Organic Process Research & Development
Article
permeability between 0.5 and 0.6 L m2− h−1 bar−1 was
observed. This was the case with two membranes of the same
dimensions and pore size but from different production
batches. Variation of the permeability rate can be achieved
either by altering the membrane parameters (surface area,
pressure etc.) or the reaction parameters. Trials were therefore
carried out to increase the rate of cyclization by increasing the
mol equivalents of iodine used from 1 to 1.5 (Table 1 entries
6−10). Nevertheless, achieving the target complete conversion
of the starting material proved to be somewhat more complex
than expected. Other methods of changing this reaction rate
such as adding some iodine to the mixer/reaction tank before
the addition of 1 was started (Table 1, entry 7) or warming of
this tank to 30 °C (Table 1 entry 8) met with mixed success.
Complete conversion was achieved using 1.5 equiv of iodine
which was added to the reaction vessel concomitantly with the
addition of 1; however, the requirement being the addition
must be a smooth, continuous addition. Indeed, in earlier
experiments the required rate of addition of the iodine solution
was sufficiently low that the addition system used tended to add
this not in a smooth continuous manner but more in spurts
with intermittent periods without iodine addition. Once this
problem was solved complete conversion was achieved (Table
1, entry 9).
(Table 1). Selectivity is simply the ratio of reaction yield to
conversion expressed as a percentage. Furthermore, because of
the high rejection at the end of the process practically all of 2 is
to be found in the separation vessel (Scheme 2b) with the
remainder being in the mixer/reaction tank.
Under the double addition mode (Table 1, entry 9) the
accumulation of peptide in the separation vessel over time is
shown graphically in Figure 1. Evidently during the reaction
time were the solutions of iodine and open chain peptide 1 are
added into the mixer/reaction tank (Scheme 2b); 2 is to be
found in both the filtration loop and the diafiltration tank
(mixer/reaction tank, Scheme 2b). On completion of the
reaction phase of the sequence, 2 in the mixer/reaction tank is
transferred via the diafiltration process, which is operating
under total return of the permeate, i.e., V-1 is open and V-2
closed, to the separation tank. The reaction time of 20 h is
comparable to the batch process which is carried out over 16−
18 h. Transfer of 2 into the separation tank was carried out by
simply allowing the diafiltration process to continue;
consequently, this transfer process is rather long. Obviously
the duration of this process can be reduced dramatically simply
by transferring all the contents of the mixer/reaction tank, once
the reaction phase is complete, in one go into the separation
tank.
Under the conditions used in Table 1, entry 9, consistently
high conversion and yield can be achieved with minimal solvent
use. This cyclization has now been performed under the
conditions used in entry 9 of Table 1 numerous times with
consistent yield and product purity. An attempt was made to
further purify the reaction product on completion of the
reaction (Table 1, entry 10). Thus, a reaction was performed in
exactly the same manner as previously (Table 1, entry 9).
However, on completion of the reaction v-1 (Scheme 2b) was
closed and v-2 opened. The contents of the filtration loop were
washed with fresh water and with the intention of washing away
some at least of the impurities found in the mixture. Though
this appeared to have minimal effect on purity of 2, there is
some effect on yield due to some losses during the washing
step, also there is obviously a direct effect on the quantity of
solvent used.
The process conditions of Table 1, entry 9, which for this
reaction gave the best result, were also carried out over an
extended time period (144 h) with the intention of monitoring
any changes in the peptide rejection profile. The results show
the stability of both the peptide rejection and accumulation of 2
within the retentate under these conditions (Figure 1b).
A final evaluation of the rejection and flux profile of the
process was performed in which the filtration loop was cooled
to below 20 °C. The reasoning behind this being that reducing
the temperature of the filtration step could reduce the rate of
possible degradation pathways. However, reducing the temper-
ature of the filtration would also affect the filtration parameters.
The experiment was carried out as previously, i.e., a reaction
phase of 20 h followed by continuation of the diafiltration to
transfer 2 to the filtration loop. As can be seen graphically in
Figure 2 and Table 1 entry 11, effects on yield and purity of 2
are minimal, though this reaction did give the best yield, purity
of 2 and lowest PMI. Rejection also remains high as with
previous reactions. The largest effect of cooling the filtration
loop is on the membrane permeance which is reduced to
approximately one-third of that when the filtration is carried
out at ambient temperature.
Rejection and Fate of Products. As the concentration of
solutes in the separation vessel increases, their rejection profile
also changes to a certain extent. This resulted in a slight
reduction in the rejection of both 1 and 2 (Table 2). The high
Table 2. Rejection Profile of 1 and 2
Product Assay. Quantitative analysis of crude reaction
product formed using the double addition OSN process shows
an almost identical impurity profile as that obtained with the
current batch process using the same batch of 1 as starting
material. Furthermore, no impurities that were not already
known from the batch process were observed when the OSN
process was used (Figure 3). Indeed, a more detailed
examination of the impurity profile formed during this
cyclization showed a slight increase in the quantity of an
impurity resulting from iodination of the tyrosine moiety of 2
but lower quantities of dimeric impurities than found in the
batch process. It should also be stated that the majority of
impurities found within the reaction product are impurities
originating from the starting material 1 whose assay was 87%.
Furthermore, post-reaction wash with pure solvent as in entry
10 of Table 1 had in this case virtually no effect on the product
purity; though there remains the possibility that with a different
time (h)
rejection 1 (%)
rejection 2 (%)
7
20
30
98.6
97.6
96.9
99.1
97.5
96.2
rejection of the reaction starting material is the reason for the
incomplete conversion when the rate of reaction is slower than
the speed of diafiltration. The high peptide rejection observed
can thus be used as an advantage to improve product purity. If
the cyclization is complete, before reaction mixture is added to
the filtration loop, via the diafiltration process, then 2 is
removed from the reaction vessel, thus preventing further
reaction of 2 to secondary products. The effect of preventing 2
from further reacting to secondary products can be seen in the
significantly improved selectivity from experiments using the
double addition mode as oppose to those from single addition
844
Org. Process Res. Dev. 2015, 19, 841−848