Elemental fluorine

Part 16. Versatile thin-film gas–liquid multi-channel microreactors for effective scale-out

Abstract

Versatile and conveniently operated microreactor devices, suitable for long term chemical synthesis for both laboratory and industrial use, are described. Direct fluorination of ethyl acetoacetate by fluorine gas is used as a model reaction to illustrate the successful application of reactors having either 9, 18 or 30 microchannels, with all channels supplied from single reagent sources, to provide significant quantities of fluorinated products. Convenient reactor maintenance, simple design, versatility and the enhanced safety features of using such modular microreactor devices, make these systems appropriate for synthesis on the large scale.

---

Introduction

Many important health-care and plant protection products contain carbon–fluorine bonds² and a variety of polyfluorinated compounds are extremely important, notably as monomers for polymerisation.^3,4 However, regioselective introduction of fluorine atoms into organic systems can be very difficult and, over the years, many procedures and fluorinating agents that enable functional group interconversion to form carbon–fluorine bonds have been developed in attempts to meet this research challenge.^5,6 In particular, methodology continues to develop for selective fluorination, where a number of reagents providing a source of electrophilic fluorine have found widespread use. Indeed, there are some excellent reagents of the N–F class,⁷ such as Selectfluor™, that are now available that act as effective sources of electrophilic fluorine and these are widely used in synthesis by organic chemists for a growing number of fluorination reactions. However, these reagents are all synthesised from elemental fluorine and, therefore, correspondingly expensive. Consequently, the direct use of elemental fluorine would be highly desirable⁸ and would be both efficient and environmentally friendly for the purpose of carbon–fluorine bond formation, because the hydrogen fluoride by-product can be recycled by electrolysis (Fig. 1).

Fig. 1 Electrophilic fluorination process.

Indeed, the technology for generation of fluorine on a large scale is highly refined in the nuclear energy industry but, so far, this world-wide major investment has only been utilised for the manufacture of uranium hexafluoride. Of course, a feature limiting the use of fluorine directly for synthesis of fine chemicals is the safe handling and control of such thermodynamically exothermic processes⁹ on an industrially viable scale. Nevertheless, many of the problems associated with the direct use of fluorine on large scale have been shown to be surmountable in the laboratory^8,10,11 and, indeed, manufacturing processes involving fluorine are currently in operation, but improvements in safe-handling and control of such processes are still keenly sought after.

Outside of industry, chemists involved in synthesis are generally reluctant to use flow systems of any type for a variety of reasons but there is a rapidly growing case to demonstrate that microreactors offer tremendous opportunities for laboratory as well as industrial syntheses.^12–16 In particular, application of microreactors to direct fluorination could bring considerable benefits to laboratory and manufacturing processes where: (a) low inventories of fluorine in contact with reactants provides a major safety feature, (b) two phase gas/liquid mixing is very efficient, (c) there is, potentially, very effective heat exchange and, as we shall see, (d) very large gas/liquid surface : volume ratios. Consequently, an excellent gas/liquid interface is possible. Furthermore, very important is the fact that what happens in the laboratory, will also occur on the plant scale, because ‘scale-out’ rather than ‘scale-up’, would apply. That is, a manufacturing process would be constructed of many of the same laboratory reactors used in parallel. Of course, reactor devices could also be mounted in series with each stage using different operating conditions, for example, staged temperature increases.

We should emphasise that, although much had been written on microreactor technology, at the beginning of the work in Durham,^17,18 nothing very helpful concerning gas/liquid phase reactors was available in the literature. Consequently, however, considerable interest has been demonstrated in the literature in recent years in the development of microreactor techniques and some of these have been applied to direct fluorination of organic compounds. Indeed, procedures involving falling film reactors, micro-bubble columns and micro-structured systems have all been used for this purpose.^19,20

Much of the driving force for the development of microreactors has derived from the idea of harnessing the techniques of the microchip industry for the mass-production of chemical reactors, a great number of which, in principle, could be linked in parallel to effectively create a manufacturing process. However, we decided at an early stage of our work that this approach would not be appropriate for our purposes and our reasoning was based on the following issues: (a) micro-channels are easily blocked in chemical synthesis and access for routine maintenance is, therefore, essential; (b) although, in principle, the manufacture of thousands of microchip-type reactors could be relatively inexpensive, the reality is that development of single ‘limited edition’ reactors is hugely expensive because designs cannot easily be modified. Therefore, we have adhered to using only common and inexpensive mechanical workshop techniques for the design and construction of our microreactors, throughout our work in this area.

Our first successful reactor (Fig. 2)¹⁸ consisted of a single groove cut into a nickel block and the groove was covered using polychlorotrifluoroethylene (PCTFE) sheet, which is relatively inert to fluorine and is, helpfully, transparent.

Fig. 2 Prototype single channel reactor.

This device enabled us to establish the flow regime that results in this system which we term ‘pipe flow’, Fig. 3, whereby a film of the liquid flows across the surface of the microchannel, with the gas flowing through the centre.

Fig. 3 Pipe and slug flow.

A ‘pipe’ may have different cross-sectional shapes such as circular, square, as in this case, or triangular, depending upon the construction of the channel and the term ‘pipe flow’ encompasses all of these possibilities. With the benefit of hindsight, pipe-flow is the logical outcome because of the opposing beneficial pressure P of the gas, and the viscous drag on the liquid adhering to the sides of the capillary. Of course, ‘pipe flow’ is the best possible scenario to obtain excellent surface ∶ volume ratios, with the attendant efficient mixing and heat transfer, rather than the alternative ‘slug flow’ in which surface contact areas are very small.

We established that this simple microchannel device could be used very effectively for a number of direct fluorination processes and, during the course of use, that the integrity of the metal fluoride coating, formed by the essential careful passivation of the metal surface in the capillary with fluorine, remained intact for most of the reactant systems that we have used so far.

The question now arises of how such a simple design may be adapted and developed for realistic use by industry for a manufacturing process. Clearly, we could have many channels on the microreactor nickel plate and we have, indeed, used such a multi-channel plate (Fig. 4)¹⁸ successfully for direct fluorination processes. However, this design is not realistic for use outside the laboratory because each channel requires a separate feed system for each reactant and multiple outlet ports, resulting in a largely unmanageable system that involves extensive and complicated fittings which, in the long term, are prone to leaks, blockings and breakages.

Fig. 4 Prototype three channel microreactor device.

The key question is how to supply each channel on a multi-channel plate from the same source ‘reservoirs’ to minimise connections, while maintaining uniform pressure to each channel, using a design that is suitable for laboratory or industrial use. Our first step towards solving this problem was to design and construct an external intermediate reservoir (Fig. 5) in which the substrates were passed via either a syringe pump or gas mass flow controller into large volume substrate reservoirs that have outlet tubes feeding all the microreactor channels simultaneously (Fig. 5).

Fig. 5 Above: Cylindrical external reservoir. Below: Use of reservoirs with a three channel microreactor.

In order to minimise pressure variations, the volume of the cylindrical reservoir must be significantly larger than the volume of the microchannel inlets for uniform flow to be possible. We confirmed that flows of liquid from each channel outlet for this microreactor system were equal, within experimental error, by measuring the amount of eluant from each channel over a period of time. Although this microreactor demonstrated that the reservoir concept to scale-up was possible, the device would still be too complex for long term use. Therefore, we sought to adapt this reservoir idea by developing a microreactor in which the reservoirs were located in positions directly attached to the microchannel substrate inlets, thus precluding the need for large amounts of unnecessary tubing and fittings.

In this paper, we describe a device that addresses this very important scale-out problem for both laboratory and large scale synthesis.

Results and discussion

After much development, we designed and constructed a modular microreactor system^21,22 and the essential elements of the new design are shown schematically in Fig. 6. The design allows for many channels to be supplied from single reservoir sources and a multichannel device to be constructed in a simple manner from a disposable channel plate. A reactor base block (Fig. 7), constructed by machining appropriate fittings from stainless steel using standard mechanical workshop techniques, is the heart of this microreactor system. Fluid ‘reservoirs’ (Fig. 7) were drilled into the stainless steel block and slots, cut through the top of the block to the reservoir, allow fluids to flow from the reservoirs through a nickel or stainless steel sealing plate to enter the channels that form the ‘channel plate’. The channel plates (Fig. 8) are constructed separately from stainless steel sheet (0.5 mm thickness) using standard electron discharge machining techniques and may have any number of channels which, in this case, are each typically 0.5 mm wide. Of course, the depth of the reaction channels are dictated by the thickness of the channel plate (0.5 mm). Obviously, more reactants could be introduced, if required, by having more reservoirs as part of the reactor ‘base block’. Each reservoir affords a volume that is substantially greater than the sum of the volumes of the channel paths and this ensures equal and accurate fluid distribution into each of the channels. After the reagents flow down the channels and reaction occurs, the product is collected in a single reservoir that connects all channels. A transparent PCTFE and stainless steel top plate complete the modular design and all plates are compressed together by screw fittings and positioned vertically when in operation.

Fig. 6 Schematic representation of modular microreactor device.

Fig. 7 Reactor base block with reservoirs, slots, screw, support and cooling holes.

Fig. 8 Channel plates with 9 (left) and 30 channels (right) respectively.

The reactor and channel plates may be each manufactured simply and inexpensively. Furthermore, the modular device may be assembled and disassembled quickly, permitting simple replacement of the channel plates and routine maintenance of the device, if required.

The completed device is shown in Fig. 9, in which the reactor block is fitted with cooling/heating coils that pass through the body of the steel base reactor block to enable effective reactor temperature control and placed in a vertical orientation. Of course, other methods of cooling could also be employed if necessary.

Fig. 9 Photograph of microreactor device fitted with a 30-channel plate.

Following the construction of the reactor device and careful passivation of the internal surfaces by passing fluorine in nitrogen in increasing concentrations (10–50%) through the device, we carried out a number of fluorination reactions of a model substrate,²³ ethyl 3-oxobutanoate, in order to establish the effectiveness of this microreactor system for gas/liquid reactions. This reaction is ideal for assessing the performance of microreactors because both the starting material and all fluorinated products are completely soluble in formic acid. A series of reactions were carried out to obtain the optimum reaction conditions before scale-out was attempted.

Fluorine gas, as a 10% mixture in nitrogen, was introduced into the top reservoir by a mass flow meter and, concurrently, a solution of ethyl 3-oxobutanoate in formic acid was passed into the lower substrate reservoir by syringe pump. After all reagents had passed through the microreactor which, in this first series of reactions, incorporated a 9-channel reactor plate, the product mixture was collected in a PTFE vessel from the product reservoir outlet. The experimental set-up is shown schematically in Fig. 10.

Fig. 10 Schematic representation of fluorination reaction equipment.

We observed that the flow regime was indeed ‘pipe flow’, as seen in our earlier reactor device and, also, it appeared that each channel was evenly supplied with the same flow of substrate mixtures giving high conversion of the starting material to several mono- and di-fluorinated products (Scheme 1).

Scheme 1 Fluorination of ethyl acetoacetate.

The effect of substrate concentrations, reactant ratio, temperature and substrate flow on the product yields and distribution were assessed by a series of fluorination reactions that were carried out using the 9-channel reactor device, with the aim of optimising the most convenient conditions for the synthesis of mono-fluorinated product 2. The results of this series of fluorination reactions are collated in Table 1, where conversions and yields of products 2–6 are listed. All products 2–6 were identified by multinuclear (¹H, ¹³C and ¹⁹F) NMR spectroscopy²³ and confirmed by NMR shifts computed from MP2/6-31G* optimised geometries of 2–6. ‘Pipe flow’ was observed in the microchannels in all cases and, since each reaction can be assessed very quickly and simply by ¹⁹F NMR and GC-MS, optimisation of the fluorination reaction using this microreactor device could be performed very conveniently.

Table 1 Fluorinations of ethyl 3-oxobutanoate 1 using the 9-channel microreactor and formic acid as the solvent

Entry	Solvent : substrate molar ratio	F2 : substrate molar ratio	conv. (%)	2	3	4	5	6	Remarks
a Conditions used unless otherwise stated: 10% F2/N2 at 10 mL min^−1 channel^−1, 10 °C. Substrate flow quoted in mL h^−1 channel^−1. b Percentages (%) are determined from ^19F and ^1H NMR data. c Percentages of 2–6 are based on product mixture.
1	Neat substrate	1.0	52	41	25	6	22	6
2	1 ∶ 1	1.0	55	43	20	5	26	6
3	1 ∶ 1	1.4	83	46	23	6	20	5
4	2 ∶ 1	1.2	34	67	22	3	7	2
5	4 ∶ 1	1.0	53	69	17	3	9	2	7 mL min^−1 F2
6	4 ∶ 1	1.4	61	66	16	3	12	3
7	4 ∶ 1	1.4	59	68	15	3	11	3	30 channels
8	4 ∶ 1	1.4	54	69	16	2	12	2	18 channels
9	8 ∶ 1	2.4	80	73	12	3	10	2
10	8 ∶ 1	3.5	95	74	13	2	7	2
11	8 ∶ 1	4.4	100	65	15	2	9	2
12	8 ∶ 1	5.7	100	51	15	2	18	4
13	8 ∶ 1	9.2	100	22	17	2	30	5
14	16 ∶ 1	3.7	94	79	9	4	7	2	At 20 °C
15	16 ∶ 1	4.1	74	76	10	3	9	2
16	16 ∶ 1	7.0	96	66	10	2	15	4
17	16 ∶ 1	11.4	100	25	15	2	23	4
18	16 ∶ 1	4.4	87	73	13	2	8	2	5% F2 used
19	32 ∶ 1	1.0	15	81	8	2	7	2	At 20 °C
20	32 ∶ 1	2.0	27	80	9	2	7	2	At 20 °C
21	32 ∶ 1	4.1	54	76	9	3	10	2	At 20 °C
22	32 ∶ 1	8.1	85	81	9	3	6	1	At 20 °C
23	32 ∶ 1	7.4	66	79	10	2	7	2

---

The effects of changing a reaction parameter on the product profile may be deduced from inspection of the data collected in Table 1. In general, as the substrate solution becomes more dilute, the yield of mono-fluorinated product 2 increases as might be expected. For example, compare entries 4, 6, 9, 17 and 22, where the yields of 2 rise from 67 to 81% with increasing dilution. Furthermore, increasing the fluorine : substrate ratio leads to higher conversion of substrate and higher yields of difluorinated by-products (compare entries 9 and 13), again as might be anticipated.

Purification of the monofluoro product 2 proved to be most convenient from reactions where 100% conversion was achieved. In the laboratory, 2,4-difluoro product 3 is separated by extraction into water whilst 2 and difluoro compounds 4, 5 and 6 were isolated by fractional distillation. Consequently, high isolated yields of monofluoro product 2 may be obtained by using the conditions listed in entry 11.

We reexamined direct fluorination of ethyl 3-oxobutanoate 1 in formic acid in standard ‘bulk’ conditions,²³ by passing fluorine gas, diluted in nitrogen, through a stirred solution of 2 in formic acid over a period of 16 h, to give 2 in 62% yield. In the case of this ‘bulk’ fluorination, the substrate : formic acid ratio was 1 ∶ 44, substrate ∶ fluorine ratio 1 : 3 and conversion was 100%. 5 and 6 were observed by ¹⁹F NMR and account for less than 0.5% yield of the product mixture whereas a more significant yield of 4 (12%) was observed. In comparison to the fluorination reactions carried out in the microreactor (Table 1), therefore, the very low concentration of substrate in formic acid for ‘bulk’ processes leads to a reduction in the quantity of difluorinated products 5 and 6 obtained.

The mechanism of fluorination of β-ketoesters is thought to be an electrophilic addition reaction of fluorine to the enol derivatives of the corresponding keto-ester. Products 2, 3 and 4 are all formed by this process while 5 and 6 are formed most likely by electrophilic aliphatic substitution of relatively electron rich carbon-hydrogen bonds via a 3-centre 2-electron bond pathway.^11,24

The 9-channel microreactor device has been in operation in our laboratories for many months and, indeed, over 700 g of crude 2 has been synthesised after repeated reactions using the same device. Continuous operation of the microreactor for the fluorination of 1 over an extended time period (150 h) has been assessed without any decrease in yield or conversion over the course of the reaction. Scale-out to 30 channels, the reasonable limit of our laboratory supply of fluorine, has also been achieved by simply exchanging a 9-channel for a 30-channel plate (Fig. 8), again, with no appreciable change in overall yield and conversion (Table 1, entry 7) Furthermore, a ‘two sided’ microreactor device, in which the fluid reservoirs feed two channel plates (e.g. 2 × 9-channel plates) located on both sides of the stainless steel reactor block, can also be operated continuously over extended periods (Table 1, entry 8).

We have demonstrated that a 30-channel microreactor device may be operated continuously for the fluorination of 1 from single reagent sources and it is instructive, at this stage, to consider the implications of using such a device for large scale synthesis. Typically, the microreactor devices described here produce ca. 0.2 g of product per hour per channel. Consequently, if operated continuously, a two-sided 30-channel device, having a total of 60 channels, would be capable of providing ca. 300 g per day. By scale-out, 10 reactors would, therefore, provide ca. 3 kg of product per day and this would be a realistic pilot-plant with the relatively low expenditure and operating just as laboratory experiments would predict. The ease of convenient maintenance, continuous operation, single source reservoir fluid delivery, convenient purification by distillation and improved safety features are major advantages of using these microreactor devices for such purposes.

Experimental

All starting materials were obtained commercially (Aldrich, Lancaster or Fluorochem). All solvents were dried using literature procedures. NMR spectra were recorded in deuteriochloroform, unless otherwise stated, on a Varian VXR 400S NMR spectrometer operating at 400 MHz (¹H NMR), 376 MHz (¹⁹F NMR) and 100 MHz (¹³C NMR) with tetramethylsilane and trichlorofluoromethane as internal standards. Mass spectra were recorded on a Fisons VG-Trio 1000 Spectrometer coupled with a Hewlett Packard 5890 series II gas chromatograph using a 25 m HP1 (methyl-silicone) column. Elemental analyses were obtained on a Exeter Analytical CE-440 elemental analyser. Accurate mass measurements were performed at the EPSRC National Mass Spectrometry Service Centre. Melting points and boiling points were recorded at atmospheric pressure unless otherwise stated and are uncorrected. Column chromatography was carried out on silica gel (Merck no. 109385, particle size 0.040–0.063 mm) and TLC analysis was performed on silica gel TLC plates (Merck). Fractional distillation was performed on a Fisher Spahltrohr MS 225 apparatus and boiling points are uncorrected.

Fabrication of the multi-channel microreactor device

Two substrate reservoirs (40 mm and 30 mm diameter) and one product collection reservoir (20 mm diameter) were cut into a stainless steel block, measuring 96 mm by 250 mm by 50 mm. A further five holes (20 mm diameter) were also drilled through the block to allow the device to be supported when in use. Slots were cut into the block surface by electric discharge machining (EDM) as shown in Fig. 7. Entry and exit openings to the reservoirs were drilled perpendicular to the substrate and product collection reservoirs. Cooling channels were drilled through the block and stainless steel tubing were attached by silver solder. Screw holes were drilled into the block and Swagelok^® fittings were inserted into the gas and liquid reservoirs and to all of the reservoir openings.

The nickel bottom plate was fabricated from 2 mm thick nickel sheet, while slots in this plate were machined by EDM. The channel plates were fabricated with the appropriate number of channels (typically, 3, 9, 30 channels, see Fig. 8) from 0.5 mm thick stainless steel sheet by the use of chemical etching and standard drilling techniques.

The top sealing plate was fabricated from PTFCE sheet (6 mm thick). The cover plate was fabricated from a 2 mm thick stainless steel sheet and viewing windows were machined into this cover sheet using a milling machine. The reverse side of the reactor was sealed with a 6 mm PTFCE sheet and 2 mm thick stainless steel sheet. All other sheet dimensions were analogous to the size of the reactor surface. Gaskets were made from copper (0.15 mm thick), lead (0.15 mm) or PTFE sheet (0.05 mm); the copper gaskets were subsequently annealed prior to use by heating to redness and then plunging into a methanol bath immediately. Stainless steel screws of 4 mm diameter were used to compress the plates to the reactor block in an air-tight seal. A torque wrench was used to ensure that equal pressure was applied to each screw.

Prior to use, the reactor was leak tested and passivated using 10%, 20%, and 50% F₂/N₂ mixtures. When in use, the microreactor was held in a vertical position by two stainless steel bars, which were inserted through appropriate holes in the block and clamped at either end.

Direct fluorination of ethyl acetoacetate

General procedure. After purging the apparatus with nitrogen, fluorine, as a 10% mixture in nitrogen (v ∶ v), was passed through the microreactor via an inlet port at a prescribed flow rate that was controlled by a mass flow controller (Brooks^® Instruments). The microreactor and the collection vessel were cooled by an external cryostat to the appropriate temperatures. Substrate mixture was injected by a mechanised syringe pump into the microreactor channel at a prescribed flow rate through the substrate inlet port. After passing through the microreactor and outlet port, excess fluorine gas and volatile waste products were passed through a scrubber filled with soda lime. Liquid products were collected in an FEP tube and the crude product mixture was examined by ¹⁹F NMR spectroscopy. The product mixture was added to water and extraction of the products from the aqueous layer into dichloromethane was continued until ¹⁹F NMR spectroscopy showed no peaks in the aqueous layer corresponding to the 2,4-fluoro derivative 3. Washing the dichloromethane layer with saturated NaHCO₃ solution, drying (MgSO₄) and evaporation of the solvent gave an oil which was analysed by ¹⁹F and ¹H NMR spectroscopy and GC-MS. The composition of a weighed crude reaction mixture was determined by ¹H NMR and GC-MS analysis and the conversion of starting material was calculated from ¹H NMR and GC peak intensities. The amount of fluorinated derivative in the crude product was determined by adding a known amount of fluorobenzene to a weighed amount of the crude product mixture. Comparison of the relative intensities of the appropriate ¹H and ¹⁹F NMR resonances gave the yield of fluorinated derivative, based upon the conversion obtained above.

Purification of products was carried out by vacuum fractional distillation to yield pure difluoro derivative 4 (0 °C at 0.01 mmHg) and the monofluoro product 2 (20 °C at 0.01 mmHg). The undistilled residue was dissolved in dichloromethane and washed with a 10% K₂CO₃ solution. The aqueous layer was then extracted with several portions of fresh dichloromethane and evaporation and drying (MgSO₄) gave compound 3. For a mixture of 5 and 6, the dichloromethane layer previously washed with K₂CO₃ was washed with water, dried over MgSO₄ and evaporated. NMR and MS data for the five products are given below.

Ethyl-2-fluoro-3-oxobutanoate. Enol form: δ_H 1.28 (t, ³J_HH 7.0, 3H, CH₂CH₃), 2.03 (d, ⁴J_HF 4.6, C(OH)CH₃), 4.27 (q, ³J_HH 7.0, 2H, CH₂CH₃), 10.01 (d, ⁴J_HF 4.4, OH); δ_C 13.9 (s, CH₂CH₃), 14.4 (s, C(OH)CH₃), 61.7 (s, CH₂CH₃), 134.5 (d, ¹J_CF 227.2, CF), 159.5 (d, ²J_CF 27.9, C(OH)CH₃), 166.2 (d, ²J_CF 24.6, CO₂Et); δ_F −179.2 (quintet, ⁴J_HF ∼4.1); Keto form: δ_H 1.28 (t, J_HH 7.0, 3H, CH₂CH₃), 2.30 (d, ⁴J_HF 3.5, COCH₃), 4.27 (q, J_HH 7.0, 2H, CH₂CH₃), 5.17 (d, J_HF 49.5, H, CHF); δ_C 13.5 (s, CH₂CH₃), 25.6 (s, COCH₃), 62.5 (s, CH₂CH₃), 91.1 (d, ¹J_CF 196.9, CF), 163.9 (d, ²J_CF 23.3, CO₂), 198.9 (d, ²J_CF 23.4, COCH₃). δ_F −193.5 (dq, ²J_HF 49.5, ⁴J_HF 4.5); m/z (EI⁺) 148 (M⁺, 10%), 106 (76), 78 (95), 43 (CH₃CO, 100).
Ethyl-2,4-difluoro-3-oxobutanoate. δ _H 1.33 (t, J_HH 7.2, 3H, CH₂CH₃), 4.33 (q, ³J_HH 7.2, 2H, CH₂CH₃), 5.21 (apparent d, J_HF 46.8, 2H, CH₂F), 5.45 (d, J_HF 48.0, 1H, CHF); δ_C 13.5 (s, CH₂CH₃) 62.5 (s, CH₂CH₃), 83.2 (dd, J_CF 184.8, J_CF 2.9, CH₂F), 89.7 (d, ¹J_CF 194.7, CF), 163.1 (d, ²J_CF 22.7, CO₂), 195.2 (dd, ²J_CF 22.7, ²J_CF 17.5, COCH₃); δ_F −203.9 (d, ²J_HF 48.5, CHF), −236.2 (td, ²J_HF 46.5, ⁴J_HF 1.9, CH₂F); m/z (EI⁺) 166 (M⁺, 20%), 140 (37), 121 (CFH₂COCFHCO⁺, 98), 93 (CFH₂COCFH⁺, 97), 60 (CFH₂CO⁺, 100).
Ethyl-2,2-difluoro-3-oxobutanoate. δ _H 1.36 (t, J_HH 7.1, 3H, CH₂CH₃), 2.41 (t, J_HF 1.5, 3H, COCH₃), 4.38 (q, ³J_HH 7.1, 2H, CH₂CH₃); δ_C 13.2 (s, CH₂CH₃), 23.4 (s, COCH₃), 62.0 (s, CH₂CH₃), 108.0 (t, ¹J_CF 264.0, CF), 160.6 (t, ²J_CF 30.7, CO₂), 194.2 (t, ²J_CF 28.2, COCH₃); δ_F −114.2 (s); m/z (EI⁺) 166 (M⁺, 6%), 121 (34), 115 (48), 87 (50), 61 (54), 28 (CO, 100).
2′-Fluoroethyl-2-fluoro-3-oxobutanoate. δ _H 2.35 (d, J_HF 3.8, 3H, COCH₃), 4.60–4.45 (apparent pattern of four quartets, 4H, CH₂CH₂F), 5.23 (d, J_HF 49.2, 1H, CHF); δ_C 25.6 (s, COCH₃), 64.9 (d, 19.1, CH₂CH₂F), 80.5 (d, ¹J_CF 170.5, CH₂CH₂F), 90.9 (d, 196.8, CFH), 163.8 (d, ²J_CF 24.1, CO₂), 198.6 (d, ²J_CF 22.7, COCH₃); δ_F −194.2 (dq, J_HF 48.6, J_HF 4.5, CHF), −225.6 (m, CFH₂); m/z (EI⁺) 166 (M⁺, 6%), 103 (75), 78 (59), 47 (CH₂CH₂F⁺, 70), 43 (CH₃CO⁺,100).
1′-Fluoroethyl-2-fluoro-3-oxobutanoate. δ _H 1.55 (apparent d, J_HF 20.8, 3H, CHFCH₃), 2.35 (d, J_HF 3.8, 3H, COCH₃), 5.20 (d, J_HF 49.2, 1H, CHF), 6.48 (apparent d, J_HF 54.8, 1H, CHFCH₃); δ_C 19.0 (d with second order effects, ²J_CF 23.4, CHFCH₃), 25.7 (s, COCH₃), 90.7 (d, 198.3, CFH), 104.2 (d with second order effects, ¹J_CF 224.0, CFHCH₃), 164.8 (d, ²J_CF 26.4, CO₂), 201.0 (d, ²J_CF 28.5, COCH₃); δ_F −123.3 (m, CFHCH₃), −194.8 (s, CFH).

Calculations–computed NMR data (GIAO-B3LYP/6-311G*//MP2/6-31G*)

All ab initio computations were carried out with the Gaussian 98 package. All geometries discussed here were optimised at the HF/6-31G* level of theory with no symmetry constraints unless otherwise stated. Frequency calculations were computed on these optimised geometries at the HF/6-31G* level for imaginary frequencies—none were found in all cases. Optimisation of these geometries were then carried out at the MP2/6-31G* level of theory. Calculated NMR shifts at the GIAO-B3LYP/6-311G* level were obtained from MP2-optimized geometries. Theoretical ¹³C and ¹H chemical shifts at the GIAO-B3LYP/6-311G*//MP2/6-31G* level were referenced to TMS: δ(¹³C) = 178.81 − σ(¹³C); δ(¹H) = 32.28 − σ(¹H). ¹⁹F chemical shifts were referenced to HF and converted to the usual CFCl₃ scale; δ(¹⁹F) = 188.0 − σ(¹⁹F). Relative energies were computed at the MP2/6-31G* level with ZPE (calculated at HF/6-31G*) corrections scaled by 0.89. Cartesian coordinates and total energies of the MP2/6-31G* optimized geometries described in this study are in the ESI.‡

Ethyl-2-fluoro-3-oxobutanoate. Enol form: δ_H 1.42 (CH₂CH₃), 2.05 (C(OH)CH₃), 4.43 (CH₂CH₃), 9.83 (OH); δ_C 13.4 (s, CH₂CH₃), 10.9 (s, C(OH)CH₃), 64.4 (CH₂CH₃), 136.3 (CF), 168.9 (C(OH)CH₃), 167.5 (CO₂Et); δ_F −177.0; Keto form: δ_H 1.47 (CH₂CH₃), 2.41 (COCH₃), 4.65 (CH₂CH₃), 5.87 (CHF); δ_C 13.1 (CH₂CH₃), 23.1 (COCH₃), 65.9 (CH₂CH₃), 94.9 (CF), 167.1 (CO₂), 204.4 (COCH₃); δ_F −197.1.

Ethyl-2,4-difluoro-3-oxobutanoate. δ _H 1.45 (CH₂CH₃), 4.41 (CH₂CH₃), 5.20 (CH₂F), 6.02 (CHF); δ_C 13.8 (CH₂CH₃) 65.7 (CH₂CH₃), 85.5 (CH₂F), 86.8 (CF), 168.1 (CO₂), 204.7 (COCH₃); δ_F −211.3 (CHF), −240.7 (CH₂F).

Ethyl-2,2-difluoro-3-oxobutanoate. δ _H 1.44 (CH₂CH₃), 2.38 (COCH₃), 4.66 (CH₂CH₃); δ_C 13.7 (CH₂CH₃), 23.7 (COCH₃), 65.3 (CH₂CH₃), 110.0 (CF), 166.1 (CO₂), 199.8 (COCH₃); δ_F −104.7.

2′-Fluoroethyl-2-fluoro-3-oxobutanoate. δ _H 2.35 (COCH₃), 4.40 (CH₂CH₂F), 4.61 (CH₂CH₂F), 5.61 (CHF); δ_C 23.3 (COCH₃), 63.0 (CH₂CH₂F), 78.9 (CH₂CH₂F), 91.9 (CFH), 169.1 (CO₂), 199.8 (d, COCH₃); δ_F −200.5 (CHF), −237.1 (CFH₂).

1′-Fluoroethyl-2-fluoro-3-oxobutanoate. δ _H 1.38 (CHFCH₃), 2.35 (COCH₃), 5.60 (CHF), 6.55 (CHFCH₃); δ_C 15.4 (CHFCH₃), 25.3 (COCH₃), 88.3 (CFH), 101.0 (CFHCH₃), 166.5 (CO₂), 201.3 (COCH₃); δ_F −116.9 (CFHCH₃), −196.5 (CFH).

Acknowledgements

We thank EPSRC, Asahi Glass Co. and EPSRC Crystal Faraday Partnership for funding.

References

1. R. D. Chambers, D. Holling, G. Sandford, A. S. Batsanov and J. A. K. Howard, J. Fluorine Chem., 2004, 125, 661.
2. J. T. Welch and S. Eswarakrishnan, Fluorine in Bioorganic Chemistry, John Wiley and Sons, New York, 1991.
3. Organofluorine Chemistry. Principles and Commercial Applications, ed. R. E. Banks, B. E. Smart and J. C. Tatlow, Plenum, New York, 1994.
4. R. D. Chambers, Fluorine in Organic Chemistry, Blackwell, Oxford, 2004.
5. J. A. Wilkinson, Chem. Rev., 1992, 92, 505.
6. Houben-Weyl Organofluorine Compounds, ed. B. Baasner, H. Hagemann and J. C. Tatlow, Thieme, Stuttgart, 2000.
7. G. S. Lal, G. P. Pez and R. G. Syvret, Chem. Rev., 1996, 96, 1737.
8. J. Hutchinson and G. Sandford, Top. Curr. Chem., 1997, 193, 1.
9. R. J. Lagow and J. L. Margrave, Prog. Inorg. Chem., 1979, 26, 161.
10. S. T. Purrington, B. S. Kagen and T. B. Patrick, Chem. Rev., 1986, 86, 997.
11. S. Rozen, Acc. Chem. Res., 1988, 21, 307.
12. H. Lowe and W. Ehrfeld, Electrochim. Acta, 1999, 44, 3679.
13. S. J. Haswell and V. Skelton, Trends Anal. Chem., 2000, 19, 389.
14. S. J. Haswell, R. J. Middleton, B. O'Sullivan, V. Skelton, P. Watts and P. Styring, Chem. Commun., 2001, 391.
15. O. Worz, K. P. Jackel, T. Richter and A. Wolf, Chem. Eng. Sci., 2001, 56, 1029.
16. K. F. Jensen, Chem. Eng. Sci., 2001, 56, 293.
17. R. D. Chambers and R. C. H. Spink, Chem. Commun., 1999, 883.
18. R. D. Chambers, D. Holling, R. C. H. Spink and G. Sandford, Lab Chip, 2001, 1, 132.
19. K. Jahnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Lowe, C. Wille and A. Guber, J. Fluorine Chem., 2000, 105, 117.
20. V. Hessel, P. Lob and H. Lowe, Chim. Oggi, 2004, 22, 10 and references therein.
21. R. D. Chambers, D. Holling and G. Sandford, Intl. Pat Appl. PCT/GB03/01993, 1992.
22. R. D. Chambers and G. Sandford, Chim. Oggi., 2004, 22, 13.
23. R. D. Chambers, M. P. Greenhall and J. Hutchinson, Tetrahedron, 1996, 52, 1.
24. G. A. Olah, G. K. S. Prakash and K. Wade, Hypercarbon Chemistry, John Wiley & Sons, New York, 1987.

---

Footnotes

† For Part 15 see ref. 1.

‡ Electronic supplementary information (ESI) available: Cartesian coordinates and total energies of the MP2/6-31G* optimized geometries. See http://www.rsc.org/suppdata/lc/b4/b416400h/

---