Statistical design of experiments for production and purification of vanillin and aminophenols from commercial lignin

Article abstract of DOI:10.1039/d0gc01234c

Lignin is a complex polyphenolic substance that collectively represents the largest renewable source of aromatic carbon on Earth. Despite low yield and purity from depolymerized lignin, bio-derived vanillin (4-hydroxy-3-methoxybenzaldehyde) is a desirable molecule in the food/beverage and fragrance industries. To maximize vanillin yield from commercially-available softwood lignin, a series of oxidation reactions were conducted using a Box-Behnken statistical design. By varying time, temperature and O2 pressure, optimal conditions were selected using the Response Surface Method, and a maximum vanillin yield of 5.3 wtpercent was achieved. Flash chromatography of depolymerized lignin was investigated by comparing anion exchange and reversed-phase resins and fractions were characterized by GC-MS, GPC-HPLC, FT-IR and NMR. Anion exchange separations using 1M NaCl as the sole mobile phase reached a vanillin purity of 55 wt percent. Lastly, the vanillin mixture was reacted via reductive amination, and facile acid-base work up led to the formation of 4-((diethylamino)methyl)-2-methoxyphenol at 94.3 wt percent purity. The aminophenol is an intermediate in the synthesis of pharmaceutical agents, ionic liquids and benzalkonium salts.

Full text of DOI:10.1039/d0gc01234c

View Article Online
View Journal
Green
Chemistry
Cutting-edge research for a greener sustainable future
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: S. Liu, L. Das, D.
N. Blauch, C. Veronee, C. Dou, J. Gladden, N. Sun and A. M. Socha, Green Chem., 2020, DOI:
1
0.1039/D0GC01234C.
Volume 20
Number
May 2018
Pages 1919-2160
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
9
7
Green
Chemistry
Cutting-edge research for a greener sustainable future
rsc.li/greenchem
Accepted Manuscripts are published online shortly after acceptance,
before technical editing, formatting and proof reading. Using this free
service, authors can make their results available to the community, in
citable form, before we publish the edited article. We will replace this
Accepted Manuscript with the edited and formatted Advance Article as
soon as it is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes to the
text and/or graphics, which may alter content. The journal’s standard
Terms & Conditions and the Ethical guidelines still apply. In no event
shall the Royal Society of Chemistry be held responsible for any errors
or omissions in this Accepted Manuscript or any consequences arising
from the use of any information it contains.
ISSN 1463-9262
PAPER
Paul T. Anastas et al.
The Green ChemisTREE: 20 years after taking root with the 12
principles
rsc.li/greenchem

Page 1 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Statistical Design of Experiments for Purification of
Vanillin from Commercial Lignin using Ion
Exchange
1
2,3
4
1
5
Shihong Liu , Lalitendu Das , David N. Blauch , Charlie Veronee , Chang Dou , John
Gladden2,3 Ning Sun , Aaron M. Socha
5*
1*
1Department of Chemistry and Environmental Science, Queens University of Charlotte, 1900
Selwyn Avenue, Charlotte, North Carolina 28274, United States
2Joint BioEnergy Institute, 5885 Hollis Street, Emeryville, California 94608, United States
3Deparment of Biomass Science and Conversion Technology, Sandia National Laboratories,
7011 East Avenue, Livermore, California 94551, United States
4Department of Chemistry, Davidson College, 405 North Main Street, Davidson, North Carolina
28035, United States
5
Advanced Biofuels and Bioproducts Process Development Unit, Biological Systems and
Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720,
United States

Green Chemistry
Page 2 of 38
View Article Online
DOI: 10.1039/D0GC01234C
INTRODUCTION
Lignin comprises a heterogenous group of organic polymers present in cell walls of all
vascular plants. As the largest renewable source of aromatic carbon on earth, lignin also
1
represents a valuable raw material for biorefining. In nature, lignin is synthesized from three
2
,3
phenylpropanoid monomers resulting in a complex, polydisperse mixture. These monolignols
vary in methoxy substitution at the 3- and 5-positions of the phenol, and their ratios depend on plant
taxonomy.4,5 The β-O-4 ether linkage between monomers accounts for approximately 50% of the
total linkages in softwood lignin and 65% of total linkages in hardwood lignin.6,7 In order to
valorize this abundant natural product, there have been many studies that have focused on

Page 3 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
lignin depolymerization to produce low molecular weight aromatic compounds such as phenols,
phenyl alcohols and benzaldehydes.8,9
Due to their reactive chemical handles, benzaldehydes represent valuable feedstocks for
chemical elaboration. Lignin oxidation to produce vanillin (4-hydroxy-3-methoxybenzaldehyde),
has been widely studied,10-13 and the bio-derived vanillin market within the food/beverage,
fragrance, and pharmaceutical industries was approximated at $150M in 2018.14 Reported
1
5,16
vanillin yields depend on lignin type, ranging from 1-8% for kraft lignin,
1-15% for organosolv
lignin,15,17 and 7-12% for lignosulfonate.18,19 Vanillin yields from softwood lignin are typically higher
than those from hardwood lignin.12 To achieve high purity vanillin from lignin, purification steps
such as distillation and multistage crystallization are required,20,21 and these methods are not
always economically viable.12
Recently, our group has used vanillin and related derivatives to synthesize protic and aprotic
benzylammonium ionic liquids in good to excellent yields.8,22 To prepare these bio-based ionic
liquids, commercially available aromatic aldehydes (vanillin, syringaldehyde, furfural, p-
anisaldehyde) were first subjected to reductive amination.23 After additional synthetic steps,8,22

Green Chemistry
Page 4 of 38
View Article Online
DOI: 10.1039/D0GC01234C
8
the ionic liquid products demonstrated the ability to dissolve cellulose, and/or effectively
pretreat biomass for enzymatic hydrolysis to fermentable sugars. 8,22
Chromatographic methods to recover vanillin and other benzaldehydes from depolymerized
lignin have recently been reported.
For example, Gomes et al. applied adsorption
chromatography to fractionate vanillin, acetovanillone, and vanillic acid after lignin
9
depolymerization. Similarly, Mota et al. investigated the downstream processing of oxidized
industrial kraft liquor, and achieved an increased concentration of vanillin and syringaldehyde.24
Both studies relied upon SP700 resin, a porous polystyrene resin. To develop a more
sustainable route to lignin-based ionic liquids, we sought to investigate oxidation of
commercially available lignin, chromatographic purification of vanillin, and in situ conversion of
depolymerized lignin mixtures into aminophenols.
BioChoice®,
a
commercially-available pine lignin was investigated in this study.
®
Approximately 60 ton/day of BioChoice lignin is produced and sold by Domtar Paper Company
®
in Plymouth, North Carolina. Characterization of BioChoice lignin was performed by Hu et al.,
and it was found to be comparable to indulin AT, another pine kraft lignin.25 Although the
production processes for the two lignin sources differ, only subtle differences in chemical

Page 5 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
structure was observed by NMR.25 To our knowledge, the only other study of BioChoice lignin
®
was performed by Jiang et. al. showing that polydispersity was reduced after sequential solvent
26
extraction. In the latter study, lignin was not depolymerized prior to fractionation.
In this study, a series of lignin depolymerization experiments were conducted using Box-
Behnken statistical design to maximize vanillin yield.27 Optimal oxidation conditions were
selected based on the Response Surface Method.28 Vanillin was separated by extraction
followed by chromatographic fractionation. Two resins were compared for the fractionation
®
®
process - Diaion HP-20 polystyrene, and AG 1-X8 anion exchange with a molecular weight
exclusion of 1kD (Figure 1). The ion exchange method demonstrated superior purification
results for vanillin recovery; the vanillin content reached approximately 55 wt%.
Chromatographic fractions from both resin types were quantified by GC-MS and further
characterized by GPC, FT-IR and NMR. Lastly, pooled fractions from the ion exchange resin,
containing 46 wt% vanillin, were reacted using reductive amination (Scheme 1). Facile acid-
base work up resulted in the desired product, 4-((diethylamino)methyl)-2-methoxyphenol (1), at
75% yield and 94 wt% purity. In addition to their utility in making bio-based ionic liquids,
aminophenols have been reported as emergency antidotes for cyanide and hydrogen sulfide
poisoning,29,30 as well as chemical intermediates, e.g. for the synthesis of isoquinolone drugs.

Green Chemistry
Page 6 of 38
View Article Online
DOI: 10.1039/D0GC01234C
More recently, an in situ synthesis of benzazepines with antibiotic properties have recently been
performed on depolymerized lignin.31
®
Figure 1. Mechanism of vanillin separation for (A) Diaion HP-20 reversed phase polystyrene
®
resin and (B) AG 1-X8 anion exchange resin.
Scheme 1: Reductive amination of vanillin as a 46 wt% mixture obtained from anion exchange
chromatography of oxidized lignin. Acid-base workup of 4-((diethylamino)methyl)-2-
methoxyphenol product (1) allowed for phase-separations of its HCl salt and free base forms to
reach a final purity of 94 wt%. STAB = sodium triacetoxyborohydride

Page 7 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
EXPERIMENTAL SECTION
Oxidative depolymerization of lignin
®
BioChoice kraft lignin was obtained from the Domtar Paper Co, LLC (Fort Mill, SC). Before
experiments, lignin was oven dried overnight at 75 °C. The dried lignin was used directly for
depolymerization in a 600 mL pressure reactor (Parr Instrument Co.). In each experiment, 1.40
g of dried lignin was loaded into the reactor vessel along with 70 mL of 2 M NaOH solution. The
reactor was sealed, and then sparged 5x with molecular oxygen (99% purity) before charging at
specified pressure. The reactor was heated to the target temperature over 15-20 min while
stirring speed was maintained at 250 rpm. Reaction times were recorded after reaching target
temperature.
As benzaldehydes are fairly reactive species, the vanillin yield was thought to be determined
by the temperature used in the reaction, the pressure of the initially charged O , and the
2
reaction time. Seven preliminary experiments were conducted at 180-220 °C for 1-3 h with 1-2

Green Chemistry
Page 8 of 38
View Article Online
DOI: 10.1039/D0GC01234C
atm of initial O pressure, and 1.8 - 5.2 wt% vanillin yields were obtained based on the dried
2
weight of lignin (data not shown). In order to optimize vanillin yield, a response surface method
was used to determine the optimal temperature, reaction time, and O pressure. Next, a Box-
2
Behnken design (JMP Pro 14, SAS Institute, Inc., Cary, N.C.) was used to optimize these
process parameters to maximize vanillin yield and extracted lignin mass. Based on the
preliminary data, three levels for the independent variables were chosen from 180 -220 °C, 1 – 3
h, and 1 - 2 atm of initial oxygen pressure (Table 1). Two sets of 14 experiments were
performed (n = 28 in total) on independent variables of temperature (X ), reaction time (X ), and
1
2
initial O pressure (X ) (Table S1). Two center points were used to determine the experimental
2
3
error and reproducibility of the data. Low and high levels of the independent variable were
coded as −1 and +1. The experimental sequence was then randomized to minimize the effects
of uncontrolled errors. The response variable for this study were vanillin yield (mass %) and
extracted lignin (mass %). The response from the independent variables was then used to
develop an empirical model corresponding to the response variable (푌) via a second-degree
polynomial equation as given in Equation 1:
2
2
2
푌 = 훽 + 훽 푋 + 훽 푋 + 훽 푋 + 훽 푋 푋 + 훽 푋 푋 + 훽 푋 푋 + 훽 푋 + 훽 푋 + 훽 푋
33 3
(Eq. 1)
0
1
1
2
2
3
3
12
1
2
13
1
3
23
2
3
11
1
22
2

Page 9 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
where 푌 is the response, 훽 is the intercept, whereas 훽 ,훽 ,훽 and 훽 ,훽 ,훽 and 훽 ,훽 ,훽 are
0
1
2
3
12 13 23
11 22 33
linear, interaction, and quadratic effect regression coefficients, respectively. The experimental
results were fitted to a polynomial model to optimize the reaction parameters for the production
of vanillin and extracted lignin using a desirability approach.
Table 1. Experimental factors and their coded levels of independent variables in the Box-
Behnken design.
Coded variable levels
Factors
Code
-1
0
1
Temperature (°C)
Reaction time (hr)
X1
X2
X3
180
1
200
2
220
3
Initial O pressure (atm)
1
2
3
2

Green Chemistry
Page 10 of 38
View Article Online
DOI: 10.1039/D0GC01234C
Separation, quantification and characterization of vanillin from depolymerized lignin
Two chromatography methods, AG® 1-X8 anion exchange (BioRad) and Diaion® HP-20
(Supelco) were compared to purify vanillin from oxidized lignin. While HP-20 required methanol
and acetone, anion exchange was achieved with 1 M NaCl as the mobile phase. Details of
post-oxidation lignin extractions and resin purifications, including schematics can be found in the
SI Section II. Briefly, each resin type was placed into separate 350 mL fritted Buchner funnels,
and sequentially washed with 200 mL of solvent. For HP-20, the sequential solvent washes
were: 20% aqueous methanol (2 x 100 mL), 40% aqueous methanol (2 x 100 mL), 60%
aqueous methanol (2 x 100 mL), 80% aqueous methanol (2 x 100 mL), 100% methanol (2 x 100
mL) and 100% acetone (2 x 100 mL). For anion-exchange, the sequential solvent washes were
5
x 200 mL 1M NaCl followed by 200 mL MeOH. Each fraction was collected and dried in
vacuo. The dried product of each fraction was weighed and then dissolved in acetone for
quantitative GC-MS and GPC analysis. Vanillin was quantified in all fractions using GC-MS, and
the full procedure for quantitative analysis of vanillin yield is provided in SI Section I.
Spectroscopic analysis methods, including GPC, IR and NMR were also used to characterize
selected depolymerized lignin fractions and are detailed in SI Section III, respectively.

Page 11 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
In situ reductive amination of vanillin in oxidized lignin mixture
®
The purified lignin fractions obtained from AG 1-X8 anion exchange method were pooled (71
mg of mixture, 46.4 wt% vanillin; 32.9 mg (0.216 mmol) vanillin) and subjected to a synthetic
procedure modified from our previous work.22 The lignin mixture was placed in a 7 mL
scintillation vial, dissolved with 3 mL of dichloromethane (DCM), and then mixed with 250 L of
diethylamine (177 mg, 2.4 mmol). The mixture was stirred at room temperature for 30 min, and
then mixed with 160 mg (0.755 mmol) of sodium triacetoxyborohydride and 50 L (52.5 mg,
0.873 mmol) of acetic acid. The vial was then tightly sealed and heated to 40 °C for 20 h. The
reaction mixture, containing DCM solvent and desired product (4-((diethylamino)methyl)-2-
methoxyphenol), was washed with 2 × 5 mL of diluted HCl (pH 1). The DCM phase contained
unreacted acetovanillone, and was discarded. The aqueous phase contained the desired
product as its HCl salt. The acidic aqueous phase was further extracted with 2 × 5 mL of diethyl
ether, which was also discarded. The pH of the aqueous phase was adjusted to pH 8.5 with 2
M NaOH and the desired product was extracted as its free base with 6 × 5 mL of CHCl . After
3

Green Chemistry
Page 12 of 38
View Article Online
DOI: 10.1039/D0GC01234C
solvent removal in vacuo, 34 mg (0.162 mmol) of 4-((diethylamino)methyl)-2-methoxyphenol
was obtained as a light brown oil, representing a yield of 74.9% yield (Scheme 1). The extracted
1
final product was characterized by H NMR and GC-MS, and the latter allowed for a purity
calculation of 94.3 wt% (Figure S10-S13).
RESULTS AND DISCUSSION
Development of the model equation
Table S1 shows the vanillin yields from oxidative depolymerization of lignin obtained from
Box-Behnken designed experiments. The results showed vanillin yields varied in the range of
1.8 - 7.5%, which are consistent with 1-10% vanillin yields for alkaline lignin oxidation reported
in the literature.12,15,32-34 Vanillin yield is greatly affected by the lignin source and alkaline
oxidation of kraft lignin typically yields less than 10% vanillin.10,12 The extracted lignin column of
Table S1 shows the mass percentage of lignin extracted with EtOAc at pH 1.5. In a control
experiment, untreated lignin provided only 48.6 wt% of extract, and 0% of vanillin yield. The
model-recommended optimum levels for the vanillin yield were temperature at 197 ℃, reaction

Page 13 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
time for 2.4 h, and initial O pressure of 2.2 atm, corresponding to a predicted vanillin yield of
2
5.3% at the 95% confidence interval ([4.1,6.4]). Similarly, under the optimum reaction
parameters, the predicted extracted lignin yield was 77.9% at the 95% confidence interval
([70.6,85.1]). The quadratic regression models for vanillin yield and extracted lignin are shown
as Equations 2 and 3 below:
2
2
2
푌푉푎푛푖푙푙푖푛 = 5.18 ― 0.21 푋 +0.29 푋 +0.26 푋 ―0.5 푋 ―0.24 푋 ―1.26 푋 +0.64 푋 푋 +0.49 푋 푋 +0.62 푋 푋
1
2
3
1
2
3
1
2
1
3
2
3
(Eq. 2)
2
2
2
푌퐸푥푡푟푎푐푡푒푑 푙푖푔푛푖푛 = 76.8 ― 8.84 푋 +0.89 푋 ―2.94 푋 ―4.39 푋 +0.09 푋 +2.04 푋 +0.01 푋 푋 ―5.9 푋 푋 ―2.5
1
2
3
1
2
3
1
2
1 3
푋2푋3
(Eq. 3)
where, X , X , and X are temperature, reaction time, and initial O pressure, respectively. The
1
2
3
2
coefficients with one factor (X , X , X ) represent the linear effects, while the coefficients with second
1
2
3
2
2
2
order term (X , X , X ) and two-factor term (X X , X X , X X ) represent the quadratic effect and
1
2
3
1
2
1
3
2
3
interaction effects, respectively.
Another lignin depolymerization experiment was conducted under the aforementioned optimal
conditions (197 ℃, 2.4 h, and initial O pressure of 2.2 atm). The GC-MS result showed the
2
vanillin yield in the depolymerized products reached 5.3% with a lignin extraction of 77.1%.
Thus, the results aligned very well with the model’s prediction. While vanillin was the major
monophenolic compound in the product mixture, acetovanillone was also detected, (3.5% yield

Green Chemistry
Page 14 of 38
View Article Online
DOI: 10.1039/D0GC01234C
3
5
based on lignin dry weight). Acetovanillone is a known product of lignin depolymerization. It is
noteworthy that vanillic acid was not detected in the product mixture, suggesting that the
optimized conditions did not lead to over-oxidation of vanillin. Information regarding the effects
of reaction temperature, time and oxygen pressure could be gleaned from the comparison of the
model’s predicted vanillin yields and extracted lignin masses with experimentally obtained
values.
Effect of reaction parameters on vanillin yield and extracted lignin.
A model was fit to the experimental data in order to find the optimized reaction conditions to
maximize vanillin yield and estimate the significance of each parameter (temperature, time, and
O pressure) towards vanillin yield and extracted lignin mass.
2
Reaction temperature: The combined effects of temperature with reaction time and initial O2
pressure on vanillin yield are shown in Figures 2 A and B. Response surface plots illustrate that
vanillin yield increased with decreasing temperature, with highest yield obtained at 197 ℃.
Additionally, results from ANOVA (Table S2) indicate that both linear and interaction effect of
temperature with reaction time and initial O pressure were not significant (p > 0.05). Figures 3
2

Page 15 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
A and B illustrate the effect of temperature coupled with interaction of reaction time and initial O2
pressure on extracted lignin (%). Results from ANOVA (Table S3) illustrate that in the design
specified levels, temperature had a significant effect (p = 0.0002) on extracted lignin.
Additionally, the interaction effect of temperature and initial O pressure was also significant (p =
2
0.0379) while the interaction effect of temperature and time was not significant on extracted
lignin mass.
Reaction time: Figures 2 A and C illustrate the combined effects of reaction time with
temperature and initial O pressure on vanillin yield. The response surface plots demonstrate
2
that changes in reaction time does not have a significant impact on vanillin yield. Similarly, the
interaction effect of reaction time with temperature and initial O pressure were not significant (p
2
>
0.05) with respect to the vanillin yield within the design specified range. The combined effect
of reaction time with temperature and initial O pressure on extracted lignin is illustrated in
2
Figures 3 A and C. Results from ANOVA (Table S3) illustrate that in the designed specified
levels, reaction time had no significant effect on extracted lignin, which is also corroborated from
the 3D response surface plots.

Green Chemistry
Page 16 of 38
View Article Online
DOI: 10.1039/D0GC01234C
Initial O pressure: The linear and interaction effect of initial O pressure with temperature and
2
2
reaction time is shown in Figure 2 B and C. Response surface plots demonstrate that vanillin
yield increased with an increase in initial O pressure reaching its maximum at 2.2 atm with
2
subsequent reduction in yield at higher pressures. Figure 3 B and C illustrate the combined
effect of initial O pressure, with reaction time and temperature, on extracted lignin. Results
2
from ANOVA (Table S3) illustrate that linear effect of initial O pressure and interaction effect
2
with reaction time has no significant impact on extracted lignin. However, interaction effect of
initial O pressure with temperature has significant effect on extracted lignin (p = 0.0379).
2
Taken together, results from the Box-Behnken designed experiments illustrate that the
specified levels for temperature, reaction time and initial O pressure had no significant effect on
2
vanillin yield. However, temperature and interaction effect of initial O2 pressure with
temperature had significant impact on extracted lignin.

Page 17 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Figure 2. Surface response plots for vanillin yield showing the effects of A) reaction time and
temperature, B) initial O pressure and temperature and C) initial O pressure and reaction time.
2
2
Figure 3. Surface response plots for extracted lignin showing the effects of A) reaction time and
temperature, B) initial O pressure and temperature and C) initial O pressure and reaction time.
2
2

Green Chemistry
Page 18 of 38
View Article Online
DOI: 10.1039/D0GC01234C
Vanillin purification using chromatographic resin.
®
Diaion HP-20 resin is a polystyrene/divinylbenzene matrix selected to exploit π-π interactions
for aromatic compounds (Figure 1A). Table 2 shows the mass of lignin and vanillin purity in
each fraction from the HP-20 purification. Because HP-20 is aromatic reversed-phase
adsorbent resin, washing with increasing concentrations of aqueous methanol was used elute to
the vanillin analyte. As the methanol concentration increased, more bulk lignin was desorbed
from the resin, but the same trend did not occur with vanillin. Vanillin showed a trend of
increasing desorption from the HP-20 resin as the methanol concentration increased from 20-
60%. These results suggest that depolymerized lignin has a greater affinity for resin than
vanillin, likely due to greater amount of aromatic ring interaction between the resin and the
larger lignin fragments. The highest purity of vanillin (16.1 wt%) as well as largest amount of
vanillin (9.84 mg) was achieved in the 60% methanol fraction, suggesting this solvent mixture is
most suitable for the desorption of vanillin from HP-20 resins. The method showed a high
vanillin recovery (89%), indicating most vanillin can be desorbed from the resin. The 60%
methanol fraction was dried in vacuo and used for FT-IR, and NMR analysis. This fraction will
hereafter be referred as the HP-20 purified fraction.

Page 19 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Table 2. Fraction mass and vanillin purity obtained by HP-20 purification using different elution
solvents. The total mass of vanillin loaded onto the resin was 29.7 mg.
Solvent Fraction
Fraction mass (mg)
Vanillin mass (mg)
Vanillin purity (%)
2
4
6
8
1
0% MeOH
0% MeOH
0% MeOH
0% MeOH
00 MeOH
15
36
61
83
87
82
364
0.4
2.7
6.5
16.1
10.9
4.8
0.8
-
2.33
9.84
9.04
4.19
0.73
26.53
Acetone
Total
®
To improve the vanillin recovery, another type of resin separation was attempted. AG 1-X8
anion exchange resin consists of styrene divinylbenzene copolymer lattice containing
quaternary ammonium cations attached to hydroxide anions. According to the manufacturer, the
resin has a 100-fold higher selectivity for a phenolate anion as compared to a hydroxide anion.
Therefore, we hypothesized that the vanillate anion would exchange with the hydroxide anion

Green Chemistry
Page 20 of 38
View Article Online
DOI: 10.1039/D0GC01234C
and adsorb onto the resin (Figure 1B). Phenolates can then be desorbed with an excess of
another anion such as chloride, which has 22-fold greater affinity than hydroxide. Because the
resin is 8% crosslinked, it also serves as a size exclusion resin for the adsorption of molecules
smaller than 1 kD.
The results of anion exchange vanillin purification are shown in Table 3. Because the pK of
a
36
the phenolic proton of vanillin is 7.4, it exists primarily in the anionic (e.g. vanillate) form under
the pH 8 conditions used. Thus, vanillate can be adsorbed onto the resin and subsequent
desorbed with 1 M NaCl. As shown in Table 3, vanillin was gradually desorbed from the resin
while washing with 1 M NaCl solution. The first 200 mL NaCl solution resulted in desorption of a
large amount of lignin from the resin, albeit only 26% pure vanillin. After the first 200 mL elution,
the fraction mass decreased gradually.
However, higher purity vanillin (45%-55%) was
obtained in NaCl fractions 2-5. The final methanol wash, which disrupts non-ionic interactions
(e.g. Van der Waals forces) with the resin, resulted in the lowest vanillin mass and purity.
These results suggest that the initial 200 mL wash with 1M NaCl caused a faster desorption of
phenolic lignin other than vanillin, and in subsequent washes the desorption rates of impurities
and vanillin were similar to one another. Because of the decreased amount of lignin remaining
on the resin, the mass transfer of impurities and vanillin decreased with additional washing

Page 21 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
steps. The methanol wash was proven to be unnecessary due to its low vanillin yield. The anion
exchange method also showed a total vanillin recovery of 82%. Further washing with NaCl
might increase the total recovery of vanillin, however, it would be difficult to desorb all the
vanillin on the resin due to decreased mass transfer. The second fraction of NaCl washing,
which had a vanillin purity of 54.8%, was dried under vacuum and used for FT-IR, and NMR
analysis. This fraction will hereafter be referred to as the anion exchange purified fraction.
Anion exchange chromatography was found to be more effective than reversed phase (HP-
20), and thus we decided to substitute artificial sea water as the mobile phase. Unfortunately,
artificial seawater was slightly inferior to 1 M NaCl (Table S5) and both the dried mass of each
fraction and vanillin purity decreased. This result is likely due to the lower ionic strength of sea
water (total salt concentration < 0.6 M), and therefore concentration of, or salt addition to,
seawater could potentially improve its solvent properties for anion exchange chromatography of
lignin.
Table 3. Fraction mass and vanillin purity in the fractions obtained by anion exchange
purification. The total mass of vanillin loaded onto the resin was 37 mg.

Green Chemistry
Page 22 of 38
View Article Online
DOI: 10.1039/D0GC01234C
Fraction
Mass (mg)
Vanillin mass (mg)
Vanillin purity (%)
NaCl wash fraction 1
NaCl wash fraction 2
NaCl wash fraction 3
NaCl wash fraction 4
NaCl wash fraction 5
MeOH wash
31
15
16
7
8.07
8.22
7.13
3.20
3.36
0.30
30.28
26.0
54.8
44.6
45.7
56.0
3.0
6
10
85
Total
-
GPC analysis
GPC estimates molecular weight (MW) distribution of a sample mixture, with compounds of
higher MW eluting faster and thus displaying shorter retention times. To obtain a relationship
between the retention time and MW, polystyrene and vanillin standards were analyzed (Figure
S4, Table S4). The estimated MW shown in Table S4 is based solely on the MWs of
polystyrene standards, and does not exactly represent the average MW of lignin due to
unknown differences in polymer branching. However, GPC data using polystyrene standards
provides valuable information on MW changes during depolymerization and fractionation.37

Page 23 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Figure 4 shows the results of GPC analysis for the fractioned lignin obtained by the two
purification methods. It can be seen that the original kraft lignin mainly consists of compounds
with MWs > 50 kDa and compounds with MWs in the range of 152 – 9000 Da (Figure 4A).
However, after the oxidative depolymerization process, the MWs in the depolymerized lignin
became smaller and all peaks have shifted to the right. A reduced signal was observed in the
region of MWs > 50 kDa, while the signal in the region of 152 – 2500 Da increased significantly.
Particularly, a peak was detected at approximately 18.5 min in the depolymerized lignin
samples, which matched the retention time of the vanillin standard (Figure S4B).

Green Chemistry
Page 24 of 38
View Article Online
DOI: 10.1039/D0GC01234C

Page 25 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Figure 4. GPC analysis at 280 nm for (A) untreated kraft lignin, oxidatively depolymerized lignin,
®
and lignin residue that was not retained by AG 1-X8 resin purification. (B) fractions generated
by HP-20 resin purification, and (C) fractions generated during by anion exchange resin
purification.
The chromatograms of the purified fractions were consistent with the results from GC-MS
analysis. Fractions with a higher vanillin purity also showed a stronger signal in the low MW
region. For the HP-20 purification, the strongest signal at 18.5 min was obtained from the 60%
methanol fraction (Figure 4B). Further increasing the concentration of methanol caused the
greater desorption of high MW compounds, leading to lower purity vanillin. Because high MW
compounds were detected in all HP-20 factions, coupling size exclusion separation or
membrane separation with HP-20 might be needed to increase vanillin purity.
Anion exchange resin effectively adsorbed the smaller MW compounds. In Figure 4A, the
chromatogram of the lignin residue fraction not retained by anion exchange resin shows a
significantly smaller peak in the region of 152 – 2500 Da. All lignin fractions generated from
NaCl elution had lower amounts of high MW compounds (Figure 4C). This is likely due to the

Green Chemistry
Page 26 of 38
View Article Online
DOI: 10.1039/D0GC01234C
size exclusion capacity of the anion exchange resin, leading to the higher purity vanillin
fractions.
FT-IR analysis. The FT-IR spectra of the resin purified fractions, as well as the untreated kraft
lignin, depolymerized lignin, and vanillin standard are shown in Figure 5. A clear structural
change resulted from alkaline oxidation. IR spectra for the depolymerized lignin and resin
purified fractions show increased C=O stretching of carbonyl/carboxyl group,38 C=O stretching
of aldehyde group,39,40 and C-C aryl-aldehyde stretching,41 at frequencies of 1700, 1663, and
-1
-1
1
170 cm respectively. On the other hand, signals at 1080 and 1030 cm showed a decrease in
absorbance for the depolymerized lignin and resin purified fractions, indicating decreased C-O
aryl-ether stretching.39,41-43 This evidence supports β-O-4 bond breaking and subsequent
formation of carbonyl compounds during oxidation.
In particular, the anion exchange purified fraction showed an increase in frequencies at 1663
-1
-1
and 632 cm , and a significant decrease in frequencies at 1700 cm compared to other
samples. This result suggests the anion exchange purified fraction was enriched in vanillin as
compared to sundry carbonyl containing lignin fragments, consistent with GC-MS analysis.

Page 27 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
Figure 5. FT-IR spectra of untreated lignin, depolymerized lignin, and resin purified fractions.
1
13
NMR analysis. H and C NMR analysis of the specified HP-20 and anion exchange purified
1
fractions showed clear presence of vanillin (Figure S5-S8). In the H spectra, singlets for 1 H (δ
=
9.83 ppm) and 3 H (δ = 3.94 ppm) were observed for aldehyde and methoxy resonances,
respectively, in fractions from both resin types. Three protons were observed on the aromatic
ring at chemical shifts of δ = 7.01, 7.50 and 7.56 ppm, consistent with the complete assignment
of vanillin. In 13C spectra, the chemical shift of the aldehyde carbon atom was observed at δ =
191.1 ppm. The carbon atom of the methoxy group was present at δ = 56.3 ppm. Another four
carbon atoms on the aromatic rings were observed at δ = 110.9, 115.9, 127.0 and 130.7 ppm.

Green Chemistry
Page 28 of 38
View Article Online
DOI: 10.1039/D0GC01234C
The quaternary carbons on the 3- and 4-positions of the aromatic ring were detected in the HP-
0 purified fraction at δ = 149.0 and 153.6 ppm, respectively. In general, the H and 13C NMR
1
2
spectra of the fractions were consistent with vanillin spectra reported in the literature,44
confirming the presence of vanillin in fractions from both types of resins. When comparing both
1H and 13C spectra for the individual resin fractions, it was observed that the HP-20 purified
fraction showed more impurities than the anion exchange purified fraction, corroborating the
GC, GPC and IR analysis.
1
13
2
D-HSQC analysis was conducted on untreated and depolymerized lignin, and shows H- C
correlations used to identify structural changes resulting from lignin depolymerization. The
results are presented in Figure S9. Figures S9A and S9C show the aliphatic regions of the lignin
samples. The spectra clearly demonstrate the presence of linkages commonly found in
1
13
softwood lignin, including β-O-4 (δ H/ C = 4.79/71.2 ppm, 4.32/84.0 ppm, 3.19-3.77/60.0 ppm), β-5
1
13
1
13
(
δ H/ C = 5.49/86.6 ppm, 3.49/53.2 ppm, 3.22-3.82/62.8 ppm), and β-β (δ H/ C = 4.63/84.9 ppm,
3
.08/53.6 ppm, 3.75/71.0 ppm, 4.16/71.0 ppm) linkages.1,45-47 These bonds were not found in the
depolymerized lignin (Figure S9C), indicating deconstruction during oxidation. As shown in
Figures S9B and S9D (aromatic regions), the untreated lignin showed strong aromatic signals of
guaiacyl and p-hydroxyphenyl units, however these signals were greatly reduced in the

Page 29 of 38
Green Chemistry
View Article Online
DOI: 10.1039/D0GC01234C
depolymerized lignin. In addition, certain sp2 C-H couplings are only observed in the
depolymerized lignin, diagnostic of vanillin, acetovanillone, p-hydroxybenzoic acid, p-coumaric
acid and ferulic acid.46 Although the HSQC results show the presences of p-hydroxybenzoic
acid, p-coumaric acid and ferulic acid, these compounds were not detected by GC-MS,
suggesting these compounds might not exist as monophenolic compounds in any appreciable
concentrations.
In situ aminophenol synthesis and purification. Because benzaldehydes undergo reductive
amination to form benzylamines, it was hypothesized that benzaldehyde containing lignin
mixtures could be further purified by reductive amination followed by a simple acid/base work
up.23 To test our hypothesis, reductive amination of a mixture containing vanillin was conducted
with excess diethylamine and sodium triacetoxyborohydride (Scheme 1) by modifying previously
reported methods.22 As shown in Figure 6, HPLC shows that reductive amination of a
depolymerized lignin mixture containing 46 wt% vanillin, followed by acid/base work up, is an
effective method for synthesizing and separating 4-((diethylamino)methyl)-2-methoxyphenol
from major impurities, such as acetovanillone, and minor impurities such as oxidized lignin