

* David L. Compton

david.compton@ars.usda.gov

1

Renewable Product Technology Research Unit, United States

Department of Agriculture, Agricultural Research Service,

National Center for Agricultural Utilization Research,

1815 N. University St., Peoria, IL 61604, USA

Introduction

2

Coconut oil is a vegetable oil obtained from the seed ker-

nel of the coconut palm consisting of mostly saturated,

short to medium chain (C6–C14) fatty acids. Coconut oil

is widely consumed in tropical populations and has gar-

nered increased global popularity as a functional oil, and

its health, metabolic, and antimicrobial properties have

been recently reviewed [1, 2]. Freshly pressed coconut oil

is a mixture of TAG (~85%), DAG (2–7%), MAG (0–3%),

and FFA (0–1%) [3, 4]. Consisting mostly of lauric acid

iActive Naturals/Biotechnology Research and Development

Corporation, 801 W Main St., Peoria, IL 61606, USA

3

Functional Foods Research Unit, United States Department

of Agriculture, Agricultural Research Service, National

Center for Agricultural Utilization Research, 1815 N.

University St., Peoria, IL 61604, USA

4

Bio-Oils Research Unit, United States Department

of Agriculture, Agricultural Research Service, National

Center for Agricultural Utilization Research, 1815 N.

University St., Peoria, IL 61604, USA

1 3

J Am Oil Chem Soc

(C12:0, 45–53%) and myristic acid (C14:0, 17–23%) with

other saturated, shorter chain fatty acids, reﬁned coconut

oil, sans DAG and MAG, is a white solid at room tempera-

ture (mp 24 °C). The saturated fatty acids limit oxidation

and rancidity making coconut oil an ideal candidate for

lipid-based products. Coconut oil has recently gained much

interest as a natural ingredient in the cosmetic and personal

care industry and is used in products ranging from facial

cleansers to deodorants to insect repellants [2, 5, 6].

enzymatic, packed-bed bioreactor over a 4.5 months pro-

duction run. The advancement from small scale, biocata-

lytic transesteriﬁcations of ethyl ferulate using alterna-

tive vegetable oils to a pilot-scale, continuous, bioreactor

production is crucial for the commercial adoption of these

feruloylated lipids by the personal care and cosmetic

industries.

Shea butter is the fat extracted from the fruit kernel of

the African shea tree that is prevalent throughout the savan-

nahs of West and Central Africa. The ecological diversity of

the expansive geographical growing region of the shea tree

results in exported oils (butter) with a wide range of fatty

acid compositions [7]. Shea butter consists mainly of oleic

acid (C18:1, 37–62%) and stearic acid (C18:0, 25–50%)

with some palmitic acid (C16:0, 2–10%) and other minor

fatty acid components. This diversity results in shea butters

with a wide range of chemicophysical properties and minor

bioactive components (e.g. triterpene alcohols, phenols,

sterols) [8]. Africa exports over 350,000 tons/year of dry

kernel (115,000 tons shea butter/year) to the United States

and Europe for mostly food and specialty fat uses [7, 9].

More recently, the solid (mp 32–45 °C), off-white, malle-

able shea kernel extract has been increasingly used as an

exotic fat and oil ingredient in the cosmetic, health, and

personal care industries [8, 10, 11].

The biocatalytic modiﬁcation of fats and oils to produce

bio-based active ingredients with enhanced functionaliza-

tion for the cosmetic and personal care industries has stead-

ily increased over the past decade [12–15]. Of particular

interest is the modiﬁcation of vegetable oils via biocatalytic

transesteriﬁcation with phenylpropanoid esters (e.g. feru-

lic, caffeic, sinapic) [16–18]. This enzymatic, solventless

process uses the vegetable oil as the reaction medium and

has been studied extensively using soybean oil, evolving

from a laboratory-scale, stirred, batch process into a con-

tinuous, enzymatic, packed-bed bioreactor process [19–21].

There has been limited examination of the transesteriﬁca-

tion using other vegetable oils, all at the laboratory scale

[22, 23]; however, none have examined vegetable oils that

are solid at room temperature and none have demonstrated

their use in a continuous, enzymatic, packed-bed bioreactor

process.

Experimental Procedures

Materials

Coconut oil used for the batch syntheses was purchased

from Better Body Foods and Nutrition (Lindon, UT, USA)

from a local grocery. Coconut oil used for the continu-

ous, enzymatic, packed-bed synthesis was purchased from

Columbus Vegetable Oils (Des Plains, IL, USA). Shea but-

ter (100% Natural African—Ultra Reﬁned) was purchased

from Green Island Traders (Chagrin Falls, OH, USA).

Non-GMO soybean oil was generously donated by Colum-

bus Vegetable Oils. Ethyl ferulate (4-hydroxy-3-methox-

ycinnamic acid ethyl ester) was purchased from Capot

Chemical Co., Ltd. (Hangzhou, Zhejiang, China). Novo-

zym 435 (Candida antarctica lipase B immobilized on

acrylic beads) was purchased from Brenntag Great Lakes,

LLC. (Wauwatosa, WI, USA). All solvents and other rea-

gents were purchased from Sigma-Aldrich (St. Louis, MO,

USA).

Transesteriﬁcation Reactions

Batch

Batch reactions on a 500-g scale were conducted with

the following general procedure. Coconut oil, shea but-

ter, or soybean oil (400–500 g depending on the vegetable

oil, 0.57 mol) and ethyl ferulate (100 g, 0.45 mol) were

weighed into a 1-L Schlenk Flask and stirred under vacuum

(11.6 Pa) at 60 °C for 18 h. The dried solution was trans-

ferred to a 2-L, bafﬂed Erlenmeyer shake ﬂask and Novo-

zym 435 (75 g) was added. The shake ﬂask was stoppered

and ﬂushed with nitrogen for 15 min. The ﬂask was shaken

in an Innova 4080 Incubator Shaker (Eppendorf Inc., for-

merly New Brunswick Scientiﬁc, Hauppauge, NY, USA) at

60 °C and 115 rpm for 21 days. The oil product was col-

lected by vacuum ﬁltration through a double layer of What-

man 54 ﬁlter paper (Fisher Scientiﬁc, Pittsburgh, PA, USA)

using a Büchner funnel (135 × 60 mm) and a 2-L collec-

tion ﬂask. The Novozym 435 was recovered and used for

subsequent reactions for a minimum of three replicates.

Transesteriﬁcations of glyceryl trioctanoate, glyc-

eryl tridecanoate, glyceryl tridodecanoate, glyceryl

We examined shaken, batch transesteriﬁcation reactions

of ethyl ferulate with the solid vegetable oils, extra virgin

coconut oil and ultra reﬁned shea butter and compared their

performance to soybean oil. The feruloylated coconut oil

and feruloylated shea butter were examined for antioxidant

and ultraviolet (UV) absorption efﬁcacy and some of their

physical properties determined. Based on these ﬁndings,

the transesteriﬁcation of ethyl ferulate and extra virgin

coconut oil was attempted in a 1/4-tonne/year continuous,

1 3

J Am Oil Chem Soc

tritetradecanoate, and puriﬁed shea butter with ethyl feru-

late were conducted using the same general procedure as

described above but on a 1/10-scale.

light scattering detection (ELSD) with nitrogen as the car-

rier gas. Parameters were set to 4 min peak width, 0.1 V

collection threshold, 40 °C spray temperature, 70 °C drift

temperature, and UV detection at 340 nm. Eluent peaks

were collected in 480-mL French bottles. Raw shea but-

ter (5.0 g) was dissolved in acetone (5.0 mL) with heating

to 60 °C for 10 min. The solution was loaded by syringe

onto a dry 220 g RediSep R_fHigh Performance Gold silica

column with 1 column volume (CV) equaling 135 mL at

83 mL/min ﬂow rate. The RediSep silica column was aspi-

rated to dryness for 1 h. The sample loaded RediSep silica

column was developed with a hexane/ethyl acetate gradient

at 83 mL/min: 0% ethyl acetate for 1 CV, increased to 2%

ethyl acetate over 4 CV, held at 2% ethyl acetate for 2.7

CV, increased to 10% ethyl acetate over 4.5 CV, increased

to 100% ethyl acetate over 0 CV, held at 100% ethyl acetate

for 3 CV, decreased to 0% ethyl acetate over 0 CV, held

at 0% ethyl acetate for 2 CV. The column was air purged

for 2 min and then aspirated to dryness overnight for the

next sample loading. The puriﬁed shea butter (containing

only TAG) eluted during the 2% ethyl acetate to 10% ethyl

acetate gradient. Solvent was removed from eluent under

vacuum to produce a white, waxy solid. The procedure was

repeated ﬁve more times. Yield: 21.4 g (71.2 wt% relative

to the raw shea butter).

Continuous, Enzymatic, Packed‑Bed Bioreactor

The four-column bioreactor used for the continuous enzy-

matic esteriﬁcation of coconut oil and ethyl ferulate was

1/4-scale of the bioreactor design reported for the produc-

tion of feruloylated soybean oil [21] with the following

modiﬁcations. The bioreactor used four 304-stainless steel

columns (6.35 cm I.D. × 81.28 cm, 2.57 L). A bleed valve

was ﬁtted between column A and column B for sampling.

The feed pump was a PULSAtron E Plus pump (model

#LPK2SA-PTC4, Pulsafeeder, Punta Gorda, FL, USA)

set at 5% stroke length, 7.5% stroke rate (~0.60 mL/min)

and regulated by a GE (Fairﬁeld, CT, USA) basic 125 V,

15 Amp, 30 min interval timer. Each column was plumbed

with a PULSAtron E Plus pump (P2, model #LE13SA-

PTC2) set at 24% stroke length, 40% stroke rate (~4.0 mL/

min) for a bottom-up ﬂow.

Batches (20 kg) of the coconut oil/ethyl ferulate sub-

strate were prepared following the same basic procedure as

previously described for soybean oil [21] except at a 4:1

coconut oil:ethyl ferulate mass ratio (1.3:1.0 mol:mol). The

heated (65–75 °C) coconut oil/ethyl ferulate substrate was

ﬁltered through a cooking oil ﬁlter (Chard, Two Rivers, WI,

USA, Model OPF-10) before being dried for 3–4 h at 60 °C

under partial vacuum with a nitrogen sparge and then trans-

ferred to the bioreactor feed tank as previously described

[21].

Each column was charged in the bioreactor heating

cabinet (60 °C) with a slurry of Novozym 435 (500 g) and

coconut oil (2.0 L) preheated to 60 °C. The excess coco-

nut oil was gravity drained to pack the column and form

the enzyme bed. The column ends were capped with glass

wool and ﬁlter screen gaskets and the system plumbed as

previously described [21]. The coconut oil was displaced

with coconut oil/ethyl ferulate substrate at a feed rate of

~0.65 mL/min. The bioreactor was then run in stop mode,

feed pump off and recirculation pumps at ~4 mL/min, for

22 days to equilibrate the Novozym 435. The bioreactor

was then switched to ﬂow mode, feed pump at ~0.6 mL/

min and recirculation pumps at ~4 mL/min, for continuous

production of feruloylated coconut oil at 0.75–1.0 kg/day.

New, dried batches of coconut oil/ethyl ferulate substrate

(20 kg) were added to the feed tank as needed.

Analytical Methods

HPLC

HPLC analysis for feruloyl soy glycerol species’ quanti-

ﬁcation was based on previously published methods [24].

Analyses were performed on two systems. HPLC₁was a

Thermo Separation Product (Thermo Fischer Scientiﬁc

Inc., Waltham, MA, USA) HPLC system consisting of a

Spectra System AS3000 autosampler, a Spectra System

P4000 pump, a Spectra System UV6000LP detector, an

Alltech (Deerﬁeld, IL, USA) 500 Evaporative Light-Scat-

tering Detector (ELSD), and a Thermo Separation Prod-

ucts SCM 1000 Membrane Degasser. HPLC₂was an Agi-

lent (Santa Clara, CA, USA) 1100 system consisting of a

G1322A Degasser, G1311A QuatPump, G1313A Autosa-

mpler, G1316A Column Heater, and G1315A Diode Array

Detector. Solvents were ﬁltered using a Whatman 0.45-μm

nylon membrane (Sigma-Aldrich) before use, and samples

were developed on a Luna Phenyl-Hexyl column (5 μm,

250 × 4.6 mm, Phenomenex, Torrance, CA, USA).

Soybean oil and shea butter products were developed

with a gradient of solvent A: water (268 mL), methanol

(70 mL), 1-butanol (11 mL), and glacial acetic acid (1 mL),

solvent B: water (93 mL), methanol (245 mL), 1-butanol

(33 mL), and glacial acetic acid (1 mL), and solvent C:

methanol. The column method at 1.0 mL/min was a 5 min

Flash Chromatography Puriﬁcation of Shea Butter

Raw shea butter was puriﬁed using a Teledyne Isco, Inc.

(Lincoln, NE, USA) CombiFlash R_f200i ﬂash chroma-

tography system. Products were monitored by evaporative

1 3

J Am Oil Chem Soc

isocratic ﬂow of 3:1 A:B, a 2 min linear gradient to 100%

B, a 5 min isocratic ﬂow of 100% B, a 2 min linear gradient

to 100% C, a 13 min isocratic ﬂow of 100% C, followed by

a 3 min linear gradient to 3:1 A:B. The UV (325 nm, 7 nm

bandpass) and diode array (325 nm) detector response was

calibrated with ethyl ferulate and ferulic acid. Injection vol-

umes were 10 μL.

Coconut oil species were determined using a modiﬁed

gradient at 1.0 mL/min with a 5-min isocratic ﬂow of 3:1

A:B, a 2-min linear gradient to 100% B, a 5-min isocratic

ﬂow of 100% B, a 2-min linear gradient to 1:9 B:C, a

13-min isocratic ﬂow of 1:9 B:C, followed by a 3-min lin-

ear gradient to 3:1 A:B. The UV (325 nm, 7 nm bandpass)

and diode array (325 nm) detector response was calibrated

with ethyl ferulate and ferulic acid. Injection volumes were

10 μL.

GC

GC analysis was conducted on a Hewlett-Packard 6890N

Series gas chromatograph (Palo Alto, CA, USA), equipped

with a ﬂame ionization detector, autosampler/injector, and

a SP 2380 30 m × 0.32 mm i.d. (Supelco, Bellefonte, PA,

USA) column. SP 2380 analysis was adapted from that

of Isbell et al. [25] with a 1.4 mL/min column ﬂow with

a helium head pressure of 138 kPa; split ratio 10:1; pro-

grammed ramp 120–135 °C at 10 °C/min, 135–175 °C

at 3 °C/min, and 175–265 °C at 10 °C/min; injector and

detector temperatures set at 250 °C. Saturated and unsatu-

rated C₈–C₃₀fatty acid methyl ester standards were used

for assignment of fatty acid methyl esters obtained from

soybean oil, coconut oil, and shea butter.

Methyl esters were prepared by reacting vegetable oils

(10 mg) with 0.5 M KOH/methanol (0.5 mL) in a sealed

vial for 1 h at 100 °C in a heating block. The sample was

cooled to room temperature and 1.5 mL of 1 M H₂SO₄/

methanol was added, the vial resealed, and the vial heated

to 100 °C for 15 min. The mixture was transferred to a 2

dram vial (7.5 mL) and 1 mL of water added. The solution

was extracted with 1 mL of hexane. The hexane extract

was dried over sodium sulfate and injected into the GC for

FAME analysis.

LC–ESI–MS

Samples were run on a Thermo Electron LTQ Orbitrap Dis-

covery Mass Spectrometer (Thermo Fisher Scientiﬁc Inc.)—

a linear ion trap (LTQ XL) MS, coupled to a high precision

electrostatic ion trap (Orbitrap) MS with a high energy col-

lision (HCD) cell—with an Ion Max electrospray ionization

(ESI) source, and a Thermo Scientiﬁc (Thermo Fischer Sci-

entiﬁc Inc.) ACCELA series HPLC system (ACCELA 1250

UHPLC pump, ACCELA1 HTC cool stack autoinjector, and

an ACCELA 80 Hz PDA detector) all running under Thermo

Scientiﬁc Xcalibur 2.1.0.1140 LC–MS software. The MS

was typically calibrated at least monthly with a standard cali-

bration mixture recommended by Thermo Scientiﬁc and the

signal detection optimized by running the autotune software

feature as needed. The MS was run with the ESI probe in the

negative mode. The source inlet temperature was 300 °C, the

sheath gas rate was set at 50 arbitrary units, the auxiliary gas

rate was set at 5 arbitrary units and the sweep gas rate was

set at 2 arbitrary units. The maximal mass resolution was set

at 30,000, the spray voltage was set at 3.0 kV, and the tube

lens was set at −100 V. Other parameters were determined

and set by the calibration and tuning process. The column

was a Luna Phenyl-Hexyl column (3 µm, 3 × 150 mm,

100 Å, Phenomenex).

NMR Spectroscopy

¹H- and ¹³C-NMR spectra were obtained on a Bruker

Avance 500 spectrometer (500 MHz H/125.77 MHz ¹³C)

1

using a 5 mm BBI probe. All samples were dissolved in

CDCl₃and all spectra were acquired at 27 °C. Chemical

shifts are reported as ppm from tetramethylsilane calcu-

lated from the lock signal (Ξ_D= 15.350609%).

Karl Fischer Titration

Water concentration was determined using a Mettler Toledo

DL 32 Karl Fischer Coulometer (Columbus, OH, USA),

following AOCS Method Ca 2e-84 [26] with 70:30 (v/v)

Hydranal AG-H/chloroform as the analyte.

For LC separation analysis, the initial solvent condi-

tions were 25% methanol in water with 0.25% formic

acid at a ﬂow rate of 0.25 mL/min. After injection (1 µL

or less) the column was developed with a linear gradient

to 100% methanol with 0.25% formic acid in 5 min then

held at 100% methanol with 0.25% formic acid for 26 min.

The column efﬂuent was monitored at 325 nm in the PDA

detector. The software package was set to collect mass data

between 100 and 2000 amu. Generally, the most signiﬁcant

sample ions generated under these conditions are [M−H]⁻

and [M+HCOO]⁻.

Total Acid Value

Total acids were determined using a Metrohm Ltd. (Her-

isau, Switzerland) 751 GPD Titrino equipped with a

Metrohm pH electrode, following the AOCS Method Cd

3d-63 [27] with ethanol substituted for methanol.

Free Radical Scavenging

Free radical scavenging capacity was determined using

a DPPH assay [28] with modiﬁcations [29] where DPPH

1 3

J Am Oil Chem Soc

R

spontaneously forms a free radical in solution. Feruloylated

soy glycerides were mixed with 0.2 mM DPPH in ethanol.

The examined concentration range of antioxidant (feru-

loylated soy glyceride) was 0.5–2.5 mM. Reduction of the

DPPH radical was followed at 517 nm for 30 min at 2 min

intervals at room temperature (22–24°C). Experiments

were performed in triplicate. The DPPH radical remaining

(%) over time was calculated as the ratio of sample absorb-

ance to a control DPPH sample at the same time, multiplied

by 100.

R

CH₂CH₃

Fr

+

β

α

ethyl ferulate (EF)

vegetable oil (R = fatty acid)

60 °C Novozym 435

Fr

OH

Fr

H

Fr

ferulic acid

(FA)

OH

diferulyol glycerol

ferulyol glycerol

(FG)

(F₂G)

Fr

R

Fr

R

Lovibond Color

R

OH

diferuloyl monoacylglycerol

ferulyol monoacylglycerol

feruloyl diacylglycerol

(FMAG)

(FDAG)

(F₂MAG)

O

Lovibond color was measured on an AOCS Tintometer

(model ca. 1950s, Lovibond Tintometer LTD., Salisbury,

England) following the AOCS Method Cc 13e-92 [30]. The

AOCS Color Comparison Tube (Kimax #45290, 15.4 × 2.0

i.d. cm, Capitol Scientiﬁc, Austin, TX, USA) was ﬁlled to

the 5 ¼″ line (13.335 cm) with sample and the color com-

pared to Lovibond colored glass ﬁlters.

O

feruloyl moiety (Fr)

HO

Fig. 1 Structures of feruloylated species obtained from the solvent-

less transesteriﬁcation of vegetable oils with ethyl ferulate. The α and

β glycerol hydrogens of the vegetable oil are labeled for reference of

¹H-NMR frequency assignments in Fig. 4

Differential Scanning Calorimetry

DSC data was obtained using a TA instruments (New Cas-

tle, DE) model 2920 V4.3A calibrated with an indium

standard (156.618 °C, enthalpy of melting = 28.86 J/g).

Data sampling and temperature controls were controlled by

the “TA Instrument Control” software program. Approxi-

mately 5.0–10.0 ( 0.1) mg of melted samples were

weighed into an aluminum DSC pans and hermetically

sealed. The sealed DSC samples were referenced against

an identical empty aluminum pan and run using modula-

tion ( 0.40 °C every 40 s) at a linear heating rate (β) of

5.0 °C/min from −55.0 to 90.0 °C. To eliminate tempera-

ture history effects and evenly distribute the samples in the

pans, the samples were ﬁrst cycled across the temperature

range. The samples were subsequently cooled to −55 °C at

5.0 °C/min, held isothermally for 1 min, and then reheated

across the melting region to obtain the melting proﬁle.

TA Universal Analysis 2000 software, version V4.3A (TA

Instruments, DE) was used to obtain heat ﬂow (W/g) versus

temperature (°C) curves from each experiment. These DSC

thermograms were subsequently analyzed to determine

peaks, onset, and total enthalpies.

with aromatic moieties (e.g., caffeoyl, feruloyl, sinapoyl)

possessing UV absorbing and antioxidant functionalities.

The transesteriﬁcation results in multiple molecular species

depicted in Fig. 1. The most studied of these reactions is the

transesteriﬁcation of ethyl ferulate with soybean oil using

soybean oil as the solvent [21, 31]. Ethyl ferulate melts and

is miscible with soybean oil at the reaction temperature

of 60 °C. These reactions typically require 14–22 days to

reach an equilibrium of ~65% conversion of ethyl ferulate

into the molecular species in Fig. 1, which have been iden-

tiﬁed and quantiﬁed by HPLC–MS [24]. The feruloylated

acylglycerols, unreacted ethyl ferulate, unreacted soybean

oil, and fatty acid ethyl ester byproducts remain a homo-

geneous solution at room temperature, except when F₂G

forms >2% of the total feruloylated species and precipitates

from the solution [32].

The transesteriﬁcation of ethyl ferulate has been con-

ducted with other vegetable oils (e.g., canola, ﬂaxseed,

olive) [18, 31] but not with vegetable oils that are solid at

room temperature. Coconut oil and shea butter are room

temperature vegetable fats that have gained prominence as

ingredients in the personal care industry. The fatty acid pro-

ﬁles of coconut oil, shea butter, and non-genetically modi-

ﬁed organism (non-GMO) soybean oil were determined

(Table 1). Coconut oil contained mostly saturated short and

medium chain fatty acids compared to the shea butter and

soybean oil. Shea butter and soybean oil contained mostly

C18 fatty acids but soybean oil is more highly unsaturated.

Feruloylated soybean oil is produced commercially in a

1-metric tonne/year continuous, enzymatic, packed-bed

Results and Discussion

Feruloylated Coconut Oil and Shea Butter

The transesteriﬁcation of vegetable oils with phenylpropa-

noic acids and their esters catalyzed by lipases results in

the exchange of fatty acid groups from the triacylglycerol

1 3

J Am Oil Chem Soc

Table 1 Fatty acid proﬁle

of vegetable oils used for the

transesteriﬁcation reactions

Fatty acid

Coconut oil (%)

Shea butter (%)

Raw

Soybean oil

(%)

Puriﬁed

C6:0

0.77 0.10

9.02 0.54

6.31 0.09

49.10 0.26

19.01 0.27

7.50 0.09

2.90 0.05

4.57 0.07

0.74 0.01

–

C8:0

–

C10:0

C12:0

C14:0

C16:0

C18:0

C18:1

C18:2

C18:3

C20:0

C22:0

C24:0

Other

Mol. wt.

–

0.21 0.01

0.14 0.00

5.43 0.01

28.41 0.02

60.80 0.10

1.83 0.01

0.36 0.00

1.12 0.01

0.14 0.00

0.09 0.00

1.47 0.10

882.94

0.21 0.03

0.15 0.02

5.69 0.04

28.36 0.22

62.24 0.27

1.73 0.03

0.32 0.03

1.09 0.03

0.14 0.02

0.08 0.01

–

10.87 0.06

3.83 0.01

21.58 0.07

53.53 0.15

7.96 0.02

0.32 0.01

0.38 0.01

–

1.71 0.02

873.38

670.39

882.74

Results are presented as means SD (n = 5 trials)

Raw shea butter as obtained from the supplier and after puriﬁcation by ﬂash chromatography

Mol. wt. was the average molecular weight (g/mol) of the vegetable oil determined by the relative quantity

of fatty acids

80

60

40

20

0

bioreactor [21]. It was not certain that lipase-catalyzed

transesteriﬁcation of ethyl ferulate with coconut oil and

shea butter would proceed similarly to that of soybean oil;

ethyl ferulate being miscible in the vegetable oil, the reac-

tion mixture remaining a solution over the course of the

reaction, and the product being a liquid. The transesteri-

ﬁcations were ﬁrst examined in 0.5-L scale shaken, batch

reactions.

Ethyl ferulate and coconut oil or shea butter were com-

bined at a 1.0:1.3 mol ratio. The substrates were heated to

60 °C and dried under vacuum to <35 ppm water. Ethyl

ferulate formed miscible solutions with both vegetable oils

but the solutions quickly solidiﬁed when allowed to cool to

room temperature. The substrate mixtures were, therefore,

kept at 60 °C for all manipulations. Novozym 435 was used

at a 0.75 wt% relative to ethyl ferulate to catalyze the reac-

tions. The time course of the conversion of ethyl ferulate

to the feruloylated vegetable oils are shown in Fig. 2, the

transesteriﬁcation with non-GMO soybean oil was done

as a comparative control. The transesteriﬁcations of each

vegetable oil were performed in quadruplicate and in series

using the same enzyme load; the enzyme was collected

from a reaction by ﬁltration and used in the subsequent

reaction. The data from the ﬁrst reaction in the series was

discarded due to the adsorption of the ethyl ferulate by the

acrylic resin of the Novozym 435 beads, which resulted in

inaccurate conversion yields [20].

0

5

10

15

20

25

Time (d)

Fig. 2 Conversion of EF to feruloylated lipids via lipase-catalyzed

transesteriﬁcations with coconut oil (circles), shea butter (squares),

puriﬁed shea butter (diamonds), and non-GMO soybean oil (trian‑

gles) in shaken batch reactions. Data points are reported as the mean

with error bars denoting SD from the mean, n = 3 trials. For illustra-

tive purposes, SigmaPlot V11.2.0.5 exponential rise to maximum best

ﬁt lines (f = y₀+ a × (1 − exp(−b × x)) + c × (1 − exp(−d × x)))

are provided, coconut oil (solid), shea butter (dot), puriﬁed shea but-

ter (dash), and non-GMO soybean oil (dash-dot)

conversion after 22 days. The solution remained clear with

no visible precipitate forming over the course of the reac-

tion. The molecular species of feruloylated soybean oil

have been identiﬁed by LC–MS [24]. The HPLC chroma-

togram of feruloylated coconut oil (Fig. 3) is comparatively

The transesteriﬁcation of ethyl ferulate and coconut oil

at 60 °C with gentle shaking reached equilibrium at 63%

1 3

J Am Oil Chem Soc

Table 2 Major products obtained from the transesteriﬁcation of tria-

cylglycerol standards with ethyl ferulate

EF

TAG

Feruloylated species

FG FA F₂G

FMAG

F₂MAG

FDAG

1

2

R_t(min)

C8

14.49

16.87

17.93

18.61

394.20

422.23

450.26

478.29

393.18

421.22

449.27

477.28

1.01

15.50

17.42

18.22

18.83

394.20

422.23

450.26

478.29

393.19

421.22

449.27

477.29

5.58

17.95

18.60

19.94

21.70

C10

C12

C14

C8

FMAG

F₂MAG

FDAG

19.26

24.08

19.97

27.49

Calculated mass

Major ion (m/z)

Peak area (%)

570.25

598.28

626.31

654.79

569.24

597.27

625.30

653.33

12.30

520.30

576.37

632.43

688.49

519.30

575.36

631.42

687.48

55.33

C10

C12

C14

C8

3

0

5

10

15

20

25

30

Time (min)

C10

C12

C14

C8

Fig. 3 HPLC chromatograms (UV detector, 325 nm) of the transes-

teriﬁcation reactions of coconut oil (1), tridodecanoate (2), and tritet-

radecanoate (3) with EF. FG, FA, F₂G, and EF were determined vs

standards. FMAG, F₂MAG, and FDG were identiﬁed by LC–MS (see

Table 2). See Fig. 1 for deﬁnitions of abbreviations

C10

C12

C14

1.36

5.98

14.35

44.40

1.10

5.40

15.13

47.15

0.99

5.52

15.59

44.46

Feruloylated species are deﬁned in Fig. 1

more convoluted due to the greater diversity and relative

abundance of different fatty acids; soybean oil, >85% C18

vs coconut oil, 9% C8, 6% C10, 49% C12, 19% C14, 8%

C16 (Table 1). EF, FA, FG, and F₂G (Fig. 1) were iden-

tiﬁed vs standards in the feruloylated coconut oil HPLC

chromatograms. The parent ion of each HPLC peak of the

feruloylated coconut oil samples could not deﬁnitively be

identify because of the coelution of molecular species. To

unambiguously determine retention times of the major fer-

uloylated coconut oil molecular species the transesteriﬁca-

tion of ethyl ferulate was conducted with TAG standards,

trioctanoate, tridecanoate, tridodecanoate, and tritetrade-

canoate, using the same shaken, batch reaction parameters

at 1/10-scale.

The neat (without solvent) transesteriﬁcations of ethyl

ferulate with the TAG standards proceeded similarly to

those with coconut oil at 60 °C in shaken, batch reactions.

Transesteriﬁcations with trihexadecanoate (mp 66–67 °C)

could not be done neat under the examined reaction condi-

tions and were not attempted. The HPLC chromatograms

of the feruloylated tridodecanoate and tritetradecanoate

reactions are shown in Fig. 3. Each chromatogram shows

three distinct peaks >15.0 min R_t(retention time) that

were identiﬁed by LC–MS. The feruloyl monoacylglycer-

ols (FMAG, see Fig. 1) eluted before the diferuloyl mono-

acylglycerol (F₂MAG) species. The feruloyl diacylglycerol

(FDAG) species eluted last. The R_tof the feruloylated spe-

cies and their major ions determined by LC–MS are listed

in Table 2. The FMAG peaks had a small preceding peak

with the same molecular weight, which was attributed to a

minor 1,2-FMAG isomer [24]. Fatty acid chain length had

R_t(min) were the retention times determined by HPLC diode array

detector (325 nm), see Fig. 3 for representative HPLC chromatograms

Major ions detected by LC–MS in negative ion mode, all determined

to be [M−H]⁻

Peak area (%) were relative to the total peak area of all feruloylated

species, including FA, F₂G and EF, obtained from individual reac-

tions of each TAG with EF

the greatest inﬂuence on the R_tof the FMAG species, ∆R_t

3.4 min for C14 vs C12 (Fig. 3; Table 2). The inﬂuence of

fatty acid chain length on ∆R_tdecreased with the decreas-

ing chain length. The difference in chain length also caused

longer fatty acid chain FMAG species to elute at the same

R_tas shorter fatty acid chain F₂MAG species (Table 2).

For example, the C12 FMAG species eluted at the same

R_t17.9 min as the C8 F₂MAG species, and similarly, the

C14 FMAG species coelutes with the C10 F₂MAG spe-

cies, R_t18.6 min. The ratio, based on their relative abun-

dance by HPLC, of the feruloylated lipid species formed

during the transesteriﬁcations were similar for all four TAG

standards (Table 2). The most abundant were the triacylg-

lycerol products FMAG (44–55%) and F₂MAG (12–16%).

The diacylglycerol species, FMAG, was the least preva-

lent (6–7%). The feruloyl, non-fatty acid products, F₂G

and FA, were each less than 1% of the total feruloylated

species for both the coconut oil and TAG standard trans-

esteriﬁcations. The relative abundance of feruloylated lipid

species from the transesteriﬁcation of the TAG standards,

FDAG ≫ F₂MAG > FMAG, was different from those

obtained from the transesteriﬁcation of ethyl ferulate with

1 3

J Am Oil Chem Soc

triolein, FMAG > FDAG > F₂MAG [24]. This was possibly

due to the inﬂuence of higher water content in the triolein

reactions, 250 ppm [20], relative to the coconut transesteri-

ﬁcations, <35 ppm. The higher water content of the triolein

transesteriﬁcations also resulted in higher quantities of FA

and F₂G, 2 and 6%, respectively. The transesteriﬁcation of

octyl dihydrocaffeate with trioctanoate in a 1:1 mol:mol

ratio catalyzed by Novozym 435 in stirred batch reac-

tions also produced FMAG > FDAG > F₂MAG [22]. The

authors did not report the water content of their phenolic

ester-trioctanoate substrate, and their experimental did not

indicate that precaution was taken to dry the substrate. This

suggested that chain length of the TAG, C8 of triocatante

vs C18 of triolein, was not responsible for the difference in

abundance of the feruloylated acylglycerol species formed

in the transesteriﬁcations of ethyl ferulate with the C8–C12

TAG standards compared to triolein. Because coconut oil

does not simply consist of a mixture of TAG standards but

instead consists of several, mixed fatty acid chain length

TAG, the number of feruloylated coconut oil species that

could potentially coelute made deﬁnitive peak identiﬁca-

tion by LC–MS parent ion implausible. In fact, the HPLC

peaks of the feruloylated coconut oil species were so con-

voluted that the FMAG, F₂MAG, and FDAG could not be

assigned general R_tranges as was done for feruloylated

soybean oil [24].

While the transesteriﬁcation of ethyl ferulate with coco-

nut oil performed similarly to that with soybean oil in the

shaken, batch reactions, the transesteriﬁcation of ethyl

ferulate with shea butter produced unexpected results.

With very similar fatty acids proﬁles (Table 2), shea but-

ter (>95% C18: 0–3) and non-GMO soybean oil (>85%

C18: 0–3) were expected to produce very similar prod-

ucts. The transesteriﬁcation with shea butter did result in

the same three-region feruloylated lipid proﬁle as deter-

mined by HPLC R_t[24]. Under the very dry transesteri-

ﬁcation conditions (<35 ppm water) examined herein, the

relative abundance of the feruloylated lipid species of shea

butter was the same as those obtained from the transes-

teriﬁcations with the non-GMO soybean oil and C8–C12

TAG standards, FDAG > F₂MAG > FMAG. After 24 h,

however, the shea butter transesteriﬁcations produced a

white precipitate not observed in the coconut oil, C8–C12

TAG standards, or soybean oil transesteriﬁcations. Also,

the ethyl ferulate transesteriﬁcations with the shea butter

reached equilibrium sooner than those with the other veg-

etable oils and reached a slightly higher conversion, 72%

(Fig. 2). The white precipitate was isolated and identiﬁed

by HPLC and ¹H NMR as F₂G, which was produced in the

shea butter transesteriﬁcations at 6–8%. This was a much

higher concentration of F₂G than was formed in transester-

iﬁcations with coconut oil (<1%), C8–C12 TAG standards

(<1%), and non-GMO soybean oil (<1%). Since the water

content was controlled for all of the transesteriﬁcations,

hydrolysis of the F₂MAG was not suspected. There is prec-

edent that quantities of MAG and DAG leads to increased

quantities of F₂G in similar transesteriﬁcations, and if

F₂G is present in concentrations >1% it can precipitate

1

from the feruloylated vegetable oil product [32–34]. H-

NMR spectroscopy of the shea butter identiﬁed 16 mol%

of DAG was present in the shea butter as obtained from

the supplier (raw) (Fig. 4). Comparatively, coconut oil and

non-GMO soybean oil did not contain DAG as determined

by the absence of 1,2-DAG and 1,3-DAG alpha- and beta-

hydrogens [35].

To conﬁrm that the DAG contaminants in the shea but-

ter were the cause of high concentrations of F₂G and the

accelerated rate to equilibrium, transesteriﬁcations of ethyl

ferulate were conducted with shea butter TAG puriﬁed by

ﬂash chromatography. Raw shea butter was developed with

a hexane/ethyl acetate gradient on silica gel to produce

three fractions (Fig. 5). The fractions were identiﬁed by ¹H

NMR (Fig. 4) and found to be aromatic/phenolic contami-

nants containing some aliphatic material (aromatics), puri-

ﬁed shea butter (TAG), and hydrolyzed shea butter (DAG

and MAG). The fatty acid proﬁle of the puriﬁed shea butter

was very similar to the raw shea butter (Table 1), consist-

ing of >95% C18: 0–3. Shaken, batch transesteriﬁcations

of ethyl ferulate with puriﬁed shea butter (5-g scale) did

not produce a white precipitate and resulted in <1% F₂G at

equilibrium. The rate of transesteriﬁcation and the percent

conversion of the puriﬁed shea butter transesteriﬁcations

were more comparable to those of non-GMO soybean oil

than those of the raw shea butter. While it was possible to

obtain gram quantities of puriﬁed shea butter for analyti-

cal purposes it would have been cost prohibitive to purify

quantities (100 kg) needed to examine the transesteriﬁca-

tion of ethyl ferulate and shea butter in a 1/4-tonne/year,

continuous, enzymatic, back-bed bioreactor.

Continuous, Enzymatic, Packed‑Bed Bioreactor

Production of Feruloylated Coconut Oil

Pilot scale (1/4-tonne/year) production of feruloylated

coconut oil was performed in a four-column, continuous,

enzymatic, packed-bed bioreactor [21]. The bioreactor

design ensured that the pumps, columns, valves, and feed

tank were maintained at 60 °C. The feruloylated coconut

oil product collection tank was exterior to the bioreactor

and kept at ambient temperature (22–24 °C). As observed

for the shaken, batch reactions, the harvested feruloylated

coconut oil remained liquid for up to 14 days before

solidifying; the feruloylated coconut oil product was eas-

ily melted and formed a homogenous oil upon reheating to

60 °C for manipulation and packaging. The melting point

of the feruloylated coconut oil is discussed below.

1 3

J Am Oil Chem Soc

1,2-DAG_β

TAG_β

1,3(2)-DAG_α1,3-DAG_β

TAG_α

*

Shea butter (raw)

1

2

3

4

Shea butter (DAG & MAG)

Shea butter (aromatics)

Shea butter (TAG)

Coconut oil

Soybean oil

5

6

Fig. 4 ¹H-NMR (500 MHZ, *CDCl₃) spectra of soybean oil (6),

coconut oil (5), and shea butter (1) as obtained from the suppliers.

Spectra 2–4 correspond to materials obtained from the ﬂash chroma-

tography puriﬁcation of raw shea butter (see Fig. 5) resulting in pure

shea butter TAG (4) and isolated aromatic (3) and DAG and MAG (2)

contaminants

Fig. 5 Flash chromatography puriﬁcation of shea butter. Far left

axis (ELS) and curved trace correspond to the ELSD signal, the UV

absorbance (254 nm) signal was zero, and the right axis and darker

linear trace correspond to the percent ethyl acetate (v:v, Percent B) in

hexane during elution. Numbers above the trace denote correspond-

ing eluent collection bottles, and the bar below the trace denotes the

peak corresponding to pure shea butter TAG. See Fig. 4 for ¹H-NMR

spectra of compounds isolated from collected peaks

The bioreactor conversion of ethyl ferulate and coconut

oil was modiﬁed from the previously described procedure

for the conversion ethyl ferulate and soybean oil to form

feruloylated soy glycerides [21]. After the columns were

packed with the Novozym 435-coconut oil slurry and

the coconut oil displaced with coconut oil:ethyl ferulate

1 3

J Am Oil Chem Soc

Table 3 Continuous, enzymatic, packed-bed, bioreactor operating

modes for the 4.5 months production of feruloylated coconut oil

substrate (1.3:1.0 mol:mol, 4:1 wt:wt) the bioreactor was

run in stop mode (feed pump off and recirculation pumps

on, see Table 3) for 22 days to allow the conversion of

ethyl ferulate to feruloylated coconut oil to reach equilib-

rium before product collection was started. This avoided

~20 kg of feruloylated coconut oil waste that would have

been unusable due to low conversion percentages, waste

that was produced with the analogous soybean oil biore-

actor runs [21]. After 22 days the bioreactor was switched

to ﬂow mode (feed pump on and recirculation pumps on)

and feruloylated coconut oil continuously collected over

a 4.5 months period. Figure 6 shows the quantity of feru-

loylated coconut oil accumulated and the percent conver-

sion of ethyl ferulate to feruloylated coconut oil over the

4.5 months bioreactor run using a single charge of Novo-

zym 435. Over the ﬁrst 30 days of the bioreactor run in ﬂow

mode, the effective substrate feed rate of the feruloylated

Mode

Stop/ﬂow times

Collection rate Collections

(h/h) (Feed pump

off)

(g/min)

(n)

A

B

C

Stop

24/0 12–12 am, pm 0.000 0.000

0/24 0.633 0.006

0

9

3

Flow

–

Stop-

ﬂow

6/18 8–11 am, pm 0.468 0.010

D

E

Stop-

ﬂow

8/16 8–10 am, pm 0.431 0.003

2–4 am, pm

3

9

Stop-

ﬂow

12/12 8–11 am, pm 0.365 0.018

2–5 am, pm

Letters correspond to the lettered boxes in Fig. 6

Stop mode = feed pump off, column recirculation pumps on and ﬂow

mode = feed pump on, column recirculation pumps on

Collection rates of the feruloylated coconut oil product are reported

coconut oil was 0.633

0.006 g/min based on the mass

as the means SD of n collections

Fig. 6 Conversion of ethyl

100

A

B

C

D

E

ferulate to feruloylated coconut

oil via lipase-catalyzed trans-

esteriﬁcation in a continuous,

packed-bed bioreactor. Top

left axis feruloylated coconut

oil product collected from the

bioreactor over the course of the

continuous run (closed circles).

Top right axis conversion of

ethyl ferulate to feruloylated

coconut oil (closed squares).

Bottom left axis relative quantity

of the sum of FMAG, F₂MAG,

and FDAG (open circles), EF

(open squares), FA (open up

triangles), and F₂G (open down

triangles). Bottom right axis

total ferulate concentration rela-

tive to initial EF concentration

in the EF-coconut oil substrate

(diamonds). Boxes: bioreac-

tor operating modes, A–E, are

detailed in Table 3

80

60

40

20

80

60

40

20

0

100

80

60

40

20

0

80

60

40

20

0

20

40

60

80

100

120

140

Time (d)

1 3

J Am Oil Chem Soc

Table 4 Physical properties of the vegetable oils and the feruloylated vegetable oils

Sample

Total ferulates (mM)

Total acid (mg/g)

T_onset(°C)

T_peak(°C)

Total ∆H_melt(J/g)

Lovibond color (Y, R)

SBO

–

0.1415 0.0046 k

0.1702 0.0110 k

0.0452 0.0004 l

0.3965 0.0158 j

4.9914 0.0374 a

1.5831 0.0298 f

0.9592 0.0235 i

4.7419 0.0888 b

2.2629 0.0422 b

1.7998 0.0111 e

1.9642 0.0411 d

1.4243 0.0213 g

1.2025 0.0216 h

–

2.0, 0.0

–

SHB_R

–

−26.5 (−2.5)

−33.6 (20.7)

7.8

−1.28 (24.9)

−0.62 (33.5)

22.1

65.1

91.9

103.9

–

SHB_TAG

CNO

–

2.0, 0.0

8.0, 1.0

–

FSBO

866 45 c

831 38 c

821 38 c

1113 54 a

1021 6 b

1006 2 b

1022 9 b

1025 17 a,b

1018 18 b

–

FSHB_R

−33.4 (28.6)

−46.2 (23.7)

−36.1 (−3.1)

−50.2 (−3.0)

−48.5 (−2.0)

−48.7 (−3.4)

−49.7 (−1.3)

−38.5 (−2.2)

4.82 (36.4)

12.6 (24.9)

−18.4 (16.5)

−18.4 (15.8)

−19.2 (16.8)

−19.2 (17.0)

−18.4 (17.4)

−19.1 (16.4)

49.9

69.8

61.7

69.9

74.4

72.3

81.5

80.4

FSHB_TAG

FCNO_SB

FCNO_CBP1

FCNO_CBP22

FCNO_CBP42

FCNO_CBP64

FCNO_CBP86

–

2.0, 0.5

2.0, 0.0

1.0, 0.0

0.5, 0.0

1.0, 0.0

Total ferulates and total acid (mg KOH/g of sample) are reported as the mean SD of n = 3 trials. Samples within a column not sharing the

same letter are signiﬁcantly different based on Differences of Least Squares Means with a Bonferroni adjustment at p ≤ 0.05

Temperature onset and peak, numbers in parentheses correspond to additional melting peaks, as determined by DSC

Lovibond color is reported as yellow and red

SBO soybean oil, SHB_Rraw shea butter, SHB_TAGshea butter puriﬁed by ﬂash chromatography, CNO coconut oil, FSBO feruloylated soybean

oil, FSHB_Rferuloylated raw shea butter, FSHB_TAGferuloylated puriﬁed shea butter, FCNO_SBferuloylated coconut oil from shaken batch reac-

tions, FCNO_CFB1‑86feruloylated coconut oil from continuous bioreactor production with number indicating total accumulated mass (kg) col-

lected when sampled

collected every 4 days (n = 9 collections). The conversion

to feruloylated coconut oil during the 22 days of stop mode

reached an equilibrium of 65% (samples obtained from the

sampling valve between column A and column B during

stop mode operation), very similar to the equilibrium con-

version obtained from the shaken, batch reactions of ethyl

ferulate and coconut oil, 63% (Fig. 2). When switched to

ﬂow mode at a feed rate of 0.633 0.006 g/min, however,

the conversion unexpectedly dropped to 54% over the next

8 days and remained at that conversion equilibrium for the

next 24 days. The total ferulate percentage relative to the

ferulate concentration of the ethyl ferulate-coconut oil sub-

strate was ~100% for the duration of the 4.5 months pro-

duction run (Fig. 6), indicating that ethyl ferulate and feru-

lated coconut oil species were not absorbed by the enzyme

substrate [20]. The enzyme substrate did absorb water as

evidenced by the very low concentration, <1%, of hydroly-

sis products, FA and F₂G, detected by HPLC (Fig. 6) and

very low total acid number of the feruloylated coconut

oil product, which decrease from 0.18 to 0.12% over the

4.5 months run (Table 4), obtained from ethyl ferulate-

coconut oil substrate that varied in water content from 250

to 1400 ppm. The absorption of water by Novozym 435 in

other packed-bed bioreactor applications resulting in lim-

ited hydrolysis product formation and low total acid num-

bers has been shown before [21, 36].

Feruloyl soy glycerides are produced and marketed at

60–65% conversion of ethyl ferulate to feruloyl soy glyc-

erides, so it was desirable to attempt to increase the con-

version of ethyl ferulate to feruloylated coconut oil. It was

previously shown that slowing the feed rate of the substrate

to the bioreactor (i.e., increasing the substrate residence

time) increased the conversion equilibrium [21]; therefore,

the effective feed rate of the ethyl ferulate:coconut oil sub-

strate to the bioreactor was decreased. Already operating

at the lowest settings of the feed pump (0.633 0.006 g/

min) the bioreactor was run in a stop-ﬂow mode where

the feed pump power was cycled on for 9 h and off for

3 h (8–11 am and pm), resulting in an effective feed rate

of 0.468 0.010 g/min based on the mass collected every

5 days (n = 3 collections), which was 3/4 of the original

feed rate (Table 3). The conversion of ethyl ferulate to fer-

uloylated coconut oil trended up to 57% for 10 days and

remained at 57% for the next 5 days. The stop-ﬂow rate

was adjusted to further decrease the substrate feed rate to

0.431

0.003 g/min for the next 3 weeks. The conver-

sion rate temporarily increased to 60% but the returned to

57% after 18 days. The substrate feed rate was adjusted to

0.365 0.018 g/min for the ﬁnal 63 days of the production

run. The substrate conversion again temporarily reached

60% conversion for 2 weeks, but slowly declined to 55%

conversion after seven additional weeks of continuous

1 3

J Am Oil Chem Soc

operation. The ﬁnal equilibrium rate of conversion of ethyl

ferulate to feruloylated coconut oil in the continuous, enzy-

matic, packed-bed bioreactor, 55%, was ~15% lower than

the equilibrium conversion of ethyl ferulate to feruloylated

soybean oil obtained in a similar, x4-scale bioreactor [21].

The results of the shaken, batch transesteriﬁcations of ethyl

ferulate with soybean oil vs coconut oil (discussed earlier),

showed that the enzymatic feruloylation of coconut oil was

slower to reach equilibrium than that of soybean oil and

that the ﬁnal equilibrium conversion was slightly lower

(Fig. 2). This suggested that the enzymatic transesteriﬁ-

cation of ethyl ferulate with the medium chain fatty acid

vegetable oil resulted in lower kinetic rates of conversion

and conversion equilibrium and that the optimal reaction

parameters for the continuous, enzymatic, packed-bed bio-

reactor transesteriﬁcation of ethyl ferulate with soybean oil

were not applicable to the transesteriﬁcation coconut oil in

a similar bioreactor at ¼-scale.

Several studies have been conducted using response

surface methodology to determine the optimal vegetable

oil:ethyl ferulate ratio for lipase-catalyzed transesteriﬁca-

tions to form feruloylated lipids [31, 33, 34, 37]. These

academic studies conclude optimal vegetable oil:ethyl

ferulate ratios of 4:1–10:1 (mol:mol) depending on reac-

tion temperature (>70 °C) and enzyme load (20–40 wt%

relative to ethyl ferulate). While interesting academic exer-

cises, the parameters determined from these small-scale,

stirred, batch reactions provide little practical insight for

large-scale, continuous, packed-bed, enzymatic syntheses

of feruloylated vegetable oils for use by the personal care

and cosmeceutical industries. To limit production costs by

avoiding post production reﬁnement of the feruloylated

vegetable oils it was necessary to limit the mass of residual

vegetable oil in the product since it lowers the concentra-

tion of the feruloylated species (i.e. efﬁcacy) in the reaction

product. The use of a 1.3:1.0 (mol:mol) ratio of vegetable

oil:ethyl ferulate at 60 °C in the bioreactor produced a feru-

loylated vegetable oil product, requiring no post produc-

tion reﬁnement, with acceptable ethyl ferulate conversion

(>55%) and useful UV-absorbing and antioxidant efﬁcacy.

1.5

1.0

0.5

0.0

UVB

UVA II

UVA I

300

320

340

360

380

400

Wavelength (nm)

Fig. 7 Ultraviolet (UV) absorption spectra of feruloylated coco-

nut oil (solid line), feruloylated shea butter (dashed line), and feru-

loylated soybean oil (dashed-dotted line) at 100 mg/L in ethanol.

The U.S. Food and Drug Administration categorizes UV irradiance

as UVC (<290 nm), UVB (290–320 nm), UVA II (320–340 nm), and

UVA I (340–400 nm) [39]

cinnamate (λ_max308) UVB absorber [32, 38], and 29 nm

less than avobenzone, a popular commercial dibenzometh-

ane (λ_max357) UVA I absorber [32, 39]. The feruloylated

vegetable oils’ λ_maxmake them excellent UVA II and good

UVB absorbers (Fig. 7) and their addition to commercial

formulations would augment the commercial active ingre-

dients’ broad spectrum coverage. From our experience, skin

care formulators design formulations recipes on an ingredi-

ent weight percent basis and have a convoluted understand-

ing of formulating on a mole basis. Thus, including feru-

loylated vegetable oils with molecular weights of >800 g/

mol requires more material than octinoxate or avobenzone,

which have relatively lower molecular weights <315 g/mol,

to obtain the same efﬁcacy, which requires equal moles

of chromophore. Feruloylated soy glycerides are used in

commercial formulations, but it would be advantageous if

a lower molecular weight triacylglyceride was used, pro-

ducing a feruloylated vegetable oil with improved efﬁcacy

on a weight basis. Comparing the three feruloylated veg-

etable oils from the shaken, batch transesteriﬁcations by

integrating the area of the UV-absorption spectra repre-

sented in Fig. 7, FCnO = 108.8 0.2, FSbO = 97.6 0.6,

FShB = 92.6 0.4 (n = 3), showed that feruloylated coco-

nut oil had an 11.5% higher absorption capacity relative

to feruloylated soybean oil and 17.5% higher absorption

capacity than feruloylated shea butter. This was expected

based on the relative molecular weights of the three veg-

etable oils (Table 1), which resulted in a higher molar

concentration of ferulate chromophore in the feruloylated

coconut oil, ~1000 mM of total ferulates compared to the

feruloylated soybean oil and shea butter, ~830 mM of total

ferulates (Table 4).

Ultraviolet Absorbance and Antioxidant Properties

of Feruloylated Vegetable Oils

Commercial, UV-absorbing active ingredients function by

absorbing UV radiation and dissipating the energy through-

out their conjugated molecular structure. Speciﬁc chemi-

cal structures possess different UV-absorbing maxima,

thus skin care formulations require multiple active ingre-

dients to provide broad spectrum UV protection [38, 39].

The feruloylated vegetable oils possessed a UV-absorbing

λ_max328 nm (Fig. 7). This maxima was 20 nm greater than

octinoxate, a popular commercial methoxy-substituted

1 3

J Am Oil Chem Soc

concentrations, 0.50, 1.00, 1.75, and 2.50 mM, over the

course of 30 min. The controls showed that none of the

vegetable oils as obtained from the suppliers exhibited any

DPPH radical scavenging capacity; even the raw shea but-

ter that contained aromatic/phenolic contaminants did not

show any DPPH radical scavenging capacity under the

conditions tested. It was also conﬁrmed that the puriﬁed

shea butter TAG did not possess DPPH radical scavenging

capacity. All three feruloylated vegetable oils were consid-

ered rapid free radical scavengers at the three highest con-

centrations tested. A rapid scavenger is deﬁned as one that

reduced 50% of initial DPPH radical concentration in less

than 5 min [28]. Only the feruloylated raw shea butter was

considered a rapid free radical scavenger at the lowest con-

centration tested. Overall, there was minimal difference in

scavenging capacity of the three feruloylated vegetable oils

at equivalent total ferulate concentrations, conﬁrming that

fatty acid chain length has no affect on scavenging capacity.

Soybean oil is a liquid at room temperature and feru-

loylated soybean oil is also a liquid at room temperature

and has been reported to remain so after 5 years of storage

[21]. The supplier reported the coconut oil and shea but-

ter, solids at room temperature, having melting points of

24.4 and 32–45 °C, respectively. The feruloylated coconut

oil from the shaken, batch reactions and from the continu-

ous, packed-bed bioreactor production had very similar

thermograms as measured by DSC (Table 4). The feru-

loylated coconut oil conveniently remained a liquid at room

temperature (23 °C) for several days before slowly cloud-

ing and eventually gelling. The feruloylated coconut oil

never completely solidiﬁed after several weeks of storage.

The gelled feruloylated coconut oil was easily liqueﬁed

by microwaving 500 mL on high (1000 W) for 30 s. The

feruloylated shea butter synthesized from both raw shea

butter and shea butter TAG isolated by ﬂash chromatogra-

phy solidiﬁed quickly upon sitting after ﬁltration from the

shaken, batch reactions. Feruloylated coconut oil was much

lighter in color than feruloylated soybean oil. Lovibond

color measurements showed that the feruloylated coconut

oil possessed a much lower yellow tint and a slightly lower

red tint (Table 4). Lighter color ingredients are often pre-

ferred when attempting to produce white colored creams

and lotions. These observations suggested that feruloylated

coconut oil would be fungible with feruloylated soybean oil

in current retail formulation protocols while feruloylated

shea butter would require modiﬁcation of current formula-

tion methods.

100

80

(A)

(B)

(C)

60

40

20

0

100

80

60

40

20

0

100

80

60

40

20

0

5

10

15

20

25

30

Time (min)

Fig. 8 Free radical, DPPH*, quenching kinetics of shea butter (a),

coconut oil (b), and soybean oil (c) as obtained from the suppliers

(closed circles), shea butter puriﬁed by ﬂash chromatography (closed

squares), and feruloylated shea butter, feruloylated coconut oil, and

feruloylated soybean oil at 0.50 mM (open circles), 1.00 mM (open

triangles), 1.75 mM (open squares), and 2.50 mM (open diamonds).

Rapid antioxidants, as deﬁned by Huang et al. [28], quench 50% of

the DPPH* within 5 min (area within the dashed box). Data points

are the means of n = 3 trials with error bars denoting SD from the

mean

The free radical scavenging capacity of the three feru-

loylated vegetable oils from the shaken, batch transes-

teriﬁcation were determined by their ability to reduce the

DPPH radical in ethanol solutions [28]. Figure 8 shows

the percentage DPPH radical remaining in the presence of

the three feruloylated vegetable oils at four total ferulate

Conclusions

The enzymatic transesteriﬁcation of coconut oil and shea

butter with ethyl ferulate were conducted in shaken, batch

1 3

J Am Oil Chem Soc

by ORAU under DOE contract number DE-AC05-06OR23100. All

opinions expressed in this paper are the authors’ and do not neces-

sarily reﬂect the policies and views of USDA, ARS, DOE, or ORAU/

ORISE. The authors would like to thank Elizabeth Lafond, Leslie

Smith, Amber Durham and Jeanette Little for their excellent techni-

cal assistance and Debra E. Palmquist, Agricultural Research Ser-

vice Midwest Area Statistician, for the statistical analysis of the data

reported in Table 4.

reactions and their kinetic and equilibrium conversions

compared with the well studied soybean oil transesteri-

ﬁcation. The ethyl ferulate transesteriﬁcation with coco-

nut oil, with a very dissimilar fatty acid proﬁle to soybean

oil, was slower to reach equilibrium, but reached the same

equilibrium conversion as soybean oil. The transesteriﬁca-

tion of coconut oil produced a feruloylated triacylglycerol

product with a higher concentration of total ferulates. The

major feruloylated coconut oil species were identiﬁed by

LC–MS using transesteriﬁcation reactions of ethyl ferulate

with medium chain TAG standards. The transesteriﬁcation

of ethyl ferulate with shea butter, with a very similar fatty

acid proﬁle to soybean oil, unexpectedly reach kinetic equi-

librium faster with a higher conversion of the ethyl feru-

late to feruloylated acylglycerol product than soybean oil.

The feruloylated shea butter reactions using raw shea butter

produced a white precipitate, identiﬁed as F₂G, that was not

present in the soybean oil, coconut, and puriﬁed shea butter

transesteriﬁcations. It was shown that the unusual kinetic

and equilibrium conversion rates and white precipitate were

due to DAG and MAG contaminants in the raw shea butter.

The objective of the shaken, batch reactions was to

determine if coconut oil and shea butter were suitable veg-

etable oils/fats to use in the continuous, enzymatic, packed-

bed bioreactor process used to manufacture feruloylated

soy glycerides. The shea butter, as obtained from the sup-

plier containing DAG and MAG, produced the precipitate

F₂G that could potentially accumulate and obstruct ﬂow in

the bioreactor. Purifying quantities of shea butter required

for a bioreactor production run was cost prohibitive; there-

fore, the pilot-scale transesteriﬁcation with shea butter was

not attempted. The enzymatic transesteriﬁcation of ethyl

ferulate with coconut oil was tested in a 1/4-tonne/year

bioreactor using the same reaction parameters used for the

previously studied pilot-scale transesteriﬁcation of ethyl

ferulate with soybean oil. While the conversion equilibrium

to feruloylated coconut initially reached similar equilib-

rium conversions as soybean oil, the equilibrium conver-

sion to feruloylated coconut oil declined by 15% over the

4.5 months production run. The feruloylated coconut oil

had the advantage of being 17% higher in total ferulate

concentration than feruloylated soybean oil or shea butter

and was a much lighter color than feruloylated soybean oil,

both advantageous characteristics preferred by skin care

formulators.

References

1. Dayrit FM (2014) The properties of lauric acid and their signiﬁ-

cance in coconut oil. J Am Oil Chem Soc 92:1–15

2. Cassiday L (2016) Coconut oil boom. Inform 27(5):6–13

3. Pham LJ, Casa EP, Gregorio MA, Kwon DY (1998) Triacylglyc-

erols and regiospeciﬁc fatty acid analyses of Philippine seed oils.

J Am Oil Chem Soc 75:807–811

4. Dayrit FM, Buenafe OEM, Chainani ET, Vera IMSD (2008)

Analysis of monoglycerides, diglycerides, sterols, and free fatty

acids in coconut (Cocos nucifera L.) oil by ³¹P NMR spectros-

copy. J Agric Food Chem 56:5765–5769

5. Anonymous (2015) Oleoline forecasts global fatty alcohol

capacity. Inform 26:229–231

6. Cassiday L (2016) Surfactants and skin. Inform 27:694–699

7. Davrieux F, Allal F, Piombo G, Kelly B, Okulo JB, Thiam M,

Diallo OB, Bouvet J-M (2010) Near infrared spectroscopy for

high-throughput characterization of shea tree (Vitellaria para‑

doxa) nut fat proﬁles. J Agric Food Chem 58:7811–7819

8. Lovett PN (2005) Shea butter industry expanding in west africa.

Inform 16:273–275

9. Lovett PN (2015) Shea butter: Properties and processing for

use in food. In: Talbot G (ed) Specialty oils and fats in food and

nutrition: properties, processing and applications. Woodhead

Publishing, Cambridge, pp 126–158

10. Honfo FG, Akissoe N, Linnemann AR, Soumanou M, Van

Boekel MAJS (2014) Nutritional composition of shea products

and chemical properties of shea butter: a review. Crit Rev Food

Sci Nutr 54:673–686

11. Hon KL, Tsang YC, Pong NH, Lee VWY, Luk NM, Chow CM,

Leung TF (2015) Patient acceptability, efﬁcacy, and skin bio-

physiology of a cream and cleanser containing lipid complex

with shea butter extract versus a ceramide product for eczema.

Hong Kong Med J 21:417–425

12. Hills G (2003) Industrial use of lipases to produce fatty acid

esters. Eur J Lipid Sci Technol 105:601–607

13. Hasan F, Shah AA, Hameed A (2006) Industrial applications of

microbial lipases. Enzyme Microb Technol 39:235–251

14. Hayes D (2004) Enzyme-catalyzed modiﬁcation of oilseed

materials to produce eco-friendly products. J Am Oil Chem Soc

81:1077–1103

15. Sharma S, Kanwar SS (2014) Organic solvent tolerant lipases

and applications. Sci World J. doi:10.1155/2014/625258

16. Compton DL, Laszlo JA (2002) Novel sunscreens from vegeta-

ble oil and plant phenols. US Patent, 6,346,236

17. Tan Z, Shahidi F (2012) A novel chemoenzymatic synthesis of

phytosteryl caffeates and assessment of their antioxidant activity.

Food Chem 133:1427–1434

18. Sorour N, Karboune S, Saint-Louis R, Kermasha S (2012)

Lipase-catalyzed synthesis of structured phenolic lipids in sol-

vent-free system using ﬂaxseed oil and selected phenolic acids as

substrates. J Biotechnol 158:128–136

Acknowledgements This research was supported in part by a Coop-

erative Research and Development Agreement with the Biotechnol-

ogy Research and Development Corporation, the PRB Foundation,

and in part by an appointment to the Agricultural Research Service

(ARS) Research Participation Program administered by the Oak

Ridge Institute for Science and Education (ORISE) through an intera-

gency agreement between the U.S. Department of Energy (DOE)

and the U.S. Department of Agriculture (USDA). ORISE is managed

19. Compton DL, Laszlo JA, Berhow MA (2000) Lipase-catalyzed

synthesis of ferulate esters. J Am Oil Chem Soc 77:513–519

1 3

J Am Oil Chem Soc

20. Laszlo JA, Compton DL, Eller FJ, Taylor SL, Isbell TA (2003)

Packed-bed bioreactor synthesis of feruloylated monoacyl- and

diacylglycerols: clean production of a ****Green**** Sun-

screen. Green Chem 5:382–386

21. Compton DL, Goodell JR, Grall S, Evans KO, Cermak SC

(2015) Continuous, packed-bed, enzymatic bioreactor pro-

duction and stability of feruloyl soy glycerides. Ind Crop Prod

77:787–794

22. Feddern V, Yang Z, Xu X, Badiale-Furlong E, de Souza-Soares

LA (2011) Synthesis of octyl dihydrocaffeate and its transesteri-

ﬁcation with tricaprylin catalyzed by Candida antarctica lipase.

Ind Eng Chem Res 50:7183–7190

23. Ciftci D, Saldaña MDA (2012) Enzymatic synthesis of phenolic

lipids using ﬂaxseed oil and ferulic acid in supercritical carbon

dioxide media. J Supercrit Fluids 72:255–262

31. Xin J-Y, Chen L-L, Zhang Y-X, Zhang S, Xia C-G (2011)

Lipase-catalyzed transesteriﬁcation of ethyl ferulate with triolein

in solvent-free medium. Food Bioprod Process 89:457–462

32. Compton DL, Laszlo JA (2009) 1,3-Diferuloyl-sn-glycerol from

the biocatalytic transesteriﬁcation of ethyl 4-hydroxy-3-methoxy

cinnamic acid (ethyl ferulate) and soybean oil. Biotechnol Lett

31:889–896

33. Sun S, Song F, Bi Y, Yang G, Liu W (2013) Solvent-free enzy-

matic transesteriﬁcation of ethyl ferulate and monostearin:

optimized by response surface methodology. J Biotechnol

164:340–345

34. Sun S, Zhou W (2014) Enhanced enzymatic preparation of

lipophilic feruloylated lipids using distearin as feruloyl accep-

tors: optimization by response surface methodology. J Oleo Sci

63:1011–1018

24. Compton DL, Laszlo JA, Berhow MA (2006) Identiﬁcation and

quantiﬁcation of feruloylated mono-, di-, and triacylglycerols

from vegetable oils. J Am Oil Chem Soc 83:753–758

35. Laszlo JA, Compton DL, Vermillion KE (2008) Acyl migration

kinetics of vegetable oil 1,2-diacylglycerols. J Am Oil Chem Soc

85:307–312

25. Isbell TA, Lowery BA, DeKeyser SS, Winchell ML, Cermak SC

(2006) Physical properties of triglyceride estolides from lesquer-

ella and castor oils. Ind Crop Prod 23:256–263

26. AOCS (2010) Moisture Karl Fischer reagent. In: Ofﬁcial meth-

ods and recommended practices of the American Oil Chemists’

Society, 6th edn. AOCS Press, Champaign, p Ca 2e-84

27. AOCS (2010) Acid value. In: Ofﬁcial methods and recom-

mended practices of the American Oil Chemists’ Society, 6th

edn. AOCS Press, Champaign, p Cd 3d-63

28. Huang D, Ou B, Prior RL (2005) The chemistry behind antioxi-

dant capacity assays. J Agric Food Chem 53:1841–1856

29. Choo WS, Birch EJ (2009) Radical scavenging activity of lipo-

philized products from lipase-catalyzed transesteriﬁcation of tri-

olein with cinnamic and ferulic acids. Lipids 44:145–152

30. AOCS (2010) Color lovibond method. In: ofﬁcial methods and

recommended practices of the American Oil Chemists’ Society,

6th edn. AOCS Press, Champaign, p Cc 13e-92

36. Laszlo JA, Compton DL (2006) Enzymatic glycerolysis and

transesteriﬁcation of vegetable oil for enhanced production of

feruloylated glycerols. J Am Oil Chem Soc 83:765–770

37. Yang Z, Glasius M, Xu X (2012) Enzymatic transesteriﬁcation of

ethyl ferulate with ﬁsh oil and reaction optimization by response

surface methodology. Food Technol Biotechnol 50:88–97

38. Shaath NA (1997) Evolution of modern sunscreen chemicals. In:

Lowe NJ, Shaath NA, Pathak MA (eds) Suncreens: development,

evaluation, and regulatory aspects. Marcel Dekker Inc., New

York, pp 3–34

39. U.S. Federal Register (2011) Sunscreen drug products for over-

the-counter human use; ﬁnal rules and proposed rules. US Fed

Regist 76(117):35620–35665

1 3

Article Doi

Feruloylated Products from Coconut Oil and Shea Butter

DOI: 10.1007/s11746-017-2953-7

Source and publish data:

Authors:

Article abstract of DOI:10.1007/s11746-017-2953-7

Full text of DOI:10.1007/s11746-017-2953-7

Products guided by the article

R&D Labs maybe for 69790-26-7

Relevant to this article

Hot Product