Reaction Experiments

This section is split up based on the various catalysts used, and arranged in chronological order as the experimental set-up evolved gradually during this investigation. It should be noted that this project was a collaboration with another researcher who, at this point, became the more active party in the project. As such, this section will be less detailed than the previous sections to respect and preserve his intellectual right to his work.

3.8.1 Silver Foil and Wire Reactor

After successful leak testing, the reactors were deemed suitable for carrying out reactions. The initial interface design was used for this experiment. The reactor containing the silver foil was tested first. A preliminary experiment was run to gauge the activity o f the catalyst. The ratio o f methanol:oxygen used was 2 (oxygen was diluted with helium) and a reaction temperature o f 360 °C was used. The experiment was then repeated at 380 °C. The results showed little or no formaldehyde formation, but significant carbon dioxide and water production, indicating that decomposition/combustion o f methanol had taken place.

An attempt at activating the catalyst by oxidising it was made. A mixture containing 7.2 % oxygen (balance helium) was passed through the reactor at 380 °C for 1 h. Methanol was then passed through the system again at the same ratio as above. A significant amount o f formaldehyde was formed, with little by-products. Yet almost no methanol was detected. It was then found that reducing the concentration o f methanol resulted in an increase in formaldehyde formation.

However, a complication presented itself when the methanol flowrate was reduced: the pressure within the system prevented methanol from entering into the system. In fact, the methanol was pushed out o f the inlet altogether. From this, it was deduced that the flowrates o f both methanol and oxygen/helium had to be balanced. Because o f this, some experiments were then conducted without the presence o f helium, in order to reduce the flowrate on the gaseous side.

In a series o f experiments, it was seen that formaldehyde production steadily declined, while carbon dioxide production increased. This suggested that the catalyst retained its activity for only a short period. The experiments using a silver wire had the same results. It is likely that both the foil and wire had similar surface characteristics, and could not remain active. Therefore, it appeared that neither catalyst was suitable.

In retrospect, it is a possibility that both the wire and foil had been poisoned by the septa that were used as gaskets at this point. In any case, the tediousness o f placing either catalyst in the reactor would have precluded further use.

3.8.2 Silver M irror Reactor(I)

Catalyst deposition in this reactor had been performed using the method involving sodium potassium tartrate. Because this reactor had already been heated in a fiimace at 400 °C for over 30 min to anneal the silver deposit, it was thought that pre­ treatment would not be necessary. The reaction was carried out at 380 °C, with a methanol: oxygen ratio o f 2. There was insignificant production o f formaldehyde. However, there was also very little methanol coming out o f the system.

Upon closer inspection, it was observed that bubbles had formed in the methanol feed at the transition between PTFE tubing and stainless steel tubing, and had blocked the flow (in small channels, bubbles require a larger pressure drop to force them through a constriction, as mentioned in Chapter 2). This vaporisation occurred because the methanol flowrate was lower than in the previous experiments and it had vaporised completely within the stainless steel tubing. As such, there w asn’t enough pressure to continuously push methanol through to the reaction channel. At this point, the two cartridge heaters closest to the inlet were removed so as to reduce the temperature there. After this, blockage due to bubble formation did not occur again.

An attempt was then made to activate the catalyst by treating it with oxygen at 400 °C. Following this, experiment was conducted at reduced methanol:oxygen ratio. As with the previous catalysts, activity was shortlived. At this point, suspicion grew that the septa gaskets could be releasing some volatile component(s) and poisoning the catalyst as prolonged use had caused the silicone centre to melt and flow.

From these two sets o f experiments, it was clear that a methanol-rich environment was detrimental to the reaction. More carbon dioxide was formed, while formaldehyde formation was insignificant. When the methanohoxygen ratio was reduced to about 0.75, formaldehyde was produced in noticeable amounts.

3.8.3 Silver M irror Reactor(II)

Here, catalyst deposition involved the second method detailed in the previous section. The catalyst was known to be active as the particulate version mentioned previously had been tested in a tube reactor and had proved to be fairly active. The second

interface design was ready at this point and was used, together with the graphite sheet gaskets. A new HPLC pump, Jasco PU-1580, was also tested. It was hoped that this pump would prevent or reduce the pulsation that was observed in previous experiments. Also, before, liquid methanol had shown some difficulty in entering the inlet tubing due to pressure resistances. This pump had a larger pumping pressure. The reaction was carried out at 450 °C, with a composition o f 5.4 % methanol, 1.5 % oxygen and balance helium. This is a higher methanohoxygen ratio than previously used, however, it had been the conditions used in the tube reactor. Unfortunately, conversion was very low.

It was now clear that there was something wrong with the system and not the catalyst deposition method, as this catalyst had been shown to be active outside o f the reactor. There were several possibilities for the low conversion:

a) insufficient catalyst surface area: The catalyst was only on the surface o f the reactor. As such, the surface area was relatively low.

b) different form o f catalyst: The catalyst in the reactor was whitish in colour whereas the catalyst on the particles looked rnore silvery. It was possible that the silver deposited on the reactor had a different crystal structure and was less catalytically active or required a different pre-treatment procedure.

c) catalyst deactivation: the pump was not very effective, as methanol was still had difficulty reaching the reactor occasionally. It has been seen previously in the tube reactor that loss o f methanol supply, even temporarily, could result in permanent loss o f catalytic activity. This may have happened to the catalyst within the reactor too, as activity slowly died down.

d) catalyst poisoning: it was found that the graphite gaskets were not pure graphite but in fact comprised o f a layer o f metal sandwiched between 2 thinner graphite layers. As it was unknown what the metallic layer consisted of, it was possible that some poisoning had occurred.

At this point the graphite sheets were found to be too porous, thus leading to air being drawn in from the surroundings. This, together with the suspicion that they were responsible for catalyst poisoning led to the construction o f another interface where the gasket material did not come into direct contact with the inlet gas flow.

3.8.4 Summary o f Further Experimental Work Carried Out

Because the mechanism o f feeding methanol to the reactor was not working as intended, major alterations were made. The methanol pumping system was replaced by an evaporation system. In this set-up, an oxygen/helium mixture was bubbled through methanol. This gas stream would then carry methanol vapour with it into the reactor, which would solve all the problems with irreproducible methanol boiling in the inlet/pre-heat section. The concentration o f methanol in the gaseous stream was dependant on the temperature o f the liquid methanol, which was kept constant in a water/ice bath.

When the new interface using graphite ferrules was put into use, rapid catalyst deactivation stopped being a problem. The method o f catalyst deposition was changed yet again as well, this time to sputtering. High conversion was achieved using a mixture o f 8.0 - 8.6 % methanol and balance oxygen, without any helium to quench the reaction. Please refer to Cao, et al (2002) for more detailed results.

3.9 Conclusion

The principles and method o f fabricating a microreactor from scratch were demonstrated. W hile a lot o f literature was available regarding microfabrication, the actual work o f fabrication was not straightforward and experience had to be gained. Much time was also spent researching the relatively unconventional catalyst deposition methods which would be compatible with microfabrication, and even then there was always the uncertainty about the behaviour o f the catalysts prior to testing.

As mentioned in the previous chapter, one o f the major problems facing the commercialisation o f microreactors was the difficulty in building an adequate interface. This proved to be a major problem in this case as well, as demonstrated by the difficulty in vaporising methanol in-situ and the susceptibility o f the catalyst to poisoning. The major cause o f this difficulty was the high temperatures required for the reaction. This was compounded by the brittle nature o f the silicon and glass. Together, this resulted in a very limited selection o f sealing methods and materials.

However, working microreactors were eventually fabricated thus completing the main goal o f this exercise. The side goal o f performing most o f the fabrication in-house had also been achieved by the end o f the project. A good foundation o f the concepts involved in microfabrication, particularly silicon microfabrication, was built thus easing future work in this area. Also, with the identification o f the difficult areas, i.e. sealing, catalyst deposition and interfacing, further work in this field will be more streamlined and it will be easier to focus attention on these points. A working knowledge o f the timescale involved in producing a microreactor was also obtained.

All the experience gained from this exercise forms the basis for producing an algorithm for designing and testing microreactors, which is presented in Chapter 8.

The next generation o f reactors has already been fabricated. The aim o f these new reactors is to increase throughput either by enlarging the reaction channel or by having many channels in parallel. The reactors were also specifically designed to allow the use o f impregnated catalyst particles.