Small is bountiful

’Disposable microreactors’, from miniature processing plants to laboratories on a chip, bring chemical manufacturing to the desktop. Cath O’Driscoll reports

Chemical manufacturing is traditionally a conservative business, notes entrepreneurial chemist Wolfgang Ehrfeld. New processing plants don’t come cheap and their 20 to 30-year lifetime leaves little scope for innovations, he says. So, most companies continue to make their chemicals in much the same way for decades.

But Ehrfeld is banking on change. He specialises in microreaction technology - the business of using miniature devices for building conventional chemical plants at the scale of the average desktop. ’The market lifetime of a microreactor will be as short as that of a microprocessor,’ forecasts Ehrfeld, cofounder of Ehrfeld Mikrotechnik in Wendelsheim, Germany and former chief executive of the government-funded Institute for Microtechnology Mainz (IMM).

’It is not expensive to replace microdevices in a chemical system. I have only to pay, say, ?30000 and I can do this and that, and then I can throw them out after one or two years and replace them with something else - and this will speed up process development dramatically.’

Ehrfeld says that he is far from alone in recognising the commercial potential of microreactors. ’A lot of work is still being done in universities and government funded research institutes, but more and more companies are now starting to use it in production or technical [production] scale.’

For Andrew deMello, the real interest in working at these small scales lies in the power of microfluidics. He is an expert on ’lab-on-a-chip’ devices - microreactors in the form of electronic processors that support both reaction and analysis. ’In microfluidic reactors, reactions take place in t iny etched channels or beds between 10 and 500 ?m wide and 10 to 100 ?m deep,’ says deMello, a senior lecturer at Imperial College, London.

’Because devices are made by microfabrication techniques, it’s just as easy to make hundreds of devices as it is to make one. So if you have hundreds of reactors running simultaneously you begin to simulate a large-scale reactor flow, and that becomes very useful.’

Though the technology is slowly establishing itself, few companies are giving too much away. Merck has used microreaction technology in a full-scale plant facility for around five years, producing 15 t pa of one speciality intermediate in Gernsheim, Germany. Since 2002, Merck has also been running a pilot plant in Darmstadt, built from modular units and designed to yield about 7 t pa of some 10 other key intermediates. Clariant, DuPont and BASF are also known to be using microreactors in process development and/or production.

’Numbering up’ instead of scaling up reactions in this way has big advantages. For a start, companies no longer need to transfer reactions from the lab to the plant - one of the most costly and time-consuming phases in bringing a new product to market. In principle, they can also deliver their chemicals directly to their customers as a nd when needed.

And there are more subtle benefits. ’The way that mixing happens on a very small scale gives you a lot of control,’ says deMello, referring to a typical microfluidic reactor. Fluids at the microscale do not use turbulence to mix, as milk does into coffee for instance; they diffuse as laminar flow into one another. ’We’ve looked at a number of multi-stage syntheses and simply by varying the flow rate (and thus mixing regime) we could tune into a particular product. This level and ease of control is simply not possible in batch reactors.’

Temperature control in microreactors is equally straightforward. High surface to volume ratios in microreactors allow very efficient heat transfer and create another useful feature for process chemists. Ehrfeld elaborates: ’Ideally most of them would like to have an isothermal process and you cannot have this in a large tank because there is no way of removing the reaction heat evenly. But if you go to small dimensions then you have this possibility and you can look at your process and see how it might operate under totally isothermal conditions. What potential does your process have if you optimise conditions?’

Higher yields and better selectivities are commonplace, but one of the major objectives of this type of work is to replace multi-stage reactions by single step processes. ’For most of the chemical systems we’ve looked at we’ve found some advantage’, deMello observes.

Companies interested in trying out the technology for themselves can now order anything from individual microdevices - millimetre-size heat exchangers, pumps and reactors - to entire desktop-size modular systems capable of carrying out all of the unit operations seen in a conventional process plant.

Microreactors must surviv e in a dirty world, but Ehrfeld is sanguine. ’If you’re looking at the most efficient, best microreactor in the world - the living cell - it lives in a very dirty environment.’

Unlike cells, microreactors lack the networks of complex feedback systems needed to control all these various processes. Most feedback relies on integrating large numbers of strategically placed sensors, with online monitoring processes such as Raman and FTIR spectroscopies. Lab-on-a-chip suppliers such as mgt microglas technik in Ma inz advocate using glass because it allows chemists ’to follow their reactions and processes optically to build up detailed information on flow patterns and the course of mixing’.

New and improved software will nevertheless be critical, write deMello and colleague Robert Wootton in a paper in Lab on a Chip (2002, 2 , 7): ’In particular, as the size of a microreactor array increases both reaction monitoring and process control become increasingly difficult. This dictates that the control architecture surrounding the "microfluidics" becomes more expensive and complex. Consequently, fluidic and electronic interfacing will most likely be the crucial issue in defining how widespread microreactors become in the area of fine chemical production’.

Moving to the mic roscale brings other potential problems. ’Certainly, when you move to these small scales you’re working in environments where the surface becomes really important and in many cases dominant. In a test-tube molecules rarely see the surface whereas in a microfluidic system surface-molecule interactions become a lot more common - and that changes the chemistry, for good or bad’, deMello says. Precipitation can be a problem, though. ’If you’re running 1000 reactions all at once and three of them get blocked, you simply replace those three and the rest keep on working’: much easier and cheaper than the ’downtime’ associated with maintaining a conventional batch reactor.

Industry can not afford to ignore the potential benefits for much longer, insists Ehrfeld. Wit hin the next five years, he predicts, suppliers of both commodity and fine chemicals will use microreactors. ’Companies will go to modular systems or they will install some microstructure or microdevice in their plant. But finally if you look at all the possibilities they will go to complete microreaction plants.’

Liquefaction of surplus natural gas could be one of the first targets, says Ehrfeld. Instead of burning off the gas at sea, oil companies could use microreactor plants on their rigs to turn the gas into methanol so that it can be safely shipped to shore.

Nor should we forget the importance of lab-on-a-chip applications, Ehrfeld stresses. ’Lab-on-a-chip devices where reaction and analysis all takes place on the same chip are also microreaction systems, and these are not for the production of large amounts of substance but mainly for the production of information’, he says. Also, the growing interest in combinatorial chemistry, in which hundreds to thousands of reactions occur in parallel, make lab-o n-a-chip devices look increasingly attractive for screening applications.

According to an estimate by French consulting company Yole D?veloppement, working jointly with IMM, the sales volume for microreactor technology in 2002 reached about €36m ( ca ?25m), only for applications in R & D and process optimisation. In 2003, Yole expects the level of industrial investment to increase by 30 per cent. Moreover, Yole estimates a world market for lab-on-a-chip in 2002 at €80m, with a 30 per cent annual growth rate .

But deMello urges a note of caution: ’Microreactor technology has become a very sexy area, but certainly there are areas where microfluidics doesn’t particularly help you, and the only reason you should use such an approach is when there is some genuine advantage or justification for doing so’. Slowly but surely, it seems that chemical companies are now beginning to decide that for themselves.