Continuous improvements

C&I Issue 8, 2009

The traditional workhorse of pharmaceuticals production, the stirred tank, has to fulfill very diverse, even conflicting, tasks during the sequence of unit operations of a chemical transformation. Every large-scale batch process is a compromise between the requirements of the chemistry and the ability of the reactor. Batch operation requires the exact repetition of a sequence of different actions and process conditions within pre-defined ranges. Product quality is therefore the result of strict controls.


Process intensification strives to remove some of the limitations of batch reactors by ensuring that each phase of a reaction proceeds in the most suitable equipment for that reaction stage. At DSM, we have been applying the principles of process intensification, both in the process development laboratory and at selected plants, for many years.


The company started to direct its attention to some of the existing exothermic processes and processes involving hazardous reagents outside its Good Manufacturing Practice (GMP) operations in 2001.


From early on, the rationale has been not to replace existing equipment by ‘smaller reactors’, but to carefully analyse existing processes to find out what limited their productivity or selectivity. In cases where transfer of mass or heat in a certain phase of the reaction proved to be limiting, for example, we started to define the ‘desired environment’ for that phase. While the ideal apparatus for the most exothermic phase of these reactions proved to be small structured reactors or microreactors, moderately exothermic phases were best performed in conventional tube reactors. Stirred tank reactors were used for the final, lengthy, barely exothermic, phase of a reaction.


The primary goal of these efforts to intensify our processes has been to bring down production costs. Process intensification brings a number of potential benefits: •Improved selectivity improves the throughput of the plant and reduces waste costs. •The higher quality of the crude reaction product simplifies work-up and increases the throughput of the work-up. •Taking the most exothermic phase of the reaction into a microreactor increases the throughput of the whole plant.


In the course of our development work we found that chemistries we had not touched at all for safety reasons became accessible to us at laboratory and production scale.


Parameter variation was done by applying one set of reaction parameters after the other while the reactor is continuously operated and its output analysed for product yield, conversion, byproducts etc. As the residence time in a typical lab set-up is short – seconds to a few minutes – the number of different sets of parameters screened can be very high, provided the analytics are capable of monitoring in or close to real time. This requires analytical methods different from those in batch experimentation.


Laboratory examples
Among some of the examples from DSM’s development lab, we investigated the continuous substitution reaction of a mesylate by an azido group, an important intermediate for the production of amines (Scheme 1). Adding an azide salt solution to a batch of mesylate solution is acceptable in the laboratory, but doing this in a multipurpose plant requires extensive safety measures, as azido compounds are toxic and prone to spontaneous decomposition under certain reaction conditions.


We found that the substitution worked smoothly in a simple set-up consisting of tubes and T-mixers at temperatures as high as 120°C within 1 minute. Continuous IR monitoring allowed us to determine the effect of any change in process conditions immediately.


The Ritter reaction (Scheme 2) is another synthesis targeted by DSM for improvements. In 2002, the company chose to improve a batch Ritter reaction process – involved in the production of acrylamides – running in the plant by using a microreactor. The original process struggled with the heat release during mixing of the starting materials, resulting in long dosing times accompanied by bad selectivity. The proof of concept and the main parameters for the microreactor reaction formed part of a PhD thesis. This was followed by fine-tuning in our development laboratories, ultimately leading to the first 300t of product produced in 2005.


The Ritter reaction involves the addition of a nitrile and a tertiary alcohol to produce an amide.The reaction starts with mixing of the alcohol with concentrated sulphuric acid to form a cation – a very exothermic process leading to extensive tar formation if the temperature is not carefully controlled. This cation is then reacted to a nitrile and subsequent hydrolysis yields the N-alkyl amide. This reaction was optimised in a laboratory reactor designed by the Institut für Mikroverfahrenstechnik (IMVT) in Karlsruhe, Germany.


Based on the lab data, the IMVT designed a reactor capable of producing 1-2 tonnes of reaction mixture/hour. The reactor was added to the plant to allow its continuous operation and because the mixing time, capacity of heat removal and residence time remained unchanged, we saw the same improvements in selectivity as in the lab reactor. In this manner it was possible to increase the yield of the process from 55% to 78%.


The reactor did not replace elements of the existing plant. It was added to contain the most exothermic phase of the reaction, the initial mixing of the reaction components. As it avoids any back-mixing and is able to control the temperature even of the most exothermic phase of the reaction – a few seconds – within a few degrees of the set-point, the reaction proceeds under ideal conditions. In the next phase, the reaction enters a tube heat exchanger, which again keeps temperature very close to the set point and avoids back-mixing. In its final, barely exothermic, phase the reaction proceeds to completion in the former fed batch equipment.


While the reactor described above was built to contain the reaction for a few seconds, the next reaction we developed was the selective mono nitrate ester formation of a compound containing a number of hydroxyl groups (Scheme 3).


This required running a number of subsequent and parallel unit operations in one device. Consequently its throughput was smaller and its complexity greater. Again we used a lab scale device to define and optimise the basic reaction parameters. This time we selected a glass reactor provided by Corning. We soon found that the effective extraction of the nitrate ester into an organic phase was crucial to avoid extensive overnitration. Therefore the reactor had to mix the reactants and keep the temperature constant. At the same time it had to continuously extract the reaction mixture until the required conversion had been reached. The mixture was then diluted and neutralised in three steps while keeping the temperature at a second level.


A pilot reactor was designed based on the lab recipe and operated to produce several 100kg of a solution containing the nitrate ester. Work-up was also made continuous where advantageous. It consisted of phase separation with mutual extraction – extracting the product into water, back into the organic phase and then concentrating the solution to make it suitable for the subsequent step. Here we used centrifugal extractors to effect the required extractions quickly while keeping the overall hold-up of nitrate esters small.


As a first step, a ‘batch’ was defined as the amount of starting material placed in a feed vessel to run the reactor for a defined period of time: 36 hours. As we had run the nitrate ester synthesis in fed batch mode under cGMP conditions using the same starting materials as before, the switch to continuous operation constituted only a minor change from a cGMP perspective. The device had not been operated in any of our plants before, nevertheless we managed to deliver in-spec product in the required quantity on time.


Process outlook

Switching to a continuous process is a viable option for roughly half of all processes. However, this comes at a cost. Usually, investment is needed at the outset because parts of the specific equipment are not available in a typical multipurpose environment. Re-configurable reactors that adapt the continuous reactor specifically to the needs of the particular synthetic step should be helpful here.


At DSM, more applications of continuous intensified processes will follow. Many of our customers have discovered the merits of continuous processing in the lab and started teaming up with reliable developers of continuous processes. While sticking strictly to transformations with immediate business applications, DSM will add more elements of ‘quality by design’ such as continuous analytics and unit


Acknowledgements: Thanks to Raf Reintjens, competence manager at DSM Pharmaceutical Products in the Netherlands for help with the preparation of this article.


Sascha Braune is a process development chemist and Peter Poechlauer is principal scientist at DSM Fine Chemicals Austria, Linz, Austria.


Further reading
1. John Kossik, Steadfast Equipment at html?page=1
3. S. Braune et al, Chemistry Today, 2009, 27 (1), 26.

Become an SCI Member to receive benefits and discounts

Join SCI