Recovery of ammonia in the dipeptide manufacturing processes

Article abstract of DOI:10.1021/op000094x

An example of an improvement in recovering ammonia in a dipeptide manufacturing process is described. The synthetic method, which makes use of the ammonolysis reaction, has been studied and found to produce dipeptides of satisfactory quality in high yield on a large scale. However, the treatment of unreacted ammonia in the ammonolysis reaction caused a reduction in the productivity and increased the production cost during actual manufacture. Therefore, a method to recover the unreacted ammonia has been investigated through simulations and trial runs using model solutions. Consequently, the modified process provided an improvement in the productivity and cost savings. In addition, the recovered ammonia could possibly be used for recycling. It was verified in a lab experiment that the reused ammonia did not lower the quality of the dipeptide.

Full text of DOI:10.1021/op000094x

Organic Process Research & Development 2001, 5, 132−135
Recovery of Ammonia in the Dipeptide Manufacturing Processes
Satoshi Kato,* Takahiro Sano, and Toru Sugaya
Sakai Research Laboratories, Kyowa Hakko Kogyo Co., Ltd., 1-1-53, Takasu-cho, Sakai, Osaka, 590-8554 Japan
Abstract:
An example of an improvement in recovering ammonia in a
dipeptide manufacturing process is described. The synthetic
method, which makes use of the ammonolysis reaction, has been
studied and found to produce dipeptides of satisfactory quality
in high yield on a large scale. However, the treatment of
unreacted ammonia in the ammonolysis reaction caused a
reduction in the productivity and increased the production cost
during actual manufacture. Therefore, a method to recover the
unreacted ammonia has been investigated through simulations
and trial runs using model solutions. Consequently, the modified
Figure 1. Synthetic routes of dipeptides 3a and 3b.
process provided an improvement in the productivity and cost
savings. In addition, the recovered ammonia could possibly be
used for recycling. It was verified in a lab experiment that the
reused ammonia did not lower the quality of the dipeptide.
Figure 2. Ammonolysis reaction in the GlyGln 3a process.
The mechanism for the ammonia wastewater generation
Introduction
is as follows: After the reaction, the reaction mixture was
The synthetic production of the dipeptides, glycyl-L-
continuously concentrated in a vacuum evaporator, and then
glutamine (3a; GlyGln) and L-alanyl-L-glutamine (3b; Ala-
the unreacted ammonia was evaporated and dissolved in the
Gln), has already been studied.1 The established process of
circulating water of the ejector pump. In the pump, freshwater
3a and 3b, which was slightly different in terms of reaction
was added to the circulating water to maintain the vacuum,
conditions and work-up procedures, consisted of the reaction
and the overflow liquid from the water pit was discarded as
of R-halogenated acid with thionyl chloride and the Schot-
ten-Baumann reaction with L-glutamine followed by an
ammonolysis reaction (Figure 1). This method was selected
wastewater. The ammonia concentration of the wastewater
was about 4000 ppm. However, it had to be controlled under
200 ppm, according to the plant bylaws which were based
as the best for dipeptide manufacturing from the viewpoint
on the allowable nitrogen concentration per the city regula-
of quality and yield. Therefore, the production was success-
tion of 240 ppm. Thus, the wastewater had to be diluted with
fully scaled-up.1 However, there was a problem about the
freshwater, and also stored until the treatment was finished.
treatment of wastewater during the actual production. In the
This treatment process required unexpected time and money
ammonolysis reaction, an excess of ammonia was required
(fee for putting wasting water in city sewer) and affected
to reduce the byproduct formation, and the unreacted
the productivity and the total production cost of the GlyGln
ammonia generated the wastewater. Its treatment resulted
in a decrease in productivity and extra cost. Therefore, we
process.
started to investigate how to recover the unreacted ammonia
Simulation
by simulations and trial runs in the plant. In this paper, the
To recover the unreacted ammonia, a scheme to install a
following is an example of the improvement mainly focused
condenser between the evaporator and the ejector pump was
on the GlyGln process, but a similar improvement will be
considered. Thus, the unreacted ammonia was expected to
applicable to the AlaGln process as well.
be recovered as a condensed liquid from the condenser during
Problem of Ammonia Wastewater. The reagents of the
the operation of concentrating the reaction mixture after the
ammonolysis reaction in the GlyGln process are 28 wt %
ammonolysis reaction. As for the condenser, the use of an
aqueous ammonia and ammonium bicarbonate (Figure 2).
idle condenser for another production area was desirable.
Feasibility studies of this scheme were then carried out by
simulations.
Simulation Method. The components, handled in the
simulations, were water and ammonia. In the simulations,
the nonvolatile components such as GlyGln were ignored.
Ammonium bicarbonate was replaced with ammonia, be-
The total ammonia source is 14 mol equiv to 2a. It is the
optimized value by the experiments in the tradeoff between
the yield and the cost of ammonia. If the ammonia mol equiv
is reduced, byproducts will increase.
* Corresponding author. Telephone: +81(722)23-5545. Fax: +81(722)27-
7214. E-mail: satosi.kato@kyowa.co.jp.
(1) Sano, T.; Sugaya, T.; Inoue, K.; Mizutaki, S.; Ono, Y.; Kasai, M. Org.
Process Res. DeVel. 2000, 4, 147.
cause it was found to mostly decompose and vaporize in
132
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development
10.1021/op000094x CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 01/06/2001

Figure 3. Simulation result: optimal condition to recover ammonia.
Table 1. Result Summary
the same way as ammonia under heated conditions during
the evaporation (35-55 °C).
ammonia concentrationa [wt %]
Aspen Plus (product of Aspen Technology, Inc.) was used
as the simulation tool, and built-in physical property methods
were used.2 However, the VLE (vapor liquid equilibrium)
of the ammonia-water system was too complicated to be
expressed by one model for a wide composition range. As a
result, two types of methods were properly used, depending
on the conditions. That is, the VLE models in the evaporator
and the condenser were different. To evaluate which model
was more appropriate, the previous experimental data3 were
referenced.
feed residual distillate wastewater recovery [%]
simulation 15
trial run 1d 14.4
trial run 2e 14.3
production 12.5
1
23
b
99
99
72
70
g
2.4
6.0
7.3
21.5
20.1
15
<0.02c
<0.02c
<0.02c
0.4
reff
17
13
g
a Total concentration of ammonia and ammonium ion. b Not simulated. c Only
confirmed to meet with the plant bylaws (<0.02 wt %). d Feed solution was
aqueous ammonia. e Feed solution was aqueous ammonia with ammonium
bicarbonate. f In the production before improvement. g No corresponding data
because of the process difference
Simulation Results. The operation of concentrating the
reaction mixture was simulated to investigate the best
operational conditions, which should achieve the following
results:
the key variables was the operating pressure, that is, the
pressure in the evaporator and the condenser, as the pressure
drop between them could be assumed negligible. High
pressure causes a high temperature in the evaporator and
decomposition of GlyGln, and low pressure leads to a low
boiling point of the vapor and the leakage of ammonia vapor
which cannot be condensed. After all, the appropriate
pressure could not be obtained under any conditions which
met all of the constraints described above, if the reaction
mixture was directly fed to the evaporator.
As there were some ways considered to provide the
appropriate pressure, a method to dilute the reaction mixture
with water before concentration was selected from the
viewpoint of cost and GMP, because there were no changes
affecting the reaction. The amount of water to dilute the
reaction mixture after the reaction and the optimal pressure
were determined from the simulations. Finally, the result is
shown in Figure 3; the ammonia concentration in the feed
liquid is 15 wt %, the optimal pressure is 150 mmHg, the
recovered ammonia as 23 wt % aqueous ammonia, and the
heat transfer area of the condenser, which might be diverted,
is large enough.
(1) Total Condensation of the Vapor. It was necessary to
minimize the ammonia concentration in the wastewater from
the ejector pump, so that the water could be directly disposed
without any treatment. Thus, the vapor from the evaporator
had to be cooled to under the boiling point and totally
condensed to minimize any leakage into the pump.
(2) Maximizing the Ammonia Concentration of the
Condensed Aqueous Ammonia. As the use of the recovered
ammonia had not been determined, there were no target
values for the ammonia concentration of the condensed
liquid. However, as high a concentration as possible was
desired to widen the use of the recovered ammonia and to
reduce its volume.
During the simulation, there were some constraints to be
followed besides achieving the above two objectives. The
critical process variables under constraint were the heating
temperature, the coolant temperature, and the GlyGln
concentration of the liquid flow from the evaporator. The
heating temperature, that is, the vapor and liquid temperature
in the evaporator, had to be kept under 55 °C due to the
instability of GlyGln. The lowest temperature of the available
coolant was -8 °C. If the concentration of GlyGln in the
liquid flow from the evaporator is higher than the solubility,
GlyGln might crystallize in the pipeline. Thus, there was a
limit in the degree of concentration. There were no solutions
that met all of the requirements based on the results from a
preliminary simulation. However, it was found that one of
Results in the Plant
Results of the Trial Runs. As the simulation results were
considered feasible, some trial runs were carried out in the
plant. In the trial runs, model solutions of the reaction mixture
were used. The model solutions were prepared with aqueous
ammonia, NH4HCO3 and water. Thus, the GlyGln reaction
product was not included in them.
The results are summarized in Table 1. The trials were
divided into two types depending on whether NH4HCO3 was
included (Table 1, “trial run 1”) or not (“trial run 2”).
In the trial when NH4HCO3 was not included, the model
solution was equal to the aqueous ammonia, and the handled
(2) The built-in methods “ELECNRTL” and “NRTL-RK” were used. While
both methods basically consist of the NRTL activity coefficient model for
the liquid phase and the Redlich-Kwong equation of state for the vapor
phase, “ELECNRTL” is an enhanced method to handle electrolytes.
(3) Sherwood, T. K. Ind. Eng. Chem. 1925, 17, 745. The experimental solubility
data of ammonia in water vs partial pressure of ammonia are described.
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
133

components were the same as in the simulations. Thus, the
result of this trial run was in quite good agreement with the
simulation result.
The result of the trial including NH4HCO3 was also in
good agreement with the simulation result except for the
residual concentration of ammonia (to be precise, ammonia
and ammonium ion). The ammonia concentration of the
liquid in the evaporator was higher than that of the simula-
tion. The difference was considered as the pH became smaller
when NH4HCO3 was included, and more ammonia was
ionized which remained in the liquid phase. However, an
accurate estimation of that concentration was not very
important, because the first objective was to recover ammonia
as a high concentration of aqueous ammonia and to reduce
the leakage of ammonia vapor into the pump. In any case,
20-25 wt % aqueous ammonia was condensed, and it was
confirmed that the leakage of ammonia vapor into the pump
was so low that the ammonia concentration in the wastewater
from the ejector pump could be low enough to meet the
regulation for actual production.
Results of the Actual Production. After the trial runs,
the operating conditions of concentrating the reaction mixture
in the GlyGln manufacturing process was modified. One of
the production results is shown in Table 1 (“production”).
While the simulated optimal feed concentration of ammonia
in Figure 3 was 15 wt %, that in the actual productions was
12-13 wt %, for example, 12.5 wt % in Table 1. The set
value of the feed ammonia concentration was lowered for
the following reasons:
Figure 4. Optimal condition to enrich the recovered ammonia.
Figure 5. Prospect for recycling: 60% of the ammonia source
can be recycled.
recovered ammonia still must be disposed of eventually.
Therefore, the beneficial reuse of recovered ammonia has
been desired. Recycling does not significantly contribute to
the cost reduction because ammonia is not as expensive as
some other ingredients, but it is environmentally desirable.
The possibility of recycling has been examined.
To be recycled, it is necessary to enrich the recovered
ammonia up to a minimum of 25 wt %. At first, we tried to
reconcentrate the recovered ammonia during the actual
production facilities. The operating condition was decided
on the basis of the results of the simulation. In this case,
there are no constraints on the temperature of the evaporator
as GlyGln is not contained. As the operating condition, which
was decided by simulation, and shown in Figure 4, the trial
run was operated at atmospheric pressure. The result of the
trial run was in good agreement with the simulation results,
and 25 wt % aqueous ammonia was obtained. It was a
possible sample for recycled ammonia. Using this sample,4
a lab experiment was carried out according to the reported
procedures.1 It was verified that the quality of synthesized
GlyGln using the sample met the specification for GlyGln.
Thus, recycling was considered to be feasible from the
viewpoint of its quality.
Figure 5 shows the prospect for the recycling derived from
the current process and the result of the trial. As a result,
60% of the source of ammonia can be recycled, and 33% is
lost. The lost ammonia mainly remains in the evaporation
residue. This is difficult to evaporate as it is almost ionized.
However, there are some ways being considered to decrease
the percentage, for example, the addition of acid, the use of
a multi-stage distillation tower, and so on.
• It was possible that the result in the actual production
might be different from that in the trial runs as the model
solution did not include solutes such as GlyGln.
• The feed concentration of 15 wt % was the upper limit
concentration to achieve the total condensation of ammonia
vapor, and the lower side was steadier.
• In the actual productions, a steadier process was
desirable, and the frequent change of the condition in the
process should be avoided from the viewpoint of GMP.
Therefore, the amount of added water for the dilution of
the reaction mixture was set to a larger value than that
required to adjust the feed concentration to 15 wt %. In each
later GlyGln manufacturing batch, the concentration proce-
dure has been operated under the same conditions. The
concentration of the wastewater from the pump was low
enough to meet the regulation, and it can be directly disposed.
As it became unnecessary to keep and treat the wastewater,
the productivity increased, and cost for the sewage charges
was reduced.
As a result, about 70% of the charged ammonia has been
recovered in about 15 wt % aqueous ammonia for each batch.
The concentration of the recovered ammonia was lower than
those in the trial runs. Some of the reasons are that the feed
concentration was lower as described above and that the
residual concentration was higher probably due to the
existence of GlyGln, which is an acid.
Summary
The ammonia recovery system in dipeptide manufacturing
processes, utilizing the ammonolysis reaction, has been
established. The optimal operational conditions were estab-
(4) As the analytical result of this sample is shown in Figure 4, the mole ratio
of N:C was 14:3. In the lab experiment, NH4HCO3 was supplemented in
the preparation of the reaction solution so that the ratio became 14:5, the
original condition shown in Figure 2.
Further Development
As described above, the process to recover ammonia has
been successfully established to save cost and time. The
134
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development

lished mainly by simulation. The implementation of recover-
ing ammonia contributed to cost and time savings by
simplifying the treatment of wastewater. The prospect for
recycling of ammonia was also developed. From the view-
point of quality, recycling was considered to be feasible from
the results of the lab experiments.
Acknowledgment
We are grateful to Satoshi Matsuoka, Hideo Kurokawa,
Kazuyasu Osada and Yurie Asano of the Yokkaichi Research
Laboratories of Kyowa Yuka Co., Ltd., for their technical
cooperation during the simulations, and we also thank the
Sakai Plant operators for the plant scale trials.
The results during the actual production showed that the
employed simulation methods are proper and will be ap-
plicable to the similar reaction system, for example other
dipeptide processes.
Received for review September 6, 2000.
OP000094X
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
135