Microfabricated reactors for on-chip heterogeneous catalysis

Abstract

Microfabricated devices constructed from glass and polydimethylsiloxane with integral heaters are described, which can be used for heterogeneous catalysis reactions. Sulfated zirconia is used as the catalyst in an open channel reactor, with either a syringe pump or electroosmotic flow being used to deliver the reactants. The results clearly demonstrate that very high conversion efficiencies are possible, however, the thermodynamics of the reactions are the same as in bulk systems. Ethanol and hexanol are dehydrated to ethene and hexene, respectively, with conversion efficiencies approaching 100%, and the esterification of ethanol is investigated. Yields of approximately 30% ethyl acetate are obtained by gas chromatographic analysis. This is the first time such a method for fabricating a catalyst micro reactor has been reported, yet it demonstrates sufficient robustness and resistance to leakage. The use of electroosmotic flow in a heated catalyst reactor is a significant advancement in reactor design.

---

Introduction

Miniaturised systems for chemical analysis and synthesis represent a rapidly developing research area, encompassing chemical and biological applications. Commercial systems are already available, most notably for DNA analysis. The chemical analysis devices are commonly referred to as micro total analytical systems (µTAS), and have been reported for a diverse range of applications. Laser induced fluorescence detection has proved to offer very sensitive detection for chip-based separations; such as reported by Hadd et al.¹ who determined a number of acetycholinesterase inhibitors after electrophoretic separation on a glass chip, and a polymer chip² that was employed for the analysis of PCR products of the hepatitis C virus. Spectrophotometric detection of analytes using fibre-optic connections provides a convenient quantification method for on-chip measurements, including the determination of nitrite via the formation of an azo dye,³ and orthophosphate via the Molybdenum Blue reaction.⁴ Electrochemical detection of analytes has been reported by a number of workers, such as Wang et al.⁵ for the detection of nitroaromatic explosives and catecholamines. There are numerous reports of microchips employing mass spectrometric detection, this offers a potentially powerful analytical tool, e.g., for rapid analysis of proteolytic digests.⁶ While electrophoretic processes prove vital for on-chip separations, electroosmotic flow offers a very convenient pumping mechanism for fluidic manipulation, such as for moving solutions through micromachined filters.⁷

However, the application of microfabricated devices to synthetic applications is deriving similar interest, one only has to look through the proceedings of recent conferences (such as IMRET 3 (1999) or IMRET 4 (2000)) to see the potential of such devices. These include highly exothermic reactions, e.g., the direct fluorination of aromatic compounds, the in situ generation of hazardous compounds, e.g., phosgene, and rapid energy transfer systems, e.g., fuel pumps. However, one area of interest to us is on-chip heterogeneous catalysis. This represents an important area in organic synthesis, and there are clear benefits in developing miniaturised synthetic reactors utilising minimum catalyst in conjunction with very limited reaction volumes. This latter feature permits very rapid mass transfer and control of reaction conditions. In addition, the potential to employ electroosmotic pumping with such devices represents a very attractive feature, permitting reagents to be pumped by a device with no moving parts. Miniaturised catalytic devices can be used to conveniently gain kinetic and thermodynamic data for reactions. They are also convenient for screening new reactions and catalysts. The screening can either be with respect to the catalyst, i.e., to determine which catalysts show activity, or with respect to the reaction, i.e., which reaction shows promise on a certain catalyst.

Within the Hull group, extensive work is on-going in the area of micro reactors employing catalyst mediation, but the work reported in this paper is the first in a heated system. This paper expands on the developments we reported earlier,⁸ and shows further reaction details and additional reactions. We report the relationship between electroosmotic flow rate and carbon number for alcohols, the dehydration of hexanol to hexene and ethanol to ethene (with some degree of cracking), and the esterification of ethanol. The catalyst used was sulfated zirconia (zirconia treated with sulfuric acid). This catalyst is known as a super acid catalyst with well documented reactions including the conversion of methanol to hydrocarbons,⁹ isobutane/but-1-ene alkylation,¹⁰ and the methoxymethylation of alcohols.¹¹ It has also been reported for the dehydration of alcohols to the associated alkene, with the added advantage that the reaction conditions are significantly milder than those required for other acid catalysts, such as silica–alumina. In a previous paper,¹² a 1 mm diameter reactor was reported to offer conversion efficiencies of only 3% for the conversion of hexanol to hexene, however, in this paper, conversion efficiencies approaching 100% are reported (compared with the industrial process that only gives approximately 30% conversion for the dehydration of hexanol). In the context of this work, conversion efficiency is the amount of product made compared with the amount of starting material. Therefore a 100% ‘yield’ would equate to 1 mol of starting material being completely converted to the product(s).

Experimental

All the alcohols were bottled in-house except for hexan-1-ol (Avocado Research Chemicals, Heysham, UK). Prior to use in the catalyst reactor, ethanol and hexanol were purged for 10 min with nitrogen to prevent coking; a frequently encountered problem at high temperatures when trace oxygen is present. The ethanoic acid and sodium tetraborate (Fischer Scientific Ltd, Loughborough, UK) were used without further purification. The sulfated zirconia (ZrO₂–SO₄²⁻) catalyst (MEL Chemicals, Swinton, Manchester, UK) was first activated by heating to 600 °C for 1 h in a microwave furnace (CEM). The polydimethylsiloxane (PDMS) was obtained as Sylgard 184 (ISL, West Midlands, UK), and required mixing and vacuum degassing for 20 min prior to use. The Nichrome wire (Goodfellows, Cambridge, UK) was obtained without insulation. The Crown white glass (Alignrite, Bridgend, South Wales) used for the fabrication of the chips was obtained pre-patterned, and the etchant was prepared from AnalaR hydrofluoric acid (BDH) and 98% ammonium fluoride (Lancaster Chemicals, UK). The Superwhite glass for the top plate was obtained from Instrument glass (Enfield, UK). Pumping was achieved using a syringe pump (Baby Bee, BAS, Cheshire, UK) or an electroosmotic pump fabricated as previously reported.¹²

Chip fabrication

The chips were fabricated from either glass alone or glass and PDMS. The latter combination was found to be most versatile since it permitted the catalyst to be easily immobilised on one wall of the channel and to allow incorporation of the heater element above. In this work, the base plate containing the channels was fabricated from glass, however, PDMS would be as applicable for many applications.

The micro reactors were constructed from two plates. The base plate was produced by photolithography and wet chemical etching of Crown white glass that had been pre-patterned and developed by Alignrite. The patterned glass was hard baked at 120 °C for 24 h prior to etching for 1 h in 1% HF–NH₄F at 70 °C. Then, the remaining photoresist and chrome was removed. Two channel designs were used. The first consisted of two intersecting channels, each 20 mm long, 200 µm wide and 80 µm deep forming a T-pattern. In the second design, the channel dimensions were 200 µm wide by 80 µm deep and 30 mm long in a Z-configuration. A schematic diagram of the channel layout can be seen in Fig. 1. The top plate for the T-chip consisted of a 17 mm thick block of Superwhite glass drilled with three 3 mm reservoirs to match the end of the channels. This was thermally bonded at 570 °C for 3 h.¹³ The top plate for the Z-chip was prepared from Sylgard 184 resin, cast in a mould containing inserts to form the reservoirs. After baking for 1 h at 100 °C, the resin was released from the mould and the inserts removed. The mating face was then coated with a thin layer of PDMS and activated catalyst dusted over the surface; it was then baked at 100 °C for 1 h. The effect of this step was to give a high surface area for the catalyst reactor, but with the catalyst firmly immobilised. When this surface was clamped onto the Z-chip, the effect was to produce a reactor with one wall of the channel being catalytically active. The in situ heater was fabricated from Nichrome wire (Ni–Cr 90∶10) 0.25 mm diameter carefully formed into a serpentine shape and immobilised in the PDMS top plate and covering the channels. In the finished chips, the heating element was 2 mm from the catalyst surface and it heated 15 mm of catalyst coated channel. The heating wire was inserted into the mould near to the mating face before the liquid PDMS was poured into the mould. The PDMS was stable to heating up to 180 °C (this temperature was not exceeded so no data exists beyond this point). Heating was achieved by a potentiostat (0–270 V) and monitored via a digital thermometer (Jencons model 2003) with the temperature probe (RS components) located close to the reaction channel. Where electroosmotic flow was used as a pumping method, a 2 cm long tube (0.5 mm id) was inserted into the base plate, butting up to the start of the reaction channel. This was filled with a porous frit¹² to act as the pump. Platinum electrodes were located at the front and rear of the tube to provide the potential, and a strip of copper was placed 1 mm away to act as a heat sink. On-chip electroosmotic pumping is at present very difficult to operate reliably due to vaporisation of the ethanol on the heated chip, however, this limitation has been overcome here with the aid of the copper heat sink. The two halves of the composite chip were held together by pressure exerted from two rigid Perspex plates, clamped together using four bolts. The bolts were tightened sufficiently to prevent leaks but without distortion to the soft polymer top plate.

Fig. 1 The layout of the microreactor channels on the T- and Z-chips. The dotted circles show the positions of the reservoirs.

Results and discussions

Electroosmotic flow study of alcohols

The glass T-chip was used for this study, with 0.5 mm diameter platinum electrodes inserted into two reservoirs. Initially, the chip was filled with 20 mM disodium tetraborate buffer. A constant voltage of 695 V (HVPS, Farnell, Leeds, UK) was applied between the electrodes, resulting in a field strength of 230 V cm⁻¹. The voltage was applied for a period of 10 min, after which time the volume of solution moved between the two reservoirs was measured. This was repeated three times with fresh solutions, then the alcohols were studied sequentially. The results can be seen in Table 1, and clearly show an overall decrease in flow rate with increasing carbon number, hexanol does not move by EOF. When a graph of carbon number versus flow rate is plotted, a linear relationship is obtained for C₁–C₄, with y = −0.78x + 3.95 (r² = −0.9941) being the equation of the line. It is interesting to note that hydrodynamic flow can also occur in the chip for the lower chain alcohols (decreasing rapidly with chain length), arising due to developing unequal heights in the reservoirs from the EOF. It is apparent that it is in the opposite direction to the electroosmotic flow, and the figures quoted do not include compensation for the effect of hydrodynamic flow.

Table 1 The electroosmotic flow of five alcohols through the T-chip (n = 3). Sodium tetraborate is included only for comparative purposes

Solvent	Flow rate/µl min^−1	σ^(n − 1)
20 mM sodium tetraborate	3.6	0.06
Methanol	3.2	0.10
Ethanol	2.3	0.10
Propan-2-ol	1.7	0.10
Butan-1-ol	0.8	0.06
Hexan-1-ol	0.0	0.0

---

Catalysis reactions

All the catalyst reactions were performed with the PDMS micro reactor, since it offered the advantages of a straightforward catalyst immobilisation method and operated with open channels thus reducing the resistance to flow.

Dehydration of hexanol

Initial studies of the influence of temperature on dehydration efficiency using a larger reactor¹² found the optimum to be 155 °C, which was in close agreement with literature values for this reaction on the industrial scale. When an optimisation study was carried out with the micro reactor, an identical result was obtained. The influence of temperature was significant, and reducing the temperature by 10 °C lead to a significant decrease in conversion efficiency. Operating at a temperature of 155 °C, the hexan-1-ol was pumped through the reactor chip at a flow rate of 3 µL min⁻¹ using a syringe pump. The products were collected and analysed by gas chromatography using a Porapak Q column at 220 °C. During a typical run time of 3 h, samples were collected every 30 min. From the gas chromatogram, a hexanol to hexene conversion of between 85 and 95% could be determined. No additional products were detected. The conversion efficiency is favourable compared with that observed with large scale reactors, where 30% is common. The reactor was used constantly over three working days, and no degradation of the performance was observed. However, without the thorough removal of oxygen, rapid coking occurred leading to a reduction in conversion to 10% in under 10 min. The reproducibility between different catalyst top plates was better than 95%, this was due to the reproducible production method for the catalyst top plate and to the fact that the channel plate was reusable (however, different channel plates were identical if taken from the same batch). In addition, the use of a torque wrench to tighten the bolts on the clamp resulted in a highly reproducible system.

Dehydration of ethanol

Ethanol has been pumped by both electroosmotic flow and by means of a syringe pump. Initially, ethanol was pumped through the chip at a flow rate of 3 µL min⁻¹ using a syringe pump and a reaction temperature of 155 °C (from an optimisation study similar to that for hexanol). The products were collected every 30 min and analysed by gas chromatography using a Porapak Q column at 50 °C. Only trace amounts of ethanol were detected, with most of the feedstock converted to ethene (68%), ethane (16%), and methane (15%). The electroosmotic pump produced repeatable flow rates of 0.9–1.1 µL min⁻¹ at a field strength of 200 V cm⁻¹, giving a significantly greater residence time for ethanol in the reactor. The effect of this could clearly be seen in the GC results obtained where the only detectable product was methane. This indicates that the reaction has gone beyond dehydration to complete cracking of the feedstock. Trace amounts of methanol were also detected. The ability to almost totally crack ethanol to methane may prove very important to the production of chemical feedstocks and energy production, for although the present system is energy intensive, developments of further miniaturised systems may reduce energy demand.

Esterification of ethanoic acid

The esterification of an alcohol and carboxylic acid has historically been performed using homogeneous catalysis (i.e., the addition of concentrated sulfuric acid to a mixture of ethanol and ethanoic acid). Since the reaction is an equilibrium, the reaction can reverse in presence of the catalyst, so using a solid catalyst should limit the reverse reaction. This concept was examined using the catalyst reactor for the production of ethyl acetate from ethanol and ethanoic acid. A 1 : 1 v/v mixture of ethanol and ethanoic acid was pumped through the Z-reactor by a syringe pump at a flow rate of 2 µL min⁻¹ over a series of reaction temperatures from room temperature to 180 °C, and the production of ethyl acetate was determined by gas chromatography. The results can be seen in Fig. 2, and clearly show that there is an activation temperature below which no reaction occurs. The reactor was used constantly over a 5 day period, and no degradation was seen in performance. Also, coking was not observed, a usual occurrence with this catalyst.

Fig. 2 The effect of temperature on yield of ethyl acetate (each data point is obtained from a mean of three 1 h experiments performed in a random order).

Conclusions

We have demonstrated the use of a micro reactor for heterogeneous catalysis. We have shown that highly efficient reactions are possible without the need for high pressure devices, and that polymeric reactors are suitable for quite demanding reactions. Perhaps of great significance is that while micro reactors have great potential, as far as catalytic reactions are concerned the thermodynamics of reaction, i.e., activation temperature, is independent of the scale of the reaction.

Acknowledgements

The authors would like to express their thanks to Mel Chemicals for supplying the sulfated zirconia catalyst, and to Terry Aspinwall and Mike Bailey for technical help with the chip production. NGW was funded by the University of Hull.

References

1. A. G. Hadd, S.C. Jacobson and J. M. Ramsey, Anal. Chem., 1999, 71, 5206.
2. Y. H. Chen, W. C. Wang, K. C. Young, T. T. Chang and S. H. Chen, Clin. Chem., 1999, 45(11), 1938.
3. G. M. Greenway, S. J Haswell and P. H. Petsul, Anal. Chim. Acta, 1999, 387, 1.
4. G. N. Doku and S. J. Haswell, Anal. Chim. Acta, 1999, 382, 1.
5. J. Wang, B. Tian and E. Sahlin, Anal. Chem., 1999, 71, 5436.
6. J. Li, P. Thibault, N. H. Bings, C. D. Skinner, C. Wang, C. Colyer and J. Harrison, Anal. Chem., 1999, 71, 3036.
7. B. He, L. Tan and F. Regnier, Anal. Chem., 1999, 71, 1464.
8. N. G. Wilson and T. McCreedy, Chem. Commun., 2000, 9, 733.
9. M. Nitta, H. Sakoh and K. Aomura, Appl. Catal., 1984, 10, 215.
10. D. Das and D. K. Chakrabarty, Energy Fuels, 1998, 12, 109.
11. A. Sarkar, Indian J. Chem., 1996, 35B, 862.
12. P. D. Christensen, S. W. P. Johnson, T. McCreedy, V. Skelton and N. G. Wilson, Anal. Commun., 1998, 35, 341.
13. T. McCreedy, Anal. Chim. Acta, 2000, 427, 39.

---