Towards a greener synthesis of (S)-3-aminobutanoic acid: Process development and environmental assessment

Article abstract of DOI:10.1039/c002721a

An improved, greener process for the enantioselective chemoenzymatic synthesis of (S)-3-aminobutanoic acid has been developed. Reaction steps comprise an initial aza-Michael addition starting from cheap prochiral compounds, subsequent enzymatic resolution via aminolysis using commercially available Candida antarctica lipase B in a solvent-free one-pot process, hydrolysis of the resulting ester and removal of the N-benzyl moiety via hydrogenation. After isolation, the desired (S)-3-aminobutanoic acid was obtained in an overall yield of 28% and with an excellent enantiomeric excess of 99% ee. Notably, this reaction sequence does not require column chromatography with organic solvents and only one purification step of an intermediate is needed. The environmental impact of this optimized process has been evaluated and an E-factor of 41 has been calculated for the overall process. A comparative assessment with the previous process was done via mass balancing using the E-factor, the selectivity index S^-1 as well as an SHE assessment.

Full text of DOI:10.1039/c002721a

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/greenchem | Green Chemistry
Towards a greener synthesis of (S)-3-aminobutanoic acid: process
development and environmental assessment†
Markus Weiß,a Tobias Brinkmannb and Harald Gro¨ger*a
Received 11th February 2010, Accepted 28th May 2010
DOI: 10.1039/c002721a
An improved, greener process for the enantioselective chemoenzymatic synthesis of
(S)-3-aminobutanoic acid has been developed. Reaction steps comprise an initial aza-Michael
addition starting from cheap prochiral compounds, subsequent enzymatic resolution via
aminolysis using commercially available Candida antarctica lipase B in a solvent-free one-pot
process, hydrolysis of the resulting ester and removal of the N-benzyl moiety via hydrogenation.
After isolation, the desired (S)-3-aminobutanoic acid was obtained in an overall yield of 28% and
with an excellent enantiomeric excess of 99% ee. Notably, this reaction sequence does not require
column chromatography with organic solvents and only one purification step of an intermediate is
needed. The environmental impact of this optimized process has been evaluated and an E-factor
of 41 has been calculated for the overall process. A comparative assessment with the previous
process was done via mass balancing using the E-factor, the selectivity index S-1 as well as an SHE
assessment.
Introduction
Enantiomerically pure b-amino acids are attractive key building
blocks for the synthesis of pharmaceuticals.1 The tendency to
use chiral b-amino acids in drug synthesis has increased over
recent years as well as the search for new synthetic methods
for their enantioselective preparation.2 For the stereoselective
synthesis of aromatic b-amino acids a range of efficient methods
exist,3,4 which in part are already applied on industrial scale.
In contrast, efficient stereoselective methodologies for aliphatic,
in particular short-chain b-amino acids, are known to a lesser
extent.5 In addition, numerous synthetic steps and tedious
work-up operations (in particular column chromatographical
separation) often limits the applicability and scalability of
several routes. In order to make those multi-step syntheses more
attractive for large-scale applications it would be desirable to:
(i) have a limited number of synthetic steps, (ii) use cheap
and easily available starting materials, (iii) avoid isolation
and (complex) purification of intermediates, and (iv) replace
column chromatography (requiring large amount of organic
solvent) by greener work-up operations. Recently we developed a
chemoenzymatic approach towards short-chain b-amino acids
exemplified for (S)-3-aminobutanoic acid ((S)-5, Scheme 1),6
which already fulfils the criteria (i) and (ii) for a sustainable and
potentially technical application.
Scheme 1 Original lab synthesis of (S)-1 according to ref. 6: scope and
limitations.
chromatographical purification step leading to a high demand
of organic solvent (in contrast to criteria (iv)). As a consequence,
this results in a high E-factor of 359 for the overall process.
In the following, we report the process development towards a
greener synthesis of (S)-3-aminobutanoic acid (as hydrochloride
salt), which overcomes the former limitations. The product (S)-
3-aminobutanoic acid hydrochloride, (S)-9 (Scheme 2), has been
chosen as a representative of chiral short-chain b-amino acids.
The process development has been influenced by environmental
considerations according to the twelve green chemistry princi-
ples by Anastas and Warner,7 addressing issues such as the use
of non-toxic chemicals, minimization of organic solvents and
reduction of isolation steps. In addition we carried out an envi-
ronmental assessment of the optimized process, in which critical
However, this process still requires isolation and purification
of intermediates (in contrast to criteria (iii)) and a column
aDepartment of Chemistry and Pharmacy, University of
Erlangen-Nuremberg, Henkestrasse 42, 91054, Erlangen, Germany.
E-mail: harald.groeger@chemie.uni-erlangen.de
bifu GmbH Hamburg, Große Bergstraße 219, 22767, Hamburg, Germany
† Electronic supplementary information (ESI) available: Material and
ecotoxicity data used for the environmental assessment, and data
resulting from the environmental assessment for the optimized and
original process. See DOI: 10.1039/c002721a
1580 | Green Chem., 2010, 12, 1580–1588
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
process steps of the original process have been replaced. Thus,
this work also contributes to increase the (still) limited number
of examples in the field of small chiral molecules describing
a process development in combination with an environmental
assessment of the process. For the assessment of the optimized
process (and the original process for comparison) we used the
software programmes EATOS8 and Umberto/Sabento.9 Both
software programmes EATOS (primarily developed for chemical
syntheses) and Umberto/Sabento (Sabento: primarily devel-
oped for biotechnological processes) rely on mass balancing,
and allow the calculation of, e.g., the E-factor according to a
concept by Sheldon.10,11 We used these software programmes,
which are based on slightly different assessment algorithms,
since they complement each other. EATOS allows an easy and
fast calculation of the E-factor and environmental impact of the
separate reaction steps within a sequence. Umberto/Sabento
allows a graphical breakdown of the whole process by the
illustration of all mass flows in a Sankey diagram.12
whole process. In addition, the conditions of each reaction step
also provide potential for optimization. Thus, in spite of general
advantages this original process results in a disadvantageous
mass balance and a non-satisfactory high E-factor of 359, which
makes a scale up of this process less attractive (for details of the
mass balancing, see Electronic Supporting Information (ESI)†
and Table 4). This high E-factor is mainly due to the use of
large quantities of organic solvents, which were necessary for
the chromatographic purification step.
To overcome these limitations we envisioned a process with
improved sustainability, which fulfils the following criteria:
(i) multiple usage (recycling) of the environmentally friendly
biocatalyst, (ii) reduced number of purification steps (ideally
only isolation of the final product (S)-5 in this multi-step
synthesis), and (iii) prevention of waste organic solvents by
avoiding the use of column chromatography. If possible, organic
solvents in reactions should also be substituted with water as
an environmentally friendly solvent with exceptional properties
(safe, non-flammable, non-toxic etc.).16
The concept and realization of such a new, greener process
is depicted in Scheme 2 as a flow sheet, and discussed in
the following. At the beginning of the process the previously
developed solvent-free one-pot reaction is carried out (steps 1
and 2). After thermal aza-Michael addition of benzylamine 2
to (E)-ethyl but-2-enoate, 1, a biocatalytic resolution through
aminolysis is initiated after addition of Candida antarctica lipase
B (CAL-B). This one-pot process resulted in the formation of the
desired b-amino ester (S)-3 with 99% ee besides N-benzyl (R)-3-
(benzylamino)butanamide, (R)-6, (E)-N-benzyl but-2-enamide,
7, and ethanol as further products. Subsequently, the immo-
bilized enzyme CAL-B was simply filtered and washed with
MTBE, followed by an additional washing step with saturated
sodium hydrogen carbonate solution.17 Since (S)-3 and (R)-6
(in the MTBE filtrate) are difficult to separate from each other
by simple, non-chromatographical procedures, we decided to
selectively hydrolyse the ester moiety of (S)-3 in the presence
of the b-amino amide (R)-6 and enamide 7. This hydrolysis
of (S)-3 (step 3), leading to the desired water-soluble sodium
salt of (S)-3-(benzylamino)butanoic acid, (S)-8, had to be fully
chemoselective for the ester moiety, since a partial cleavage
of amide (R)-6 would lead to the opposite enantiomer (R)-8,
thus decreasing the optical purity of the resulting sodium (S)-3-
(benzylamino)butanoate, (S)-8.
Thus, a screening of different reaction conditions for such a
chemoselective hydrolysis was conducted to explore the opti-
mum conditions. As a substrate, rac-3 was used for this purpose
(instead of (S)-3 for practical reasons). Besides a chemoselective
course of the reactions we focused on a minimization of the
required amount of sodium hydroxide (used under mild reaction
conditions) and the feasibility of conducting the hydrolysis in
MTBE, since this solvent was used in the previous unit operation
step (filtration of the immobilized enzyme CAL-B).
The results are summarized in Table 1. It turned out that both
the concentration of the NaOH solution and the equivalents
of NaOH (referring to the amount of rac-3) are crucial for
the outcome of the hydrolysis. In terms of minimized inorganic
waste and amount of water as well as high conversion entry 3
showed the optimal reaction conditions. Next, we studied the
chemoselective course of the reaction when using a mixture
Results and discussion
Process development towards a sustainable synthesis
Our recently developed lab scale process for the preparation of
the enantiomerically pure (S)-b-amino acid (S)-5 (Scheme 1),
represented the starting point for the process development
work. This process is a multi-step synthesis with a solvent-
free, chemoenzymatic one-pot reaction as the initial step.
This key step led to the desired intermediate (S)-3 in high
enantiomeric excess of 99% ee after separation of the immo-
bilized enzyme through filtration, washing with methyl tert-
butyl ether (MTBE), and purification via column chromatog-
raphy (Scheme 1). The ester moiety in (S)-3 was subsequently
hydrolysed under harsh acidic conditions under formation of the
hydrochloride salt of the acid (S)-4. After deprotection of the
N-benzyl protecting group via hydrogenation and subsequent
ion-exchange chromatography, the desired (S)-3-aminobutanoic
acid, (S)-5, was obtained in 25% overall yield and enantiomeri-
cally pure form (> 99% ee).
The use of cheap and readily available (prochiral) substrates
(1 and 2) is advantageous, as is using a commercially available
biocatalyst, limiting the number of synthetic steps, using solvent-
free and highly enantioselective initial chemoenzymatic synthe-
sis of (S)-3, and the satisfactory overall yield of 25% (taking
into account the presence of a resolution step within a four-step
synthesis). Since organic solvents, which are used as reaction
media and for product isolation and purification (extraction,
column chromatography), typically are the main contributors
of “auxiliary waste” produced during chemical processes,13 an
improved multi-step process should exclude organic solvents.14,15
The avoidance of organic solvents in reaction steps has been
addressed in our original work by the development of a solvent-
free protocol for the initial key step.
However, critical steps of this lab process with respect
to sustainability are in particular (i) the large amounts of
waste, which result from organic solvents used for the column
chromatographical step (as a typical lab purification procedure),
(ii) single use of the biocatalyst, and (iii) isolation of the resulting
intermediates and product after each reaction stage during the
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1580–1588 | 1581
©

View Article Online
Scheme 2 Improved, greener synthesis of (S)-3-aminobutanoic acid hydrochloride, (S)-9.
Table 1 Screening of conditions for hydrolysis of rac-3
(S)-3 (>99% ee) and 1.5 equivalents of amide (R)-6 (60% ee) as
starting material under basic conditions (1 M NaOH) led to a
highly chemoselective formation of the sodium salt (S)-8 with
>99% ee (Scheme 3), thus indicating - as desired - hydrolysis
of only the ester (S)-3 (and not amide (R)-4). It should also be
noted that in contrast acidic hydrolysis (1.0 M HCl) of such a
mixture consisting of ester (S)-3 (>99% ee) and amide (R)-6
(60% ee) led to a dramatic decrease in the enantiomeric excess
of the resulting acid compared with the enantiomeric excess of
the starting material (S)-3 (data not shown).
With such a chemoselective hydrolysis in hand, next we
applied these conditions for the hydrolysis of product (S)-3
(99% ee), which results from step 2 of the chemoenzymatic
one-pot process as a solution in MTBE (Scheme 2). After
treatment of this organic layer containing (S)-3 with a solution
of NaOH (1.0 M), the resulting hydrolytic reaction led to a
highly chemoselective formation of the desired intermediate (S)-
8 with unchanged 99% ee (Scheme 2, step 3). After completion
of hydrolysis (indicated by TLC) the sodium salt of (S)-3-
(benzylamino)-butanoate, (S)-8, was obtained via simple sep-
aration of the aqueous layer from the hydrophobic components.
By means of an anion-exchanger N-benzylated (S)-b-amino
acid hydrochloride (S)-4 was isolated from the aqueous layer
from step 3 in pure form (Scheme 2, step 4). Subsequent
Solvent
Entrya Base
ratiob
Eq.c pH Cleavaged E-Factore
1
2
3
4
5
6
NaOH (1.0M)
1.9/1
0.4/1
0.8/1
1.5/1
0.4/1
5.0 14
2.2 14
2.0 14
2.0 14
1.1 14
+
~
+
+
~
3.9
6.0
2.0
2.9
1.4
14.3
NaOH (2.0M)
NaOH (1.0M)
NaOH (0.5M)
NaOH (1.0M)
KCl/NaOH (0.2M) 10/1
5.2 12.6 +
a Typical reaction conditions: rac-3 (42.0 mg, 0.19 mmol), 0.5 mL
MTBE, 18 h stirring at room temperature. b Solvent ratio: ratio of
the volume of aqueous solution of the base to the volume of MTBE.
c Equivalents = eq. refers to the amount of rac-3. d Denotation of
symbols: “+” complete cleavage, “~” partial cleavage, the reaction was
monitored by TLC (ethylacetate:2-PrOH), 95 : 5 (v/v), 0.2% diethy-
lamine. e E = waste (kg)/product (kg); the E-factor was calculated based
on the assumption of complete hydrolysis of rac-3 and a recycling rate
of the aqueous phase of 90%.
of ester (S)-3 (>99% ee) and amide (R)-6 (60% ee). We
were pleased to find that stirring a mixture of isolated ester
1582 | Green Chem., 2010, 12, 1580–1588
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
of reactivity. After 5 reaction cycles, conversion drops to 52%
(Table 2, entry 5) compared with 60.6% after the same reaction
time when using fresh enzyme (entry 1).
As a main reason for the decreased enzyme activity we assume
the loss of CAL-B due to mechanical abrasion during the
reaction (caused by the magnetic stirrer used in our experiments)
and subsequent filtration. Potential process options for the
future in order to circumvent this critical issue are the use
of stirrers other than magnetic stirrers for such batch-type
operations.
Environmental assessment of the optimized process
Scheme 3 Chemoselective hydrolysis of (S)-3 (control experiment).
After the successful preparation of (S)-3-amino butanoic acid
hydrochloride, (S)-9 (28% overall yield, 99% ee), we started the
environmental assessment of the developed process with the
software programmes EATOS and Umberto/Sabento. Since not
all material data was provided by the material databases of both
programmes, a collection of such ecological information was
done first (for material data, see ESI†). For this purpose we
used the free-access databases of BGIA gestis, BGIA report,18
eChemportal,19 ecotox data20 and the material data safety sheets
(MSDS) of the commercial chemicals suppliers Acros,21 Sigma-
Aldrich22 and Merck.23
Based on the collection of the material data, the results
of the chemical reactions of the overall process (shown in
Scheme 2) and the process assumptions given below, a material
flow network was modelled. This fully specified material flow
model is shown in Scheme 4 as a Sankey diagram. In the Sankey
diagram the intermediates are shown after each reaction step for
clarity. In general there are three types of places “P”, which are
mapped by colored circles: green input places (materials enter
the material network here), yellow connection places (materials
are passed from one transition to another) and red output
places (materials leave the system boundary here). The blue
boxes denote “transition sites”, which are sites where changes in
the mass composition occur through chemical reactions, work
up procedures and recycling steps. This enables an overview of
the mass flow in the process, and an analysis of what happens
to each material. In contrast to the reaction equations of
Scheme 2 one can immediately recognize mass intensive parts of
a reaction sequence, since the line width of every material arrow
is proportional to the mass of each material used. The dashed
lined boxes show areas of special interest, i.e. side-products and
recycling areas.
hydrogenation of (S)-4 over palladium in 1 M HCl then gave
the desired (S)-3-aminoutanoic acid hydrochloride, (S)-9, with
excellent enantiomeric excess of 99% ee (Scheme 2, step 5), and
a good overall yield of 28% (after five process steps starting from
inexpensive prochiral compounds and including one resolution
step). Notably, all the synthetic steps can be combined in an
elegant manner without the need for a complex purification unit
operation for either intermediates or product. The selected
downstream-processing steps are based on simple unit operation
steps such as extraction and ion-exchange chromatography,
thus avoiding the solvent-consuming (and waste-producing)
column chromatography as “standard work up protocol” for
lab procedures.
In order to further improve the overall process economy we
studied the re-usage of the immobilized enzyme CAL-B for the
one-pot reaction. The conditions of the recycling experiments
and the results are shown in Table 2. After each one-pot reaction
under formation of (S)-3 the immobilized CAL-B was filtered,
washed with MTBE, and dried prior to re-use. The one-pot
reaction starting from prochiral compounds 1 and 2 has been
repeated four times with the recycled enzyme immobilisate.
The results shown in Table 2 indicate that enzyme recycling is
possible in principle, however at the expense of a continuous loss
Table 2 Recycling of CAL-B in the one-pot reaction
For the mass balancing the following assumptions have been
used as boundary conditions: (i) 90% recycling rate for pure
organic solvents,24 aqueous solutions (HCl (1 M), NaOH (1
M)) and water,25 (ii) 85% recycling rate for CAL-B,26 (iii) full
reusability for palladium catalyst and ion exchanger,27 and (iv)
85% recycling rate for mixture of organic solvents.
The results of the mass balancing visualized in the Sankey
diagram in Scheme 4 are discussed in the following in more
detail. An overview about two types of indices for the input
materials is shown in Fig. 1. Therein, the blue bar is related to
the used amount (mass) of each material (in kg) theoretically
required for the hypothetical preparation of one kg of final
product (S)-9, whereas the red bar shows the fraction of
the (substance-specific) potential environmental impact for the
Entrya Reaction Conv. Step 1 (%)b Conv. Step 2 (%)b ee (S)-3c
1
2
3
4
5
1
2
3
4
5
95.5
93.0
94.0
93.5
93.0
60.6
56.0
55.0
55.0
52.0
99
96
93
91
88
a Typical reaction conditions (E)-ethyl but-2-enoate, 1 (114.2 mg, 1.0
mmol), benzylamine, 2 (235.5 mg, 2.2 mmol), lipase CAL-B (50 mg
mmol-1). b Conversions were determined by means of 1H-NMR spec-
troscopy. c Determination of enantiomeric excess via chiral HPLC.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1580–1588 | 1583
©

View Article Online
Scheme 4 Sankey diagram (material network) of the Umberto/Sabento modelling according to Scheme 2; mass flows are indicated as follows: light
blue = HCl (1 M), dark blue = water, grey = NaOH (1 M), purple = auxiliary materials, yellow = organic solvents, pink = enzyme, green = (organic)
material transfer, red = (organic) waste transfer.
PEI, whereas its mass proportion (4.5%) is relatively small
compared to the other materials. In contrast 1 M HCl (HCl
used for work up, 25.4%) represents the second largest mass
proportion, together with the second largest substance-specific
PEI. This underlines the importance of the reduction of organic
solvents from syntheses and work up procedures. The substance-
specific PEI of the substrates benzylamine, 2(14.4% of total mass
input) and (E)-ethyl but-2-enoate, 1 (6.9% of total mass input),
was calculated to be nearly equal within this process, although
the required mass of benzylamine is significantly higher.
In addition, the different impact categories of the process
have been studied and evaluated. The results are visualized in
Fig. 2. Therein the different impact categories that arise from
Fig. 1 Selected mass indices (in kg) and substance-specific PEI indices
(in PEI/kg) of input materials according to the Sankey diagram in
Scheme 4 (in the text section the resulting percentaged mass indices (in
%), calculated by dividing the substance specific mass index by the total
sum of mass indices, are discussed).
input side (PEI(input)) of each material after safety health
environment weighing (SHE). Accordingly, water (28.2%), 1 M
HCl (HCl used for work-up, 25.4%) and benzylamine, 2 (14.4%),
determine the three main parts of the total input material mass
flow. In addition it should be noted that only 68% of the total
material demand is needed for product isolation.
Since from a sustainability perspective the impact of a
material, in particular waste material, is not only related to its
required mass but also to its hazard properties, it is important to
examine the PEI of each material besides its amount. According
Fig. 2 Impact categories of the input materials according to the Sankey
to Fig. 1 methyl tert-butyl ether is identified to have the highest
diagram in Scheme 4.
1584 | Green Chem., 2010, 12, 1580–1588
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Table 3 Mass balancea of the process according to the Sankey diagram
the substances used within the process (input materials) are
displayed. Accordingly, the major impact potentials with respect
to the input materials are assigned to thermal risks (34%) and
acute toxicity (44%).
in Scheme 4
Input
Quantity/kg Output
0.0708 Hydrogen
Quantity/kg
Furthermore, an environmental assessment of the output
materials (which represents waste material) was carried out in
analogous manner. The resulting mass and PEI indices of the
output materials are displayed in Fig. 3. As can be seen in
Fig. 3, a 1 M solution of HCl (25.5%), (R)-3-(benzylamino)-
butanamide (R)-4 (9.8%), a 2M solution of NaOH (7.3%) and
methyl tert-butyl ether (4.5%) represent the main proportion of
the total mass output (waste). The substance-specific PEI values
follow the same order. It is also noteworthy that small quantities
of organic solvents like toluene (2% of total mass output) and
methyl tert-butyl ether (4.5%) show a relatively high substance-
specific PEI value compared to the aqueous mass flows (1 M
HCl and 1 M NaOH), which would require much more mass to
cause a similar substance-specific PEI value. In addition, 66%
of the total output waste volume is of aqueous origin, whereas
only 4.5% are caused by the only organic solvent MTBE.
Hydrogen
NaOH (0.1 M) 0.2531
NaOH (1 M)
NaOH (2 M)
MTBE
CAL-B
Benzylamine, 2 5.9620
(E)-Ethyl
but-2-enoate, 1
NaHCO3
HCl (1 M)
Water
0.0569
0.2531
1.7341
3.0506
1.8734
0.1898
1.8101
6.3807
NaOH (0.1 M)
NaOH (1 M)
NaOH (2 M)
MTBE
2.0632
3.9236
1.8734
0.1898
CAL-B
Benzylamine, 2
Coupling-/Side
products
Inorganic salts
HCl (1 M)
Water
2.8860
0.3645
12.1645
11.7721
2.8328
10.5696
11.7721
(S)-3-Aminobutanoic 1.0
acid hydrochloride,
(S)-9
Total
41.5232
Total
41.5232
41
E-factorb
a Mass balance calculated for the hypothetical production of 1 kg (S)-
3-aminobutanoic acid hydrochloride, (S)-9. For a more detailed mass
balance, see ESI†. b E-factor = (amount of total waste)[kg]/(amount of
product)[kg]; the E-factor is given above without decimal places (E =
41); the calculated E-factor with decimal places is 40.5232.
Accordingly, about two thirds of the total risks of the produced
waste output can be assigned to thermal risks (38%) and acute
toxicity (26%). Among further risks there are also risks for
eutrophication (11%) and chronical toxicity (11%).
In addition to this type of EHS assessment, which considers
the properties of the substances, a comparative assessment only
based on mass balances can also be done. Here an important
indication for the sustainability of a chemical process is the so-
called E-factor which describes the amount of waste (in kg)
produced per kg of product. Thus, the E-factor, which has
been introduced by Sheldon,10,11 reflects the performance of a
preparative process considering, e.g., stoichiometry of reagents,
product yield, amount of catalyst and solvents. Thus, for the
calculation of the E-value a detailed mass balance of the process
is required. Such a mass balance of the overall process for
the synthesis of (S)-9 (according to Scheme 4) is given in
Table 3. Based on this mass balance an E-factor of 41 was
calculated. For comparison, typical E-factors in the industrial
synthesis of pharmaceuticals are in the range of 25–100. Thus,
the achieved E-factor of 41 can be considered to be in a satisfying
range, especially when taking into account the relatively low
molar mass of the target molecule (S)-3-aminobutanoic acid
hydrochloride, (S)-9, and the presence of a resolution within the
five-step synthesis.
Fig. 3 Selected mass indices (in kg) and substance-specific PEI indices
(in PEI/kg) of output materials according to the Sankey diagram in
Scheme 4 (in the text section the resulting percentaged mass indices (in
%), calculated by dividing the substance specific mass index by the total
sum of mass indices, are discussed).
The overall risk of the process with respect to the output
materials (waste materials) has also been classified into the
different impact categories. The results are displayed in Fig. 4.
In multi-step syntheses a further interesting aspect is the
evaluation of which steps contribute to the consumption of
substances and formation of waste, and to what extent. Thus, for
our process we calculated the distribution of the mass volume of
the used resources (represented by selectivity S-1; S-1 = (amount
of resources [kg])/(amount of product) [kg]) and the waste
(represented by the E-factor) on the individual process steps
1 to 5 (according to Schemes 2 and 4). The results of this
evaluation (S-1, E-factor referring to each step) are shown in
Fig. 5. The separate steps contribute to the overall amount of
Fig. 4 Impact categories of the output materials according to the
Sankey diagram in Scheme 4.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1580–1588 | 1585
©

View Article Online
Table 4 Comparison of key data of the optimized process according to
operations for intermediates in the optimized process also turned
out to be beneficial and contributed to a reduction of waste. This
is underlined by the significantly improved E-factor (reaction)
of the new, optimized process.
Schemes 2, 4, and the original process according to Scheme 1 and ref. 6
New optimized
process (this work) (according to ref. 6)
Original processd
Entry
The advantages of the optimized process (Schemes 2, 4) in
comparison with the original process (Scheme 1) in terms of
sustainability is also underlined by the analysis of the potential
environmental impact PEI. The PEI (total) summarizes all
substance-specific PEIs (for selected data, see Fig. 1 and 3), and
is therefore a key indicator for the SHE assessment of a process.
The PEI (input) and PEI (output) are referring to all substance-
specific PEIs from the input and output materials, respectively.
In comparison to the original process,6 the developed optimized
process is about 62 times more benign in terms of environmental,
health and safety aspects (Table 4).
E-factora (reaction)
13
41
45
359
366
6580
6666
6632
E-factora (incl. isolation)
Selectivityb S-1 (incl. isolation) 36
PEIc (input)
PEIc (output)
PEIc (total)
86
121
107
a E-factor = (amount of total waste[kg])/(amount of product[kg].
b Selectivity S-1 = (amount of resources[kg]/(amount of product[kg].
c PEI = potential environmental impact = PEI/(kg product), calculated
with Umberto/Sabento. d For details of the calculation for the original
process according to ref. 6, see ESI†.
Conclusion
In conclusion, we reported a successful process development
towards an efficient approach with improved sustainability to
the hydrochloride of (S)-3-aminobutanoic acid, (S)-9. This
compound is a representative of the pharmaceutically interest-
ing product class of short-chain aliphatic b-amino acids and
corresponding salts. The target product (S)-9 has been obtained
within a five-step chemoenzymatic synthesis in an overall yield of
28% and with excellent enantiomeric excess of 99% ee. Notably,
column chromatographical work-up steps with organic solvents
were not required, and isolations and purifications are based
on simple unit operations such as extraction or ion-exchange
chromatography. In addition, the developed synthesis of (S)-
9 was analysed via mass balancing and SHE assessment with
the software programmes EATOS and Umberto/Sabento. The
analysis of key environmental indicators such as, e.g., S-1, E-
factor, PEI, revealed a significantly improved sustainability of
the new optimized process in comparison to the original process.
A reduction of the E-factor by more than eight times was
achieved in addition to a reduction of the PEI (total) by more
than 60 times. This fact also indicates a significant diminishment
of hazardous substances for the optimized process.
Fig. 5 Distribution of the mass volume of resources (S-1) and waste
(E) on the different process steps 1–5 according to the Sankey diagram
in Scheme 4 (calculated with EATOS).
used substances (resources) and produced waste as follows: The
chemoenzymatic one-pot synthesis of (S)-3 (steps 1 and 2) as
a key step only demands 28.1% of used substances and only
causes 23.9% of formed waste. Whereas subsequent hydrolysis
(step 3) also contributes to a minor extent to the demand of
resources (6.7%) and formation of waste (7.0%), desalination
(step 4) plays a major role in this field with 45.6% of demand of
resources and 47.5% of formed waste. The hydrogenation (step
5) as a final step contributes to 19.6% of demand of resources
and 21.6% of formed waste. The improvements from the process
development becomes evident when comparing the optimized
process (according to Scheme 2 and 4) with the original process
(shown in Scheme 1). In Table 4, the key data for both the
optimized process (described in this work) and our previously
published original process6 are shown (for details see ESI†).
Comparing the calculated E-factors (including isolation) of both
processes a reduction of more than eight times (41 versus 359)
could be achieved with the new optimized process.
Experimental
General
1H NMR (13C NMR) spectra were recorded at room temperature
on a Bruker Avance 300 or JEOL JNM GX 400 spectrometer
operating at 300 MHz or 400 MHz. Chemical shifts (d) are
reported in parts per million relative to tetramethylsilane (TMS)
(d 0.00). NMR raw data was analyzed with the program MestreC
4.9.9.9. HPLC spectra were recorded on a Jasco HPLC system
with a PU-1587 Intelligent Prep. Pump and a LC-Net II and
ADC device at room temperature. As stationary phase the
R
following Daicelꢀ columns were used: OD, OJ-H, AD-H, IA
and IB. Mass spectra (MALDI) were recorded on a Shimadzu
Biotech Axima Confidence spectrometer. Mass spectra (EI) were
recorded on a Micromass Zabspec spectrometer with electron
ionization. Elemental analysis were measured on a EA 1119
CHNS, CE instrument. IR spectra were recorded on a Thermo
scientific Nicolet IR-100 FT-IR spectrometer. The absorption
The same tendency has been found for the resource demand
(represented by S-1). The decrease of isolation process unit
1586 | Green Chem., 2010, 12, 1580–1588
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
-1
˜
is given in wavelength n; [cm ]. The optical rotation of chiral
compounds was measured on a Perkin Elmer instruments
polarimeter (model 341) with a Na-lamp (wavelenght l = 589
nm). Preparative (flash) column chromatography was performed
over Merck Silica gel 60 (230–400 mesh) as stationary phase.
Thin layer chromatography (TLC) was performed on silica
Preparation of (S)-3-(benzylamino)butanoic acid hydrochloride,
(S)-4, and (S)-3-aminobutanoic acid hydrochloride, (S)-9,
according to Schemes 2 and 4 (optimized process)
Steps 1 and 2. In a 10 mL round bottom flask with
glass stopper (E)-ethyl but-2-enoate (1, 20 mmol, 2.28 g) and
benzylamine (2, 44 mmol, 4.81 mL) was heated at 60 ◦C and
stirred for 30 h. After completion of the thermal Michael
addition the conversion (step 1) was determined (step 1: 93.0%
total conversion, 91.0% product related conversion) by means
R
gel TLC cards (Alugrammꢀ SIL G, layer thickness 0.20 mm,
Fluka) with fluorescens indicator. Chemicals and materials were
purchased from commercial suppliers Acros and Sigma-Aldrich
unless otherwise stated. The used enzyme Candida antarctica
lipase B in immobilized form is a commercial product from
1
of H-NMR spectroscopy (5 ml sample). Then, lipase Candida
antarctica B (CAL-B) was added (50 mg mmol-1, 1.0 g) and the
R
Novozymes A/S (product name: Novozymꢀ435).
◦
reaction was stirred for another 16 h at 60 C. The conversion
(step 2) was determined (step 2: 61.2% conversion) by means of
1H-NMR spectroscopy (5 ml sample). The enantiomeric excess
of ethyl (S)-3-(benzylamino) butanoate, (S)-3, was determined
by chiral HPLC chromatography. Enantiomeric excess: 99%
ee. Retention time (HPLC): 11.80 min ((S)-3), 13.38 min
((R)-3) (Daicel column IB, hexanes:2-PrOH, 99 : 1 (v/v), 0.1%
diethylamine, flow 1.0 ml min-1, 254 nm)). The reaction mixture
was separated from the immobilized enzyme by filtration,
and subsequently washed with methyl tert-butyl ether (MTBE,
20 mL). The resulting organic solution was washed three times
with a solution of aqueous saturated NaHCO3 (1 mL). The
organic layer containing the product (S)-3 was used directly for
the next step. The assignment of the absolute configuration was
done by comparison with data in ref. 6.
Procedure for hydrolysis reactions of rac-3 according to Table 1.
In a 5 mL round bottom flask rac-3 (0.19 mmol, 42 mg)
was diluted in 0.5 mL MTBE. After addition of 1.1–5.0 eq
of a base the reaction mixture was stirred for 18 h at room
temperature. Reaction progress was monitored by TLC control
(ethyl acetate/2-PrOH, 95 : 5 (v/v), 0.2% diethylamine).
Procedure for the control experiment for chemoselective hydrol-
ysis of (S)-3 according to Scheme 3. In a 5 mL round bottom
flask (S)-3 (99% ee) (0.19 mmol, 42 mg) and (R)-4 (60% ee)
(0.28 mmol, 80 mg) was diluted in 0.5 mL MTBE. After addition
of 0.76 mL NaOH (2 M) the reaction mixture was stirred for 18 h.
The completion of the reaction was indicated by TLC control
(ethyl acetate/2-PrOH, 95 : 5 (v/v), 0.2% diethylamine). The
aqueous phase was extracted four times with 1 mL ethyl acetate.
The remaining water was removed at 60 ◦C under reduced
pressure. After cooling to room temperature thionyl chloride
(0.38 mmol, 45 mg) was added. The reaction mixture was cooled
to -10 ◦C (ice/NaCl-bath) and ethanol was added dropwise.
After stirring for 30 min at -10 ◦C, the reaction mixture was
stirred for 1 h at room temperature. After heating to 60 ◦C
the reaction mixture was stirred again for 1 h, diluted in 5 mL
NaHCO3 and extracted four times with 2 mL dichloromethane.
After drying over MgSO4 the solvent was evaporated under
reduced pressure to yield ethyl (S)-(3-benzylamino) butanoate,
(S)-3, as a yellowish oil. Yield: 18 mg (43%). Enantiomeric
excess: >99% ee. Retention time (HPLC): 11.80 min ((S)-3),
13.38 min ((R)-3) (Daicel column IB, hexanes:2-PrOH, 99 : 1
(v/v), 0.1% diethylamine, flow 1.0 ml min-1, 254 nm).
Step 3. An aqueous solution of NaOH (1 M, 14.4 mL)
was added to the organic layer resulting from step 2 in a
10 mL round bottom flask. The resulting reaction mixture was
stirred at room temperature for 16 h. The reaction progress was
monitored according to the consumption of (S)-3 by TLC (ethyl
acetate:2-PrOH, 95 : 5 (v/v), 0.1% diethylamine, Rf 0.60). After
completion of the reaction, the aqueous layer was separated in
a separatory funnel. Finally the organic phase was extracted
two times with an aqueous solution of NaOH (0.1 M, 1 mL).
Step 4, (S)-3-(benzylamino)butanoic acid hydrochloride (S)-
4. To the aqueous layer resulting from step 3 was added ion
exchanger Merck-III (14.4 g). The resulting mixture was stirred
for one hour. After washing with water (40 mL) the (S)-3-
(benzylamino)butanoic acid hydrochloride, (S)-4, was eluted
with an aqueous solution of HCl (1 M, 70 mL). Pure (S)-3-
(benzylamino)butanoic acid hydrochloride, (S)-4, was yielded
as a white solid after evaporation of the aqueous hydrochloric
acid under reduced pressure at 60 ◦C. Yield: 1.30 g (29%, related
to the amount of (E)-ethyl but-2-enoate, 1). The enantiomeric
excess of (S)-4 was determined by chiral HPLC chromatography
of its ethyl ester, (S)-3 (after derivatization of (S)-4 with
thionylchloride according to the procedure for the control
experiment for chemoselective hydrolysis of (s)-3 according to
Scheme 3). Enantiomeric excess: 99% ee. Optical rotation [a]D25
+15.7 (c 1.0 in H2O). Elemental analysis: Found: C, 57.09;
H, 7.03; N, 6.13%. C11H16ClNO2 requires C, 57.52; H, 7.02;
N, 6.10%. IR nmax/cm-1: 3061 (CHarom.), 2911 (CHaliph.), 2735
Procedure for recycling experiments with CAL-B in the synthe-
sis of (S)-3 according to Table 2. In a 5 mL round bottom flask
with glass stopper (E)-ethyl but-2-enoate (1, 2 mmol, 228 mg),
and benzylamine (2, 4.4 mmol, 480 ml) was heated at 60 ◦C
and stirred for 30 h. After completion of the thermal Michael
addition the conversion (step 1) was determined by means of
1H-NMR spectroscopy (5 ml sample). Then, lipase from Candida
antarctica B (CAL-B) was added (50 mg mmol-1, 100 mg) and the
◦
reaction was stirred for another 18 h at 60 C. The conversion
1
(step 2) was determined by means of H-NMR spectroscopy,
the enantiomeric excess was determined by chiral HPLC (IB
column, hexanes:2-PrOH, 99 : 1 (v/v), 0.1% diethylamine) from
a 5 ml sample of the reaction mixture. The enzyme was washed
eight times with 1 mL MTBE. After evaporation of the solvent,
the enzyme was re-used again.
(CHaliph.), 1723 (C O), 1456 (C C), 1173 (C–C). 1H-NMR
(400 MHz, D2O): 1.42 (3H, d, JH,H = 6.7 Hz, Me), 2.78 (1H,
dd, J = 6.6 Hz, J = 17.4 Hz, CHCH2CO), 2.85 (1H, dd, J =
17.4 Hz, J = 6.2 Hz, CHCH2CO), 3.70 (1H, m, 1H, CH), 4.24,
=
=
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1580–1588 | 1587
©

View Article Online
4.31 (2H, 2d, JH,H = 13.2 Hz, CH2), 7.47–7.50 (5H, m, CH). 13C-
NMR (100 MHz, D2O): 16.08 (CH3), 36.87 (CHCH2COOH),
L. G. Sevillano, C. J. Lock, P. D. Tiffin, N. Tremayne and S. Winter,
Tetrahedron Lett., 2000, 41, 2679–2681; (g) H. Gro¨ger, O. May, H.
Hu¨sken, S. Georgeon, K. Drauz and K. Landfester, Angew. Chem.,
2006, 118, 1676–1679, (Angew. Chem., Int. Ed., 2006, 45, 1645–1648);
(h) H. Gro¨ger, H. Trauthwein, S. Buchholz, K. Drauz, C. Sacherer,
S. Godfrin and H. Werner, Org. Biomol. Chem., 2004, 2, 1977–1978;
(i) Reviews: ref. 2.
=
48.89 (CHCH2COOH), 50.78 (NH2-CH2-C6), 129.62 (C C),
=
=
=
129.98 (C C), 130.02 (C C), 130.89 (C C), 171.60 (-COOH).
MS (MALDI): m/z = 194 (M+).
Step 5, (S)-3-aminobutanoic acid hydrochloride, (S)-9. The
product resulting from step 4, (S)-4 (1.5 mmol, 290 mg), was
dissolved in 3 mL HCl (1M) in a Fischer–Porter bottle, followed
by the addition of Pd(C) (~10%, 87 mg). The bottle is evacuated
and flushed with inert gas for three times. After final evacuation
the bottle was filled with (65 psi corresponding to 0.44 MPa)
of hydrogen. The reaction mixture was heated to 65–70 ◦C and
stirred for 22 h. Then, the reaction mixture was separated from
the Pd(C) catalyst via filtration and eluted with water (15 mL).
The pure product (S)-3-aminobutanoic acid hydrochloride, (S)-
9, was yielded as a white solid after evaporation of the solvent
at 60 ◦C under reduced pressure. Yield: 209 mg (99% related to
the amount of (S)-3-(benzylamino)butanoic acid hydrochloride,
(S)-4, 28% related to the amount of (E)-ethyl but-2-enoate, 1,
5 For selected efficient enzymatic enantioselective syntheses of
aliphatic b-amino acids, see: (a) R. Stu¨rmer, K. Dietrich, W. Siegel,
US Pat., 60636615, 2000; (b) Ref. 4b; (c) Reviews: ref. 2.
6 M. Weiß and H. Gro¨ger, Synlett, 2009, 8, 1251–1254.
7 P. T. Anastas and J. C. Warner, Green Chemistry: Theory and Practice,
Oxford University Press, 1998, p. 30.
8 EATOS: environmental assessment tool for organic syntheses; this
software has been developed in the Metzger group, see: M. Eissen
and J. O. Metzger, Chem.–Eur. J., 2002, 8(16), 3580–3585.
9 (a) This study has been performedwiththe software Umberto (version
5.5), which was complemented with data sets and ecological assess-
ment tools from the software Sabento (version 1.2). These software
programmes have been developed by ifu Hamburg GmbH, see:
http://www.ifu.com; (b) For the EHS assessment the methodology
of Heinzle was chosen, see: E. Heinzle, final report for the project
AZ0312752A, German Federal Ministry of Education and Research;
(c) The further developed methodology as implemented in Sabento is
described in: T. Brinkmann, H. Rubbeling, P. Fro¨hlich, M. Katzberg
and M. Bertau, Chem. Eng. Technol., 2010, 33, 618.
3
1H-NMR (400 MHz, D2O): 1.34 (d, J = 6.7 Hz, 3H, Me),
10 R. A. Sheldon, Chem. Ind. (London), 1992, 6, 903–906.
11 R. A. Sheldon, Green Chem., 2007, 9, 1273–1283.
2.73 (dd, J = 7.4 Hz, J = 17.5 Hz, 1H, CHCH2CO), 2.78 (dd,
J = 5.7 Hz, J = 17.5 Hz, 1H, CHCH2CO), 3.71–3.78 (m, 1H,
CH3CHCH2). 13C-NMR (100 MHz, CDCl3): 17.91 (C-4), 37.95
(C-3), 44.61 (C-2), 174.38 (C-1). The spectroscopic data are in
accordance with those reported in the literature.28 Enantiomeric
excess: 99% ee (determined by optical rotation of the free (S)-3-
amino butanoic acid and comparison with literature data).29
12 For an overview and comparison of ecological assessment tools, see:
F. Eiden and A. Schmid, transcript, 2008, 14, 29.
13 B. M. Trost, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 13197–
13202.
14 J. O. Metzger, Angew. Chem., 1998, 110, 3145–3148, (Angew. Chem.,
Int. Ed., 1998, 37, 2975–2978).
15 K. Tanaka, Chem. Rev., 2000, 100, 1025–1074.
16 I. T. Horvath, Green Chem., 2008, 10, 1024–1028.
17 To remove traces of (R)-3-(benzylamino)butanoic acid (presumably
formed via hydrolysis of rac-3 with residual water on immobilized
CAL-B), the organic layer was extracted two times with an aqueous
solution of NaOH (0.1 M, 1 mL).
18 http://www.dguv.de/bgia/de/gestis/stoffdb/index.jsp (checked
29.01.2010).
19 http://webnet3.oecd.org/echemportal (checked 29.01.2010).
20 http://cfpub.epa.gov/ecotox (checked 29.01.2010).
21 http://www.acros.com (checked 29.01.2010).
22 http://www.sigmaaldrich.com (checked 29.01.2010).
23 http://www.merck-chemicals.de (checked 29.01.2010).
24 M. Eissen, G. Geisler, B. Bu¨hler, C. Fischer, K. Hungerbu¨hler, A.
Schmid, E. M. Carreira, in A. Lapkin, D. Constable, (Ed.), Green
Chemistry Metrics: Measuring and Monitoring Sustainable Processes,
Wiley, Chichester UK, 2008, 200-227.
25 Although being aware that according to the concept of Sheldon (see
ref. 10 and 11) water is typically not considered in the calculation of
the E-value, in this work the amount of water from aqueous solutions
has been considered in the calculation of the E-value by assuming
a recycling rate of 90% of these solutions (which results in a higher
E-value compared to the E-value without considering water from
aqueous solutions).
26 The regeneration of the enzyme was tested over five cycles according
to Table 2. From these experiments a recovery rate of 85% has been
calculated.
27 Reusability of the ion-exchanger requires regeneration of the ion
exchanger after usage. For the regeneration of the ion exchanger,
we assumed an amount of 17.28 mL of a solution of NaOH (2 M).
Calculation, assuming for the ion exchanger 1.2 val per L a density
of 1 g L-1: Capacity of the ion exchanger: 1.2 mmol g-1; Capacity of
14.4 g ion exchanger: 14.4 ¥ 1.2 mmol g-1 =17.28 mmol; Minimal vol-
ume of NaOH (aq., 2 M) for regeneration: 17.28 mL/(1mmol mL-1)
= 17.28 mL. As, for all other pure organic and aqueous solutions a
recovery rate of 90% is assumed.
Acknowledgements
We thank ifu Hamburg GmbH and Evonik Degussa GmbH
for generous support. A grant for M.W. by the Deutsche Bun-
desstiftung Umwelt (DBU) within the scholarship programme
“Nachhaltige Bioprozesse” (“Sustainable Bioprocesses”) is
gratefully acknowledged. We also thank Dr M. Eissen for
generous help and personal communications.
Notes and references
1 E. Juaristi, (Ed.), Enantioselective Synthesis of b-Amino Acids, Wiley-
VCH, New York, 1997.
2 For reviews see: (a) M. Liu and M. P. Sibi, Tetrahedron, 2002, 58,
7991–8035; (b) A. Liljeblad and L. T. Kanerva, Tetrahedron, 2006,
62, 5831–5854.
3 For selected chemocatalytic examples, see: (a) Y. Hsiao, N. R. Rivera,
T. Rosner, S. W. Krska, E. Njolito, F. Wang, Y. Sun, J. D. Armstrong,
E. J. J. Grabowski, R. D. Tillyer, F. Spindler and C. Malan, J. Am.
Chem. Soc., 2004, 126, 9918–9919; (b) A. Berkessel, F. Cleemann and
S. Mukherjee, Angew. Chem., 2005, 117, 7632–7635, (Angew. Chem.,
Int. Ed., 2005, 44, 7466–7469); (c) A. M. Clausen, B. Dziadul, K. L.
Cappuccio, M. Kaba, C. Starbuck, Y. Hsiao and T. M. Dowling, Org.
Process Res. Dev., 2006, 10, 723–726; (d) M. Sibi and K. Itoh, J. Am.
Chem. Soc., 2007, 129, 8064–8065.
4 For selected biocatalytic examples, see: (a) S. G. Cohen and S. Y. We-
instein, J. Am. Chem. Soc., 1964, 86, 725–728; (b) V. A. Soloshonok,
V. K. Svedas, V. P. Kukhar, A. G. Kirilenko, A. V. Rybakova, V. A.
Solodenko, N. A. Fokina, O. V. Kogut, I. Y. Gulaev, E. V. Kozlova,
I. P. Shishkina and S. V. Galushko, Synlett, 1993, 339–341; (c) M.
Prashad, D. Har, O. Repic, T. J. Blacklock and P. Giannousis,
Tetrahedron: Asymmetry, 1998, 9, 2133–2136; (d) G. Cardillo, L.
Gentilucci, A. Tolomelli and C. Tomasini, J. Org. Chem., 1998, 63,
2351–2353; (e) S. Katayama, N. Ae and R. Nagata, Tetrahedron:
Asymmetry, 1998, 9, 4295–4299; (f) S. J. Faulconbridge, K. E. Holt,
28 M. Nejman, A. Sliwinska and A. Zwierzak, Tetrahedron, 2005, 61,
8536–8541.
29 S. G. Davies, A. W. Mulvaney, A. J. Russell and A. D. Smith,
Tetrahedron: Asymmetry, 2007, 18, 1554–1566.
1588 | Green Chem., 2010, 12, 1580–1588
This journal is
The Royal Society of Chemistry 2010
©