Direct Fluorination In Mini And Microreactors

Transcript

Direct Fluorination In Mini And Microreactors INAUGURALDISSERTATION zur Erlangung des Doktorgrades der Fakultät für Chemie, Pharmazie und Geowissenschaften der Albert-Ludwigs-Universität Freiburg im Breisgau vorgelegt von Mathias Hill aus Bremervörde 2013 Die vorliegende Arbeit wurde von April 2009 bis März 2013 am Institut fur Anorganische und Analytische Chemie der Albert-Ludwigs Universität Freiburg unter der Anleitung von Prof. Dr. Ingo Krossing angefertigt. Vorsitzender des Promotionsausschusses: Referent: Prof. Dr. Thorsten Koslowski Prof. Dr. Ingo Krossing Korreferent: Prof. Dr. Peter Woias Datum der mündlichen Prüfung: 26.04.2013 I Für meine Oma Ingeborg Hill II Table Of Content 1. Introduction .......................................................................................................................... 1 2. Background ........................................................................................................................... 3 2.1 Fluorine ............................................................................................................................. 3 2.1.1 Physical properties .................................................................................................... 3 2.1.2 Chemical properties ................................................................................................... 3 2.1.3 Fluorine in nature ...................................................................................................... 4 2.1.4 Synthesis of fluorine ................................................................................................... 4 2.1.5 Application in chemistry ............................................................................................ 6 2.2 Fluorination ...................................................................................................................... 9 2.2.1 Mechanism of the fluorination ................................................................................. 10 2.2.2 Perfluorination ......................................................................................................... 13 2.3 Indirect Fluorinations ..................................................................................................... 16 2.3.1 Selective fluorinations .............................................................................................. 16 2.3.2 Perfluorination ......................................................................................................... 17 2.4 Microreactors .................................................................................................................. 18 2.4.1 Definition of microreactor: ...................................................................................... 18 2.4.2 Fluidic specialties: ................................................................................................... 18 2.4.3 Advantages of microreactors ................................................................................... 20 2.4.4 Disadvantages of microreactors .............................................................................. 23 2.4.5 Fabrication techniques: ........................................................................................... 23 2.4.6 Stabilities of materials against fluorine and hydrogen fluoride: ............................. 26 2.4.7 Microreactors in general chemistry......................................................................... 28 2.4.8 Direct fluorination in microreactors ....................................................................... 29 3. Equipment And Methods................................................................................................... 33 3.1 Fluorine Fume Hood ....................................................................................................... 33 III 3.1.1 Basic fume hood equipment: .................................................................................... 34 3.1.2 Fluorination equipment: .......................................................................................... 36 3.2 Microreactors .................................................................................................................. 40 3.2.1 Minireactor #1 ......................................................................................................... 40 3.2.2 Minireactor #2 ......................................................................................................... 44 3.2.3 Capillary reactor ..................................................................................................... 46 3.2.4 Characterisation of the used microreactors ............................................................ 47 3.2.5 Set up for silicon chip coating tests ......................................................................... 50 3.3 Fluorination Of Metal Salts ............................................................................................ 51 4 Results .................................................................................................................................. 52 4.1 Silicon Chip Coating Tests ............................................................................................. 52 4.1.1 Gold coating............................................................................................................. 53 4.1.2 Copper coating......................................................................................................... 53 4.1.3 Nickel coating .......................................................................................................... 53 4.2 Direct Fluorination Experiments .................................................................................... 55 4.2.1 Choice of substrates ................................................................................................. 55 4.2.2 Fluorination of acetonitrile...................................................................................... 61 4.2.3 Fluorination of toluene ............................................................................................ 62 4.2.4 Fluorination of methyl tert-butyl ether: ................................................................... 66 4.2.5 Fluorination of iso-propylacetate: ........................................................................... 66 4.2.6 Fluorination of n-butyl trifluoroacetate................................................................... 69 4.2.7 Fluorination of ethyl acetoacetate ........................................................................... 70 4.2.8 Fluorination of ethylene carbonate: ........................................................................ 71 4.2.9 Fluorination of propylene carbonate: ..................................................................... 81 4.3 Ethylene And Propylene Carbonate As Solvent For The Fluorination Of Ionic Substrates .............................................................................................................................. 89 IV 4.3.1 Comparison of EC and PC ...................................................................................... 89 4.3.2 Fluorination of closo-K2[B12H12] ............................................................................ 89 4.3.3 Fluorination of tetraalkylammonium salts............................................................... 91 5. Summary ............................................................................................................................. 93 6. Experimental Section ......................................................................................................... 98 6.1 Used Materials: ............................................................................................................... 98 6.2. Line Operations ............................................................................................................. 98 6.2.1 Leak testing .............................................................................................................. 98 6.2.2 Measuring lines volume ........................................................................................... 99 6.2.3 Purging of the line ................................................................................................... 99 6.2.4 Connecting of PFA tubes ......................................................................................... 99 6.2.5 Preparation of a hydrogen fluoride containing NMR sample ............................... 100 6.3 General Comments On Working With Elemental Fluorine ......................................... 101 6.3.1 Risks, toxicity and treatment .................................................................................. 101 6.3.2 Passivation of tubing and containments ................................................................ 103 6.3.3 Use of iodides for fluorine detection ...................................................................... 103 6.3.4 General safety advises ........................................................................................... 103 6.4 Chip Coating Tests ....................................................................................................... 104 6.5 NMR Conditions........................................................................................................... 104 6.6 Fluorinations ................................................................................................................. 104 6.6.1 Fluorination of toluene and TBME ........................................................................ 104 6.5.2 Fluorination of iso-propylacetate .......................................................................... 106 5.6.3 Fluorination of ethyl acetoacetate ......................................................................... 107 5.6.4 Fluorination of ethylene carbonate ....................................................................... 107 6.6.5 Fluorination von 4-methyl-1,3-dioxolan-2-on (PC) .............................................. 109 6.6.6 Fluorination of closo-K2[B12H12]: ......................................................................... 112 V 6.6.7 Fluorination of tetra-n-Butylammonium Tetrafluoroborate.................................. 113 7 Literature ........................................................................................................................... 114 8 Danksagung........................................................................................................................ 122 VI List Of Abbreviations '' Inch ρ Density [g cm-3] σ Surface tension [kg s-2] η Dynamic viscosity [kg m-1 s-1] ν Kinetic viscosity [m2s-1] A Surface [m2] Bo Bond number Ca Capillary number DAST Diethylaminosulfurtrifluoride EC 1,3-Dioxolan-2-on FEC 4-Fluoro-1,3-Dioxalan-2-on GC Gas chromatography HPLC High pressure liquid chromatography ID Inner Diameter iPrOAc iso-Propylacetate MS Mass spectrometry MTBE Methyl-tert-butyl ether NMR Nuclear Magnetic Resonance Spectroscopy OD Outer Diameter PC Proplyene carbonate PEEK Polyether ether ketone PSU Polysulfone PTFE Poly(difluoromethylene) Q Heat energy SEM Scanning electron microscope [J] VII Re Reynolds number TBAF Tetra butyl ammonium fluoride u velocity We Weber number [m s-1] VIII 1. Introduction The search of chemists to find more effective substances in life and material sciences leads often to fluorinated molecules.[1] Fluorine, with its very high electronegativity and very stable bonds, offers a multitude of possibilities for application in chemical industrial processes. For example, fluorination is employed in order to vary the lipophilicity in pharmaceutical products. Today, ten of the 30 top selling pharmaceutical products contain at least one fluorine atom.[2, 3] Other examples are the enhancement of the polarity of liquid crystals for LCDs,[4] the weakly coordinating anions [5] or the fine tuning of the electrochemical stability of solvents for lithium ion batteries like in (fluorinated) ethylene carbonate.[6] In order to perform fluorination of organic compounds, expensive reagents, such as DAST, Selectfluor® or the Ruppert-Prakash reagent are used.[7] Another commonly applied method is electro-fluorination, which works well for perfluorination but is rather limited in terms of selectivity and the tolerance of functional groups.[8-10] After the initial work by Lagow [11] and Rozen et al.,[12] direct fluorination using elemental fluorine, became a real alternative and several groups successfully demonstrated its application.[13, 14] Nevertheless there are two main difficulties that are commonly encountered when working with fluorine: i) The strongly exothermic character of the reactions: for example, CH4 + F2 → CH3F + HF (∆rH ≈ -430 kJ mol–1). A single substitution already provides enough energy to cleave a C-C bond in the molecule, which usually has a bond energy of 351–368 kJmol–1.[14] The fast rate of the direct fluorination reaction often leads to local hot spots or even explosions,[15] which is problematic with respect to selectivity and degradation reactions, not to mention operational safety. To overcome this, very fast energy transport and efficient temperature control is essential. ii) The second challenge is the toxicity of fluorine. A leak can lead to a release of highly toxic fluorine or hydrogen fluoride gases. Thus, a minimization of hazardous reagents would be desirable, given that the overall product yield is still acceptable. One elegant possibility to handle the problems of elemental fluorine is the use of microstructured reactors. These reactors, having reaction channels less than 1 mm in diameter, were used successfully for a variety of liquid organic chemical reactions since the late 1990s.[16] They are used as tool for process intensification through all fields of chemistry. Their main benefits are increased yields and selectivities, increased reaction rates, simplification of 1 process flows, accessibility of exothermic and thermal runaway reactions, increased safety and increased efficiency.[17] The direct fluorination reactions profits mainly from the high surface to volume ratio of microreactors, which provides very fast heat transfer. This is necessary to remove the high thermal energy released during the direct fluorination.[18] Additionally, the small channels guarantee fast mixing in two-phase systems.[19] The risks of working with fluorine are minimized due to the reduced volumes. Despite of these advantages, so far very few groups have developed micro-reactors for direct fluorinations. Chambers and Sandford et al. developed channel and falling film reactors and used them for various selective direct fluorinations. Their reactors are based on nickel or stainless steel, and are to date the only broadly tested micro-reactor systems for direct fluorination. They obtained very good results for the selective direct fluorination of organic compounds such as β-ketoesters,[20] aromatic systems [21] and ethers.[22] Jähnisch et al.[23] developed a falling film microreactor for direct fluorination and Jensen et al.[24] invented a silicon based microreactor. The channel widths were between 500-250 µm. Both groups tested their reactors with toluene in solvents like acetonitrile as their only substrate. In this thesis a new type of mini and microreactor developed in cooperation with the Woias group (IMTEK, Freiburg i. Br.) is presented. These reactors were used for the direct fluorinations of several organic solvents and ionic substances. An experimental focus was laid on the fluorination of the solvents ethylene carbonate and propylene carbonate and the direct fluorination of closo-K2[B12H12]. Parallel to the direct fluorination reactions a silicon based microreactor was in construction at the IMTEK. For this reactor fluorine resistance tests of different protective layers were investigated as part of this thesis. 2 2. Background In this chapter a short overview on fluorine and fluorination as well as the basic micro fluidics are given. Afterwards these two topics will be combined and what already has been done in direct fluorination reactions using microreactors by other research groups will be described. 2.1 Fluorine Fluorine is an element, which has many unique properties, compared to the other halogens and all other elements. 2.1.1 Physical properties Fluorine is a slightly yellow gas with a boiling point of –188.13 °C. The liquid is weakly yellow coloured and crystallizes at -219.62 °C to a colourless crystal. At 0°C fluorine has a density of 1.696 g L-1.[25] In nature fluorine is found 100 % as the 19 9F isotope, 18 9F can be produced in a cyclotron by conversion of H218O.[26] 2.1.2 Chemical properties Fluorine is the most reactive and most electronegative element in the periodic table. The Pauling electronegativity value of fluorine is 3.98. Its oxidation potential is in acidic solution ε0= 3.05 V, in alkaline solution ε0= 2.87 V. This reactivity allows fluorine to react readily with almost every element in the periodic table, except helium, neon and argon. Some metals form passivating metal fluoride layers, which slower the process of fluorination and some (like krypton and xenon) need energy by irradiation to react.[25] Table 1 Molecule distances (Dx-x) + dissociation energies (Ediss) of halogens (see also graph on the right) [25, 27] and [28] the temperature of 1 % dissociation + ionization energy F2 Cl2 Br2 I2 1.435 1.988 2.284 2.666 EDiss.[kJ mol ] 158 243 193 151 1% Diss. at T. [°C] 765 975 775 575 1681 1251 1139 1008 Dx-x [Å] -1 -1 EIonisation.[kJ mol ] -1 EDiss [kJ mol ] Figure 1 Dissociation energies of halogens 3 The dissociation energies of the halogens (Table 1, Figure 1) are constantly increasing from iodine to chlorine, but fluorine’s dissociation energy is with 159 kJ mol–1 extraordinary weak.[29] This is based on large lone pair orbitals compared to the atom size. Because of this, a repulsion of the lone pairs in the F2 molecule follows, this leads to a weakening of the bond. 2.1.3 Fluorine in nature Fluorine makes up, as fluorides, 0.065 percent of the earth’s crust, which makes it the second most occurring halogen after chlorine. Most fluorine is bound as fluorspar (fluorite, CaF2), but also cryolite (Na3[AlF6]) and apatites (e.g. Ca5[F/(PO4)3]). Fluorspar was the first fluorine mineral being described (1546 by Agricoloa)[30] One of the most prominent appearances of fluorine inside the human body is in teeth. In 1938 Armstrong et al. found in carious parts of human teeth only 62 % fluorine content relative to healthy teeth.[31] This was one of the first evidences of fluorine being crucial for the teeth health. Today the positive effect of fluorine on the teeth enamel is proven. The enamel is mainly based on hydroxyl apatite (Ca10(PO4)6(OH)2 or Ca5(PO4)3OH), but if fluoride is added fluoro apatite (Ca5(PO4)3F) can be formed. The fluoride ion can, because of its smaller ionic radius compared to hydroxyl groups, form a more dense crystal structure. This leads to an improved hardness and chemical resistance.[32] But not only teeth do need fluorine. It is classified as an essential element for the human body, though only small doses are necessary. For an adult the estimated safe and adequate daily amount is 1.5-4.0 mg.[33] In 2012 the group of Kraus published the first proven existence of elemental fluorine in nature. It was detected by 19 F-NMR in CaF2 crystals. They suspect the radioactivity of uranium traces to have released the fluorine.[34] 2.1.4 Synthesis of fluorine The use of fluorspar as a flux additive in melts was first mentioned in 1529 by Georgius Agricola, who also gave fluorine its name by naming the fluorspar (flux→fluss (ger.)→fluor, a latinization by Agricola). After 1771 the synthesis of hydrogen fluoride by the reaction of strong acids with fluorspar was discovered by Carl Wilhelm Scheele. In 1810 AntoineLaurent Lavoisier predicted the existence of fluorine. André-Marie Ampère then named the unknown element fluorine. 4 Electrochemical synthesis: Many chemists tried to synthesize elemental fluorine by electrolysis of hydrogen fluoride. But many were injured or even killed in this process. The main problems were the dangers of the highly toxic HF and the explosive recombination of fluorine and hydrogen.[30] In 1886, Henry Moissan first was able to produce elemental fluorine by electrolyzing HF (Equation 1).[25] By adding molten potassium fluoride into an anhydrous hydrogen fluoride bath, he was able to increase its conductivity. The Equation 2 shows the reaction of hydrogen fluoride and sodium or potassium fluoride. 542.6 kJ mol-1 + 2HF → F2 + H2 (1) HF + MF → M[HF2] (2) Additionally he used a platinum electrolyser. To lower corrosion, he reduced the temperature. With this set up he found a yellow green gas at the anode, the first elemental fluorine.[35] This technique, developed by Moissan, is up to now the fundament of elemental fluorine production. Today the medium temperature process is used, in which 1.0 mol KF combined with 1.8 to 2.5 mol HF are electrolyzed at 70-130 °C. This is a continuous process in which the lost hydrogen fluoride is constantly replaced. The electrodes are made of petrol coke and separated by an iron plate. The produced fluorine still contains up to 10 % hydrogen fluoride, which is removed by freezing out at –100 °C and by passing the gas through sodium fluoride.[25] A schematic description of an electrolysis cell for the fluorine production is Steel Cathode H2 Gas Separation Carbon Anode HF + KF Lid F2 HF shown in Figure 2. Figure 2 Set up of a hydrogen fluoride electrolysis cell for fluorine production. 5 [28] Chemical synthesis: In 1986, 100 years after Moissan’s invention, Karl Christe published a chemical route to synthesize elemental fluorine. It is based upon the thermodynamically unstable high oxidation state of transition metal fluorides. Those can be stabilized by their corresponding [MnF6]2anions. Additionally he used the fact that weaker Lewis acids, e.g. MnF4 can be displaced from its salts by a stronger Lewis acid. In his case SbF5 was used to react with K2[MnF6] to MnF4 which then decomposed into MnF3 and a half fluorine molecule. The first synthesis of fluorine on a chemical route in shown in Equation 3.[36] SbCl5 + 5HF → SbF5 + 5HCl K2[MnF6] + 2SbF5 → MnF3 + ½F2 (3) 2.1.5 Application in chemistry Fluorine itself is used only for the synthesis of fluorinated compounds. Important are fluorine containing molecules in the fields of fluorocarbons, nuclear power technology (e.g. UF6), fluorographite, perfluorinated polymers (e.g. Teflon®), surface fluorination of polymers (fuel tanks),[25] weakly coordinating anions,[37] pharmaceuticals and pesticides. Nuclear power technology: The synthesis of UF6 is by quantity, the main application for fluorine. It is used for the enrichment of 23592Uranium, by gas centrifugation. Uranium hexafluoride has a melting point of 64 °C (triple point) and a sublimation temperature of 56.5 °C.[25] For the centrifugation the UF6 mixture is transferred into the gas phase and then centrifugated with speeds up to 700 m/s. The mass difference of 23592UF6 and 23892UF6 is 0.85 % Perfluorinated polymers: Perfluorinated polymers like PTFE (Teflon®), PFA or FEP show a very good chemical stability, are non sticky and having a high melting point (Tm > 250 °C, PTFE≈ 327 °C). The structures of the three sample polymers are shown in Figure 3. Possible applications for example are water repellent membranes, or tubing and containers for highly corrosive chemicals. F F F F C C C C F F A n n F F F F F O B CF3 Figure 3 Chemical structures of PTFE (A), PFA (B), FEP (C). 6 m F F C C n F F C F F F CF3 m Fluorocarbons: They are used up to now as important refrigerants. Because of their large global warming potential (GWP) there has been a lot of effort to reduce it. The GWP describes the global warming potential of a substance relative to CO2, which is set to 1. For example the fluorocarbon CF3CFH2 (1,1,1,2-Tetrafluoroethane, R-134a) has a GWP of 1430, which is too high for today’s EU norms. A possible alternative is the CF3CF=CH2 (HFO-1234yr), which shows a GWP of only four.[38] Fluorographite: When well crystallized graphite is fluorinated at elevated temperatures fluorine is added to the carbon-carbon double bonds. This can lead to molecules with the composition from CF4 over (CF)x to (C4F)x. (CF)x is a non conducting and chemically inert white solid, which is even at space conditions a very good dry lubricant. The properties of (CF)x are very unlike to those of (C4F)x, it is a blackish solid, which is still conducting and used as electrodes in high energy density batteries.[25] Hydrogen fluoride: Hydrogen fluoride is a colourless gas/ fuming liquid, with a boiling point of 19.51 °C. Its melting point is -83.36 °C. It can be produced by the reaction of fluorspar with sulphuric acid (Equation 4). ∆rH = +59 kJ mol-1 CaF2 + H2SO4 → CaSO4 + 2HF (4) Below 90 °C gaseous HF partly exists in a hexameric form, (HF)6, additionally to the monomeric form. Liquid contains hydrogen fluoride unbranched chains of on average seven molecules. The chains are build on very strong hydrogen bridges HF can form due to its high dielectric constant of 83.5 at 0 °C (water 78.3, 25 °C). In the solid state zick zack formed chains are formed. Hydrofluoric acid is a weak aqueous acid (pKS= 3.2), which is based on the formation of a H2O-Hδ+···Fδ- ion pair, which is less acidic as solvated hydronium ions. A special property of hydrogen fluoride and hydrofluoric acid is its ability to react with glass to gaseous silicon tetra fluoride (Equation 5).[25] SiO2 + 4 HF → SiF4 ↑ + 2 H2O (5) 7 Surface fluorination of polymers: For already 30 years the fluorination of polymer components are widely carried out in industry to improve gas barrier properties. The main application is the fluorination of polymer fuel tanks to reduce evaporation of the fuel through the polymer wall. The polymers only are fluorinated at the surface and very thin fluorine mixtures of approx. 2-3 % are used. Normally fluorine atoms are introduced into an only a few nano meter to a few micro meter thick layer. The fluorination duration needed to form a layer is strongly depending on the type of the polymer. Low density polyethylene, a pure aliphat, needs around 30 times more time to form a certain fluorination degree as poly styrene, which contains phenyl rings. Additionally to the gas barrier properties, the adhesion properties and the colourability can be improved by a fluorination together with oxygen. This mixture can form polar -FC(=O) groups, which react in contact to water to -COOH groups. Fluorination also can lead to improved chemical resistance of the polymers.[39] Weakly coordinating anions: They show only weak interactions towards their cation. These anions usually have a low charge and are highly symmetrical. To achieve this, the anions are often enclosed by fluorine atoms, whether directly bound to fluorine atoms or by highly fluorinated ligands. This reduces the interactions in two ways. First the negative charge of the anion is σ-delocalized by the electronegative fluorine atoms. Second, the fluorine’s lone pairs repell all other atoms approaching the anion. This can be used in fundamental research as well as industrial applications. As first example, in the Krossing group anions of the type [Al(ORF)4]– were used to stabilize highly reactive cations like the [P9]+ cation[40], or [Ga(PPh3)3]+.[5] A second more applied example is the widely used [PF6]- anion. Its main application is to help dissolve the lithium cation into a polar organic solvent, for example for applications in batteries.[37, 41] Pharmaceuticals and pesticides: Fluorine has many properties, which can help to make pharmaceuticals and agrochemicals more efficient. 1. Fluorine has a similar Van-der-Waals radius (1.35 Å) as hydrogen (1.2 Å) or hydroxy groups (1.4) and can be used as a mimic. It can also mimic metabolic label groups like nitro groups. 2. Taking advantage of the inductive effect of fluorine, pKa values can be modified. 3. The stable carbon fluorine bond can help to increase metabolic stability. 4. The lipophilicity can be increased. 5. Fluorine can be used to form hydrogen bridges, to stabilize a 8 conformation. An example of conformation control by fluorine substituents is shown in Figure 4.[1] H F NH2 O B HO NH2 HO O A H HO HO F Figure 4 Noradrenalin-analogues showing the effect of fluorine substituents on hydrogen bridges. A) β-analogue, B) α- analogue [1] In 2012 seven (28 %) of the top 25 retail selling drugs in the USA contained at least one fluorine atom.[42] In Figure 5 three important fluorine containing drugs are shown.[7] H N O O O F A OH F3C N N B HN O OH N H OH COO- C Ca2+ F Figure 5 Structures of A: Prozac (antidepressant), B: Ciprobay (antibiotic), C: Lipitor (cholesterol lowering). [7] In 2006 Theodoridis showed the development of fluorinated agrochemicals from by comparing the fluorinated and non fluorinated in the Pesticide Manual. In 1977 4 % of the agrochemicals were containing fluorine atoms. In 2003 this number increased to 14 %. Sorted by types were 45 % herbicides, 33 % insecticides and 18 % fungicides. Most of this substances contained fluorinated or trifluoromethylated aromatics.[43] 2.2 Fluorination First fluorinations with elementary fluorine were already carried out by Henri Moissan himself. On the 26 of June 1886 he discovered an electrolysis method in which he used aqueous hydrogen fluoride and traces of potassium fluoride to produce gaseous fluorine. Soon he first described the immense reactivity of fluorine. In his experiments most substances brought in contact with fluorine were ignited or decomposed.[44] 9 Its reactivity is mainly based on the weakness of the fluorine’s F-F bond. In Table 2 the dissociation energies and the reaction enthalpy of the direct fluorination of hydrocarbons and boranes are shown. In contrast to the weak fluorine bond, the fluorine hydrogen bond, a molecule which is often formed in fluorinations, possesses a dissociation enthalpy of 569 kJ mol–1.[45] In case of a direct fluorination of a hydrocarbon additionally a carbon hydrogen bond must be cleaved (366 - 431 kJ mol–1) and a strong fluorine carbon bond gets formed (–452 - –485 kJ mol–1, for methane[46]). After all this would result in a reaction enthalpy for a direct fluorination of around –394 kJ mol–1. This is approx. the range of the combustion energy of methane using 1 mol oxygen (445 kJ mol–1 [47]). Table 2 Dissociation energies of F2, CH, HF, CF, BH, BF. [46] Energies of the direct fluorination of hydrocarbons and boranes [kJ mol-1] F-F C-H H-F C-F Ediss. 158 366-431 569 452-485 F-F+C-H/F-H+F-C 524-589 ∆E (F2+C-H→C-F+C-H) 1021-1054 -432-530 [kJ mol-1] F-F B-H F-H B-F Ediss 158 327 569 759 F-F+B-H/F-H+F-B 485 ∆E (F2+B-H→B-F+B-H) 1328 -843 For a long period of time it was not believed it is possible to fluorinate hydrocarbon with fluorine directly in acceptable yield.[48] The main argument was the carbon-carbon bond with its dissociation energies between approx. 250 and 350 kJ mol–1.[49] A single fluorination would be sufficient to cause decomposition. In 1976 the groups of R. H. Hesse and S. Rozen and showed that if low temperatures and a polar uncreative solvent, a selective fluorination is possible. They also first interpreted the fluorination as an electrophilic reaction.[50] 2.2.1 Mechanism of the fluorination In 1986 Lagow und Margrave discussed the mechanism of the fluorination more detailed. They stated there are two different initiation steps for radical mechanism of the fluorination. The first one is a homolytic cleavage of the fluorine-fluorine bond, with a following attack of a fluorine radical on a hydrogen atom. The cleavage of a fluorine molecule needs 10 157 kJ mol-1. This means at a temperature of 298 K the amount homolytically cleaved fluorine is less than 1 %. Experiments at –78 °C carried out in the dark still showed fast fluorinations. Because of this, Miller et al.[51] postulated a second start of the radical chain. In this scenario a fluorine molecule attacks the hydrogen atom and a hydrogen fluoride and a methyl as well as a fluorine radical is formed. Rozen et. al. reported in 1988 a possible fluorination of protected 4-methylcyclohexanol in a 1:1 mixture of monochlorotrifluoromethane and chloroform. The tertiary carbon atom was fluorinated with a 60 % yield. With non polar solvents only tars were obtained.[12] The relative selectivities fluorinating primary secondary and tertiary carbon atoms were described by J. M. Tedder et al.. [52] and are shown in table 1.2.. Compared to the selectivity of chlorine, fluorine shows, due to its reactivity, a lowered selectivity. Rozen further found that in his system stereo information is preserved during fluorination. This indicates an electrophilic mechanism of the fluorination. The chloroform acts as an acceptor for the developing F through hydrogen bonding.[12] A scheme of this reaction is shown in Figure 6. OCOR R H F2 OCOR R CHCl3 H F F HCl3C OCOR R -HF F Figure 6 First ever published example of direct fluorination keeping stereo information untouched. [12] For more detailed information about the possible mechanism of the electrophilic fluorination in 2002 H. Fukaya und K. Morokuma calculated with ab-initio methods the fluorination of CH4+ F2 [+ CHCl3] and (CH3)3CH+ F2 [+ CHCl3]n n=1,3. In this work they also investigated how a polar solvent can influence the reaction. For their calculation they used as methods MP2/6-31+G*//MP2/6-31+(F)G* and ONIOM(MP2/6-31+G*//HF/6-31+G*). By their calculations they conclude the electrophilic pathway to be a two step process. First a hydride abstraction leads to a R+δ•••HF•••Fδ like transition state. Second a complicated rearrangement takes place, which leads to the final product. A scheme showing the rough structures of the different steps for the reaction with one chloroform molecule in the reaction is shown in Figure 7. For the fluorination of methane they found an activation energy of 142 kJ mol–1 (Table 3) without added chloroform and 96 kJ mol–1 for the system including three chloroform molecules. They state both activation energies to be too high for the reaction to take place under mild conditions. When they studied the system (CH3)3CH+ F2 they found calculated activation energies of 105 kJ mol–1 without and 63 kJ mol–1 with an chloroform molecule. 11 Both activation energies are still too high. But when they applied chloroform molecules to the reaction they found the ionic process to be clearly favoured to the radical one. The transition state 2 (Figure 7) is the alternative route to the formation of the radical pair t-Bu• + HF + F•. The relative energy of the formation of this radical pair is –13 kJ mol–1. When going from two to three chloroforms the relative energy of the second transition state is lowered from 13 kJ mol-1, which is just equal the radical formation energy, to –46 kJ mol-1. During this reaction the calculations showed a full retention of the t-butanes configuration.[53] pre-reaction complex H transition state 1 F F intermediate F F H H Cl product complex Cl Cl Cl Cl transition state 2 F F F H2 C H Cl Cl H F H H Cl Cl F F H2 C H Cl Cl H H2C H H H H Cl Cl Cl Cl Figure 7 Schematic sequence of the direct fluorination of t-butane based on the calculation of H. Fukaya und K. Morokuma. [53] For the fluorination of methane they found an activation energy of 142 kJ mol–1 (Table 3) without added chloroform and 96 kJ mol–1 for the system including three chloroform molecules. They state both activation energies to be too high for the reaction to take place under mild conditions. When they studied the system (CH3)3CH+ F2 they found calculated activation energies of 105 kJ mol–1 without and 63 kJ mol–1 with an chloroform molecule. Both activation energies are still too high. But when they applied chloroform molecules to the reaction they found the ionic process to be clearly favoured to the radical one. The transition state 2 (Figure 7) is the alternative route to the formation of the radical pair t-Bu• + HF + F•. The relative energy of the formation of this radical pair is –13 kJ mol–1. When going from two to three chloroforms the relative energy of the second transition state is lowered from 13 kJ mol-1, which is just equal the radical formation energy, to –46 kJ mol-1. During this reaction the calculations showed a full retention of the t-butanes configuration.[53] 12 Table 3 calculated highest activation energies in the direct fluorination Reaction Highest activation [53] Reaction –1 energy [kJ mol ] Highest activation energy [kJ mol–1] MeH + F2 142 t-BuH + F2 105 MeH + F2 + CHCl3 121 t-BuH + F2 + CHCl3 63 MeH + F2 + 2 CHCl3 MeH + F2 + 3 CHCl3 t-BuH + F2 + 2 113 CHCl3 t-BuH + F2 + 3 96 CHCl3 38 21 2.2.2 Perfluorination A perfluorination is a fluorination reaction in which all hydrogen atoms of a molecule are exchanged by fluorine atoms. Since the 1970th methods were developed to perfluorinate larger molecules. The main difficulties are the common direct fluorination problems, the occurring reaction heat and the acidic conditions due to the formation of hydrogen fluoride. Additionally for the perfluorination steric and electronic hindrance can appear, which might prevent a full perfluorination.[14] This is based on two effects. First the free electron pairs of the fluorine do shield the reactive site against further fluorination. Secondly does a fluorine atom lower the electron density at the surrounding hydrogen atoms. A series of reactivity can be formulated: CH3 > CH2F > CHF2.[54, 55] Up to now several methods were developed for perfluorination reactions. Examples are the La-Mar-process (1970) by Lagow and Margrave, the direct aerosol fluorination by Adcock (1987), the photo fluorination in liquid phase (1990) and the Exfluor-Lagow method (1992).[9] For the perfluorination of ethers Okazoe et al. published the PERFECT- process in 2001.[13] La Mar- Process: R. J. Lagow and J. L. Margrave published in 1970 a mild method for the perfluorination of hydrocarbons. They decided to minimize the concentration of fluorine to minimize the amount of radicals. And they considered the higher stability of poly fluorinated materials. They invented a procedure, which started at low temperature and with highly diluted fluorine and increased both over a long time. The low fluorine concentration was achieved by dilution with inert gas. The La-Mar process is necessarily a time consuming batch process and by this limited in industrial applications.[4, 14] 13 4- Zone Cryogenic- Reactor: In 1972 N. J. Maraschin and R. J. Lagow published the use of a 4-zone-cryogenic-reactor. A substrate is evaporated and condensed at the first two of 4 cold traps (-78 °C). Now diluted Fluorine was led over the condensed material. After this the first trap was brought to room temperature and the third one was cooled. This was stepwise continued until the last trap was warmed to room temperature. Behind the reactor a -78 °C cooling trap equipped with sodium fluoride trapped all products. They successfully fluorinated tetramethylethane with a yield of 9.3 %.[56] Two years later J. L. Adcock and R. J. Lagow used this reactor to fluorinate 1,4dioxane. Tests to achieve perfluorination via electro fluorination only yielded in the non cyclic perfluoro-1,2-dimethoxyethan. With the cryogenic reactor they achieved 38.5 % of perfluoro-1,4-dioxane.[11] Direct aerosol fluorination: 1980 J. L. Adcock et al. developed a method in which adsorbed organic material was directly fluorinated. At 850 °C sodium fluoride was brought as 17 Å particles into a carrier gas stream. This gas was cooled down to –198 °C and a second gas stream added, which carried the evaporated organic substrate. The substrate condensed onto the particles and was adsorbed. The loaded gas stream was now led through areas of increasing fluorine concentrations and rising temperatures. To finalize the perfluorination a photochemical step at the end was added. This method had several advantages. By the immobilization of the substrate onto a solid particle, lattice relaxation was possible. Additionally recombination of decomposed material was possible. Both decreased the amount of decomposed material. Another advantage was that produced HF was directly converted into NaHF2. In contrast to most direct fluorination methods, the direct aerosol fluorination is a continuous procedure.[54] Exfluor-Lagow: The main characteristic feature of the Exfluor-Lagow method is the perhalogenated inert medium in which gaseous fluorine is used for the fluorination. As solvent for example chlorofluoroethers (e.g. (CF2Cl)2CFOCF2OCF(CF2Cl)2) are used. Temperature between -40 and +150 °C can be used. An increase of fluorine at the exit indicates the almost completion of the reaction. Normally an excess of 10-40 % of fluorine is needed.[57] 14 PERFECT: In 2001 the group of Okazoe developed a method to synthesize perfluorinated ketones.[13] They called it PERFECT cycle. PERFECT stands for: PERFluorination of an Esterified Compound then Thermal elimination. A perfluoroacyl fluoride (e.g. FCO- CF(CF3)O(CF2)2CF3) was first converted into an ester with a non fluorinated alcohol. After fluorination with diluted elemental fluorine in an inert solvent (e.g. 1,1,2-trichloro-1,2,2trifluoroethane (R113)) the now perfluorinated former alcohol moiety was eliminated as a perfluoroacyl fluoride. The In the same step the original perfluorocayl fluoride is recovered. A scheme of the reaction cycle is shown in Figure 8. By using long fluorinated esters the volatility can be reduced, which prevents dangerous gas phase reactions. The reactions can be carried out at room temperature and easily be upscaled.[58] H O H R1H R2F O n F2 HF H R1 H H esterification perfluorination OH thermal elimination O F n HF R2F Figure 8 PERFECT cycle by T. Okazoe et al.. R1F O R1F F O F O R2F F [58] Microreactor: Chambers et. al. published in 1999 a method for the perfluorination of cyclic polyfluorinated substrates (Figure 9). A microreactor was used which is described in section 2.4.8. They used a liquid flow of 0.5 ml h–1 and a gas flow of 15.0 ml min–1 with a 50 % fluorine nitrogen mixture. The fluorination was done twice. The first time at room temperature, the second time at 280 °C [59] This resulted in a estimated fluorine excess of 370 % (Assumed density of the liquid of 1.0 g cm3). The obtained yield was 91 %, with a recovery rate of 52 %.[59] 15 F 3C F2 C C HF O CH CH H2C CH 2 F2 C C HF CF 3 -1 1. 0.5 ml h , 50 % F 2 in N2 with 15 ml min-1 ,room temp 2. 0.5 ml h-1, 50 % F 2 in N2 with 15 ml min-1 , 280°C F 3C F2 C C F2 O CF CF F 2C CF 2 F2 C C F2 CF 3 Figure 9 Reaction condition for perfluorination in a microreactor by Chambers et. al.. 2.3 Indirect Fluorinations To reduce the high energies of the direct fluorination often fluorinating agents are used. This reagents often react more selective and in a milder manner, but are more expensive. 2.3.1 Selective fluorinations Nuclephilic fluorination: There are several reagents to achieve a nucleophilic fluorination (“F-“). Most important for a good nucleophilic fluorination reagent is the solubility in organic solvents. When being coordinated to a polar solvent fluorides lose nucleophilicity. The solubility of alkali fluorides for example can be improved by using crown ethers. Alternatively voluminous cations like tetra butyl ammonium can be used to lower the ionic bond strength. A non ionic based example is (Diethylamino)sulfur trifluoride (DAST). It is used to convert hydroxyl groups into fluorides. The mechanism is based on the attack of the oxygen on the sulphur, in which a fluoride is released to attack the carbon.[60] Such fluorination agents are broadly commercially available, but often expensive. The costs for 25 g DAST are 181.50 € (Chem Pur, 2012, 1169 € mol-1) Electrophilic fluorination: There are several reagents to achieve an electrophile fluorination (“F+”). One large group are the N-F and O-F type reagents. Examples for those are CsSO4F, CF3OF, CH3COOF, Selectfluor®, Synfluor® and NFTh®. Some examples are shown in Figure 10. They do react at positions with acidic hydrogen atoms and with activated aromatic systems.[61] The costs for 25 g Selectfluor are 107.50 € (Sigma Aldrich, 2012, 770 € mol-1) 16 A B C NaF nBu F N S F F nBu CH2Cl O O S NF S O O N N F D nBu FnBu N O HO S O F O F E Figure 10 A:DAST, B: sodium fluoride, C: TBAF, D: Selectfluor, E: NFTh, F: CsSO4F. [1, 60] 2.3.2 Perfluorination For the perfluorination of organic molecules, there are two systems used in industry for the non direct perfluorination fluorination: the fluorination with CoF3 and the electro fluorination. CoF3: The cobalt trifluoride process was developed during the Manhattan project. It separates the fluorination of hydrocarbons into two steps. First, cobalt difluoride is fluorinated to cobalt trifluoride at 305 °C with elemental fluorine, with a reaction enthalpy of –222 kJ mol–1. The second step is the gas phase fluorination of the hydrocarbon releasing approx the same amount of energy. One problem of this process is the occurrence of rearrangements, which can occur when abstracting of a hydrogen atom forms to form a carbocation.[1] A drawback of this method is the low content of active fluorine. In theory 16.38 % of CoF3 are active, but only 2.5± 3.2 % of the CoF3 can be found as active species. This low content is mainly caused by the formation of dense CoF2. A possibility to minimize this effect is using a fluidized bed in which the denser CoF2 sinks to the bottom and can be refluorinated. The yields of this process are between 20- 40 % [62] and the substrates need to be volatile. Electro fluorination: In this process a dissolved substrate is fluorinated by a fluoride containing precursor by applying a voltage to the bath. There are three main procedures for electro fluorination. First the Simons Process in which nickel electrodes and anhydrous HF is used. This method is well established and used in industry.[63] The mechanism is expected to be based on higher fluorinated nickel at the anode.[1] Secondly, KF·HF with carbon electrodes can be used. This system is well situated for smaller molecules. 17 The third system is milder and also used for selective fluorinations. Triethylamine-HF or pyridine-HF is dissolved in an organic solvent like ethanol or acetonitrile.[63] The drawbacks, especially for the Simons process are the yields and low reproducibility. The yields are usually around 7 to maximal 70 %, which also leads to a large amount of waste. The unstable results are due to the corrosion of the electrodes which first, during an initiation phase, leads to an increasing yield, but after certain time the yield decreases. This is also accompanied by the formation of tarry side products over the time.[62] A reason for the low yield can be the isomerisation of the carbon backbone. This isomerisation can be either a cyclisation or branching.[64] 2.4 Microreactors 2.4.1 Definition of microreactor: Microreactors are defined as a reactor with three dimensional structures of an inner diameter of one millimetre and smaller in the relevant direction. A complete and fully accepted definition does not exist up to now. The size rule of less than one millimetre is nevertheless a good choice to point at the most important feature. In Figure 11 the scales of microreactors compared to chemical compounds and the diffusion length in one second is shown. micro reactors glas flask Diffiusion length in one second animals liquids gases small molecules proteins nm 10-9m cells µm 10-6m nanomicrotechnology technology Figure 11 large scale comparison m mm 10-3m macroworld [65] 2.4.2 Fluidic specialties: The small channel volumes and diameter lead to several effects, which are not observable in the macroscopic scale. 18 Laminarity: When going from macro to micro the flows are getting more laminar. Laminar flow has no convection vertical to the flow direction. In flow direction there is a parabolic flow profile, which is shown in Figure 12. Instead of convection diffusion is often the only way for vertical mass transfer. The contrast of a laminar flow is the turbulent flow. In which chaotic flows are dominant. Because there is a vertical mixing the flow profile is much more flat than parabolic. Figure 12 Scheme of a laminar flow profile in a straight tube. The laminarity can be described by the Reynolds number, which describes the ratio of inertial forces to viscous forces (Equation 5). Increasing velocity (u), channel diameter (d) and density (ρ) increase the turbulences, increasing viscosity (η) reduces it. For smooth linear round tubes the border of laminar to turbulent flow is approx. at Re= 2300. (5) Because the Reynolds number is directly proportional to the diameter of the channel and also the liquid flow, both are usually relatively low in micro fluidics, the flows in microreactors can be extremely laminar. Surface tension: The surface tension in micro channels can get dominant over other forces like gravity, inertia or friction. Connected to it is the interaction of a liquid with the wall. The main value to describe this interaction is the contact angle Θ. This is the angle between liquid and surface at the contact of a drop on a certain material. This can reach from cosΘ= +1, a complete wetting to cosΘ= –1, a ball with a contact surface → 0. The capillary pressure p is directly connected to the surface tension (σ) and the contact angle Θ and given by the Young-Laplace equation (Equation 6). (6) If cosΘ= +1 the pressure is positive, when cosΘ= -1 the pressure is negative. A schematic picture of this effect is shown in Figure 13. This pressure can on the one hand be used to 19 create a driving force or on the other hand to construct a stop position for the liquid. For a stop position a part of the channel is (for water) coated with a hydrophobic layer. This causes a pressure against the flow direction, which holds back the liquid until the pushing pressure exceeds the capillary pressure.[66] hydrophilic wall hydrophobic wall Θ Θ Figure 13 Interfacial interaction of a liquid with the wall. 2.4.3 Advantages of microreactors Reactions in micro structured reactors benefit of the small sizes. Heat transfer: In microreactors the heat transfer is superior to classic approaches because typical values for the specific surface of microreactors are between 10 000 m2m–3 and 50 000 m2m–3. A traditional reactor has a specific surface around 100 m2m–3.[67] A comparison of different reactors and their volume to surface ratios are shown in Figure 14. Volume to surface ratio [m²m -3] 15000 13333 10000 4000 5000 2 46 100 100 L pilot plant reactor (80*200 cm, cylinder) 1000 ml round bottomed f lask 100 ml round bottomed f lask 0 1.0x1.0 mm mini channel 0.3x0.3 mm micro channel Figure 14 comparison of the volume to surface ratios of different reaction containments. The heat transfer coefficient is inversely proportional to the specific area, values of 10 kW m-2 K–1 are possible. In the Equation 7 h is the heat transfer coefficient, Q the transferred heat, A the surface and ∆T the temperature difference of wall and liquid. (7) ∆ 20 These high values for the heat transfer allows a very efficient cooling or heating of a reaction. This property is crucial in highly exothermic reactions which otherwise can form many side products or lead to a thermal runaway of the reaction. The fast heating and cooling rates also allow to have very a defined times of heating or cooling. An example for an innovative cooling system was published 1999 by Löwe et al.. The problem they faced was a gas which reacted at 1200 °C and needed to be cooled down rapidly after the reaction to avoid side reactions. They constructed a cooling device, which was able to cool the gas down to 120 °C within 0.2 ms. This corresponds to a theoretical cooling rate of 10,000,000 K s–1.[68] Mass transfer: A further argument using a microreactor can be its ability to mix two substrates in times orders of magnitude shorter than in common reactors. Using micro mixers, the mixing time can be reduced from seconds to milliseconds. Mass transfer is especially relevant for two types of reactions. First, fast reaction in which the mass transfer is dominating, like the catalysed hydration of cyclohexene.[69] Second, it is important in two phase systems. For those very large interfacial surfaces can be realised. In multiphase system there are many possibilities to vary the flow profiles and with it the interfacial surfaces. A series of possible flows within a tube are shown in Figure 15. A B C D E Figure 15: Left: Flow regimes in gas liquid two phase systems. A: bubbly flow, B: slug flow, C: wavy annular flow, D: annular flow, E: churn flow [70] Right: Picture of a slug flow in a 1/8’’ inch ID PFA tube exiting the minireactor #2. The slug flow, for example adds an additional current to the main convection, which increases radial heat and mass transfer (Figure 16). The rate of circulation is highly depending on the slug length and its velocity. In general it is valid that the smaller the slugs are and the faster they are moving the better the heat and mass transport is inside the slug.[71, 72] The size of the bubble is on the one hand influenced by the gas and liquid flows and on the other hand by the gas inlet geometry. The geometry has large influence of the shearing off of the bubbles which has influence on bubble size and frequency.[73] Outside of the bubbles a small liquid film remains. This leads to a larger contact area compared to the surface of the slug caps only. The 21 liquid slugs have a slower velocity than the gas flow.[74] An intensive study on the gas bubble formation using different nozzle geometries were carried out by P. Lang at the IMTEK in Freiburg as part of the Direct Fluorination Project.[75] An example for a very high interfacial area was realized by Jähnisch et al.. They constructed a falling film reactor which forms superficial gas liquid surfaces of up to 40,000 m2m–3. This is based on liquid films thicknesses of down to 25 µm.[23] F2/N2 slug Liquid F2/N2 slug Figure 16 Scheme showing the flows within a gas liquid slug flow system. [71] Number up: To increase the amount of product in a certain time a scale up in the classical sense is not possible in micro reaction technology. Typically a reaction is upscaled, by increasing the volumes. The amount of product starts at a few hundred gram scale in laboratories over a pilot plant to a multi ton per year industry plant. For microreactors an increase of volume is not possible, instead the reactors are numbered up. This means a parallel use of multiple reactors. The construction of these systems can be divided in internal and external number up. In the external way, the whole set up is multiplied in total, which is easier, but not as parallel as the internal number up, in which several of the key reactors or mixers are combined to a new piece, which has only one supply and collection container. The external number up is advantageous for slow or hazardous reactions, for multi phase problems, when an after reaction processing is necessary or if only a small increase of productivity is needed (≈ <10x).[76] Safety: Finally there is a safety aspect in using microreactors. Their use is advantageous when high pressure or explosive or toxic substances are involved in the reaction. On the one hand the amount of dangerous material handled within one moment is reduced to a minimum, making damages on the reactor system less dangerous. On the other hand improved heat transfer rates help to keep even very exothermic reactions within safe temperature parameters. A runaway of the reaction is less likely.[17] Additionally the micro channels are much more stable at high 22 pressures than larger reactors, which reduces accidents caused by overpressure or material failures.[67] 2.4.4 Disadvantages of microreactors Reactions need to fulfil some criteria to be suitable to be carried out in micro devices. They must not contain solids. Though nano particles can be prepared in microreactors [77] any macroscopic solid would lead to a clogging of the reactor channels. The reaction needs to be fast enough to be finished after passing through the micro system. Reactions which need longer than 30 minutes for completion are not suitable. The realisation for production is much more expensive than common techniques. This needs a strong process intensification for expensive products to financially be profitable.[17] Additionally to the limited reaction range the corrosion of the reactors materials is an important factor. Micro scaled walls and segments are much thinner than in macroscopic devices. Because of this, corrosion can have a greater effect, which only can be countered by using more expensive material. 2.4.5 Fabrication techniques: There are various techniques to produce micro structured devices. Photolithography, LIGA, injection moulding, hot embossing, powder blasting, laser micro forming are presented. Photolithography: In the photolithography a thin layer of a photosensitive material is coated onto the reactor material, which often consists of glass or silicon. This is typically done by spin coating. The materials of this layer are often polymers, which do react when being irradiated. For the irradiation of the photosensitive layer a mask decides where the polymer is activated. There are positive and negative photoresists, which decides whether the activated or the non activated parts of the layer are removed.[78] Examples for negative photoresists are poly(vinyl cinnamate)s , polystilbenes and allylic esters. The poly(vinyl cinnamate) for example forms dimers when exposed to UV light (Figure 17). 23 Polymer Polymer O O O C C O C H C H C H C H O C H C H C O C C H C H hν O O Polymer Polymer Figure 17 A possible photoresist. A cinnamic ester linked to a polymer can dimerise when irradiated. [79] Positive photoresist are more common for the lithography. In this process a hardened polymer layer can be made soluble again by integrated photoactive groups, which cause decomposition of the polymer backbone.[79] After the photoresist step, a mask is placed on the reactor material, and an etching step follows. Some suitable etching reagents are applied to the surface. They can only react where the photoresist has been removed. For example for glass, a mixture of 1% HF and 5% NH4F in water at 65 °C results in an etch rate of 0.3 to 0.5 µm min-1. A scheme of this process is shown in Figure 18.[78] For silicon the use of KOH solution as etching reagents leads to anisotropic etching. The 100 planes of the silicon are etched 200 times faster than the 111 plane. This leads to an angle of 54.7° in which structures like Vshaped channels can be etched into the silicon.[80] The selectivity is based on the different binding situations inside the silicon planes.[81] Photoresist Chrome Mask Photoresist Chrome Photoresist Chrome Glass Glass Glass Chrome masked glass coated Mask attached Photoresist exposed to UV light Irradiation with photosresist through a mask Glass Photoresist and metal removed Removing Etching of developed chrome Photoresist was Photoresist Chrome Photoresist Chrome Glass Glass Etching of exposed glass Glass Two glass plates bonded to form a channel Bonding Glass Figure 18 Sequence of process of a photolithographic process. Etching of glas Etching of exposed chrome [78] LIGA: Lithographie, Galvanioformung and Abformung (lithography, electroforming and moulding) is a technique to produce a form for micro polymer structured pieces. A photoresist layer, placed on a conducting plate is irradiated by a strong light through a mask. After this the mask 24 and the activated parts are removed. The parts with no resist can be filled with a metal by electroforming. When the base plate and the remaining photoresist are removed, a metal mould remains. This can be used to produce micro formed polymer pieces. By this process structures of lateral geometry with dimensions of 1 mm to 0.2 µm can be produced. A roughness Rα= 30 nm can be achieved.[82] Injection moulding: This is the most common method to form plastic components in macro scales. Hot polymer is pressed through a hole into a mould. Micro injection moulding is the similar to the common process, but products of less than one gram can be produced. In principle this can directly be adapted to micro scales, but there are several mainly technical difficulties. For example the fast cooling of the polymer melt in small channels and the fragility of the objects produced.[83] Hot embossing: This process is similar to the injection moulding. Between two moulds a thermoplastic foil is placed. The whole set up gets heated under vacuum and the moulds are pressed together until a defined pressure is reached. After cooling the product can be released. A scheme of this process is shown in Figure 19. The vacuum is needed to fully fill the cavities. The main advantage of these technologies, comparing to the injection moulding is the production of pieces with high aspect ratios. This needs high pressure to fill in a highly viscous polymer into the moulds. Additionally the shorter ways and slower movement of the polymer reduce sheer stress, which can for example be necessary for optical properties of the polymer.[84, 85] Pressure + Heat Cavities Upper mold Imprinted Parts Thermoplastic foil Lower mold Figure 19 The hot embossing process. [84] Powder blasting: A photoresist mask (see Photolithography) is placed on the reactor material (often glass). Then a fine powder of e.g. aluminium oxide is accelerated in a gas stream towards the surface. The channels were created by erosion. Channels with a width of below 100 µm can be manufactured this way. The main advantage of this method is its cost efficiency and the 25 possibility to work outside a clean room. The main disadvantage is the relatively high roughness of the walls.[86] Laser micro forming: It is possible to form channels and microstructures by laser ablation. This method is preferred when handling harder materials like metals and ceramics, but also polymers like PEEK and PSU can be processed. With a excimer laser a roughness around Rα= 100-200 can be achieved for PEEK and PSU. One large advantage is the possibility to directly write on a substrate without a previous mask need to be placed onto it.[87] 2.4.6 Stabilities of materials against fluorine and hydrogen fluoride: The common materials to handle fluorine or hydrogen fluoride are copper, nickel, Alloy 400 (Monel), perfluoro polymers and stainless steel. Nickel is treated more detailed in this section, because of it important use in this thesis. Copper: Is the cheapest of the metals and below 320 K stable against 1000 hPa fluorine. A corrosion of 23 µm a–1 can be observed at 303 K. This increases rapidly at elevated temperatures. At a temperature of 373 K the loss is 1.1 mm a–1. Presence of water leads to a reduced fluorine inertness. The passivation layer is hydrolysed from copperdifluoride to Cu(OH)F•CuF2, CuF2•H2O or Cu(OH)2 Against hydrogen fluoride, copper is stable even at elevated temperatures. At a temperature >1000 K a corrosion of only <0.25 mm a–1 are observed.[88] Nickel: Against Hydrogen fluoride nickel is stable until 620 K, above this temperature a constant weight loss occurs. The corrosion rate for gaseous fluorine at 400 °C is 8 mm a–1. In liquid fluorine (–196 °C) the corrosion is less than 1 mm a–1.[88] Nickel also resists hot anhydrous hydrogen fluoride. At 500 °C the corrosion rate is 36 mm a-1.[88-90] Alloy 400: This material is also known as Monel. Alloy 400 is composed of ca. 65 % nickel, 33 % copper and 2 % iron. The resistance against fluorine is very good until a temperature of 800 K. The corrosion rate for gaseous fluorine at 400 °C is 6 mm a-1. In liquid fluorine (–196 °C) the corrosion is less than 1 mm a–1.[88] 26 Monel also resists hot anhydrous hydrogen fluoride. At 500 °C the corrosion rate is 42 mm a-1. Iron: Is slightly lightly stable against fluorine.[91] In liquid fluorine (–196 °C) the corrosion of stainless stainl steel type 304 is less than 1 mm a–1.[88] Stainless steel has only limited stability against hot anhydrous hydrogen fluoride. fluori At 500 °C the corrosion rate for the best suited steel 309cb is 230 mm a–1.[88] Silicon: Silicon does react with fluorine to the gaseous silicon tetrafluoride (Si + 4 F2→ SiF4).[92] The Jensen group measured the etching rate of silicon (100, boron doped 1-50 Ω) Ω in a set up using a one bar pressed flow with a fluorine partial pressure of 250 mbar. They found etching rates of 119 mm h-1 and a very rough surface after the exposure to fl fluorine. uorine.[93] In Figure 20 a silicon chip with a nickel layer on its bottom side is shown after the exposure for 66 hours in an atmosphere of 500 mbar partial fluorine pressure. The he nickel was not affected by the fluorine. The silicon on the opposite side was heavily etched, showing also the rough surface the Jensen group reported. Figure 20 12 µm nickel coated silicon chip after 66 h in 50 % F2 atmosphere at 1 bar. 27 2.4.7 Microreactors in general chemistry Microreactors have been shown to be advantageous in a series of chemical reactions. Some examples are shown in Table 4. Table 4 Reaction types which were carried out in microreactors Class Examples Liquid reactions Synthesis of azo dyes, Knoevenagel condensations Gas liquid reactions direct fluorination, hydrations Liquid solid reactions heterogeneously catalysed reactions, synthesis of nano particles. [16] [94] [23, 24, 95, 96] [97] [77] [97] Microreactors offer the possibility to integrate online or offline analytical tools into the system. Those can offer data, important for an improved insight into the reactions kinetics or the simple optimization of yield, time and selectivity of the reaction. Most often reaction mixtures are analyzed by single or combined methods of MS[98], GC[99] and HPLC[100]. But there are many more possible analytical tools. For example it is also possible to measure viscosity[101], heat conductivity [102] , liquid flow [103] or pressure [104] on micro scale with integrated sensors. Example 1: sodium azide reaction: a Bis-(trifluoromethyl)benzyl azide: To give a example for a chemical reaction using a microreactor, the conversion of bis(trifluoromethyl)benzyl chloride to bis-(trifluoromethyl)benzyl azide is briefly described. Using a high-pressure continuous-flow thermal tube reactor Kopach et al. carried out azide processes. They dissolved bis-(trifluoromethyl)benzyl chloride in DMSO and sodium azide in water. Both solutions reacted in a stainless steel tube reactor (0.64 mm ID x 63.1 m length) at 90 °C and 13.8 bar. The residence time in the reactor was 20 min. Afterwards the liquid was cooled to room temperature and was collected into a tank. After a extraction and two times washing they got a yield of 94 %. Batch experiments of the reaction are also described by Kopach et al.. The batch reactions were carried out at a maximal temperature of 40 °C. Because of the much lower temperature the reaction times were 1.5-4.0 h. This is 4.5-12 times longer as when the microreactor was used. The yield of the batch experiments was the same as using the reactor. For the optimization of the flow condition they attached a sample loop for a gas chromatograph to the tube after cooling to room temperature.[99] 28 This example shows the abilities of a microreactor to run potentially dangerous reactions much safer, even at higher temperatures. The increase of temperature then leads to significantly shorter reaction times. They also used the chance to link a GC to the system, for a fast reaction condition screening. Example 2 Fluorination with DAST: Microreactors have been used for fluorination reaction using diethylaminosulfurtrifluoride (DAST). DAST is a difficult to handle reagent, because it undergoes readily disproportionation to SF4 and (Et2N)2SF2. Explosions are possible. This happens at Temperatures above 90 °C. Additionally by-products are highly toxic and corrosive. DAST is used to convert alcohols into monofluorides or aldehydes and ketones into difluorides. Using the high heat transfer rates of microreactors reactions at temperatures of 70-90 °C are possible, reducing the reaction times dramatically. Batch reactions are normally carried out at –70-0 °C [105, 106] 2.4.8 Direct fluorination in microreactors Three scientific groups investigated the direct fluorination since 1999. Three different types of reactors have been developed since then: Nickel or stainless reactors with a cut straight channel, an etched silicon straight channel reactor and a stainless steel falling film microreactor. The research groups are presented chronologically to the date they started to work on this field. Chambers / Sanford Group: In 1999 R. D. Chambers et al. developed first a micro structured reactor for the direct fluorination. They cut a 100 mm long, ca. 500 µm deep and ca. 500 µm deep straight groove into a nickel or copper block, in which additionally cooling channels and thermocouple channels were drilled. The channel lid was made of polychlorotrifluoroethene.[59] Later this design was expanded to multichannel reactors.[107] Later also the falling film design of Jähnisch and Hessel et al. was used. The groups of Chambers and later Sandford have investigated and synthesized various fluorinated organic compounds via direct fluorination in microreactors.[20, 21, 107-113] A selection of substrates fluorinated by Chambers/ Sandford is shown in Figure 21. Their main focus was always the regio selective fluorination. Nevertheless also some experiments towards stereo selective fluorinations and perfluorination were carried out. 29 O O O A R'' O R' R O B R'' R' H RF R O F O O RF O G S R C H NO 2 E O 2N R D NO 2 O H O S OCH 3 NC I Figure 21 Substrates fluorinated by the groups of Chambers and Sandford: A: 1,3-ketoesters diketones [107, 114] , C: n-hexyl derivatives [112] , D: 3-nitrophenyl disulfide 2,5-(1,1,2,3,3,3-fluoropropane) tetrahydrofurane dicyanobenzene. [59] [107] , G: meldrums acid CN [107, 114] , B: 1,3- , E: substituted nitrobenzenes [113] [107] , F: , H: p-metoxy- benzaldehyde, I: p- [21] Oxidation by elemental fluorine: Elemental fluorine is a strongly oxidizing agent. Chambers and Sandford showed the possibility to oxidize alcohols into the corresponding ketones. Using a microreactor, 10 % fluorine and acetonitrile as solvent, they synthesized e.g. cyclohexanone (84 % conversion, 74 % yield) from cyclohexanol.[109] Oxidation by in situ synthesis of HOF-MeCN: The group of G. Sandford used a microreactor to synthesize the unstable, but strongly oxidizing agent HOF-MeCN. They focused on the synthesis of epoxide containing compounds, the oxidation of amines and Baeyer-Villiger like reactions. The double bond of e.g. (E)-diphenylethen (99 % yield) and both double bonds of (Z,Z)-1,5cylcooctadiene (94 % yield) were successfully converted into the corresponding epoxides. The oxidation of amines was investigated with the oxidation of e.g. 1-amino-n-alklyles (6095 %yield) and substituted aminobenzenes (51-74 % yield).[115] In the literature for batch approaches maximum reaction temperatures of –40 °C are recommended to prevent explosions. For easily oxidizable substrates a temperature of –78 °C should be used.[116] Using a metal microreactor for the synthesis of HOF-MeCN followed by the addition of the oxidizable substrate and a PTFE capillary tube reactor, the reaction could be carried out at room temperature and in high yields.[115] Before the two step microreactor was developed, an early example was the conversion of cyclohexanone into δ-valerolactone. This was one of the first reactions in which HOF was used for oxidation purposes. They used in a batch approach 10 % fluorine in nitrogen and wet 30 formic acid as solvent and investigated the role of the water within the reaction.[117] Later these conditions were transferred into microreactor experiments.[109] Jähnisch / Hessel Group: In 2000 K. Jähnisch and V. Hessel et al. published their work on the direct fluorination using a falling film microreactor and a multi channel microreactor. The falling film reactor had various 100 µm x 300 µm channels. The film thicknesses with acetonitrile were 30-37 µm. The used microreactor’s channels also had the size of 100 µm x 300 µm and could have slug flow as well as annular flows. They studied the fluorination of toluene in acetonitrile and methanol, using a toluene concentration of 1.1 mmol mL–1. Acetonitrile was superior to methanol in yield, selectivity and conversion in almost every experiment. For the fluorination experiments a fluorine excess of 20-100 times F2 to toluene was used. The maximal yields of monofluorinated toluenes with the falling film reactor were 20-28 %, which were two to three times better than those achieved with the classic minireactor. They reported selectivities around 30-50 %. As side products especially additions and polymerizations are mentioned. When the fluorine concentration was increased from 10 to 50 vol%, conversion and yield increased almost linearly. The selectivities were not influenced by the fluorine concentration.[23] Jensen Group: In 2003 DeMas et al. published a new type of microreactor for direct fluorination reactions. Instead of nickel or stainless steel they used a silicon chip as main material of their reactor. Because of the rapid formation of gaseous silicon tetrafluoride, when fluorine gets in contact to silicon, its surface had to be protected by an inert metal layer. The formation of the protective layer as well as the microstructures common silicon manipulation techniques were used (Figure 22). The channels were, after having a silicon nitride mask, KOH etched.[24] Etching of Silicon with KOH is an anisotropic process in which the etching rate is depending on the different orientation planes in the silicon crystal. This dependence leads to triangular structures.[118] In case of the microreactor an (100) etching was used. The channels did have a width of 435 µm, a depth of 305 µm and a length of of 2 cm. The angle of this triangle was 54.7° in respect to the plane of the wafer. To form a nozzle the backside of the chip was also etched till both triangles got in contact and hole was formed. The final protective layer was nickel with a minimum thickness of 200 nm on an adhesion layer of chromium (10 nm). An operation time of several hours without the observation of corrosion was reported. 31 Pattern using dry nitride etching Silicon with silicon nitride coating Nitride removed KOH channel etching Oxidized + metalized (Ni/Cr) Anodic bonding to metalized Pyrex Figure 22 Procedure to manufacture silicon based two channel microreactors for the direct fluorination. [24] They used this reactor for the direct fluorination of toluene. As solvents acetonitrile, methanol, octafluorotoluene and 1,1,2-trichloro-1,1,2-trifluoroethane were used. Typical concentrations used were between 0.1 and 1.0 mol ml–1. The use of acetonitrile gave the highest yield, which was about 10 % better compared to methanol. Octafluorotoluene resulted in a high conversion, but showed a high concentration of chain fluorination, which DeMas et al explained by free radical reactions. The use 1,1,2-trichloro-1,1,2-trifluoroethane as solvent led to a clogging of the reactor within 20 minutes. The selectivity towards single isomers was only weakly affected by the choice of solvent.[24] Later the improved the direct fluorination of toluene by a new 20 channel reactor and a variation of the flow conditions. The channels of the new reactor did have the same dimensions like the reactor before, but the output was enhanced. The liquid output was around 450 µl min–1. By increasing the velocity by the factor of five for the gas (10→50 mL min–1) as well as for the liquid (45→223 µL min–1) they improved the conversion from 63 to 76 % and the yield from 18 to 20 %.[119] 32 3. Equipment And Methods The fluorine fume hood and the reactors details will be discussed within this chapter. The reactor development and construction was made at the Institute for Micro System Engineering in Freiburg as part of Dr. P. Lang’s PhD theses. 3.1 Fluorine Fume Hood The fluorine fume hood at the Institute for Inorganic and Analytical Chemistry of Freiburg was a modified standard fume hood. The walls of the fume hood as well as the full ventilation system were made of V4 A stainless steel. The amount of ignitable material was reduced to a minimum. A 5 L fluorine gas bottle is placed inside the fume hood. For improved protection it is positioned in an additional stainless steel box, which directs a possible fluorine stream to the back of the fume hood. For handling of the gases, a monel line was available. All parts of it were delivered by Swagelok and using imperial sizes. A flow chart and a picture of the fume hood set up (grey background) are shown in Figures 23 and 24. Figure 23 The fluorine fume hood set up and the inside of the fluorine bottle box. Soda Lime Plow Phigh N2 or Ar NaF Peristaltic Pump Pulsation Soda Dampener Lime MFC MFC F2 Cryostat Figure 24 Flow chart of the gas delivery set up (grey background) and the set up for the minireactor #2. 33 3.1.1 Basic fume hood equipment: Fluorine Bottle The 5 L fluorine (99.98 %) was donated by Solvay Fluor GmbH. The pressure regulator (Matheson Valves & Fittings Ltd.) is attached to the fluorine bottle by using a lead seal. This seal was self stamped out. To check, if there is any leak, filter papers, wetted with sodium iodide solution, were placed close to all connections. In case of a leak the iodide gets oxidized by fluorine and forms coloured elemental iodine. All valves within the fluorine bottles extra steel box are equipped with length extended levers, to make it possible to use them from the outside of the extra box. There are three of those valves, the main bottle valve, the pressure regulating valve and a needle valve. Outside the box, another needle valve is connecting the fluorine system to the line. A larger bottle was not allowed to be installed, because of safety considerations. It needed to be placed at a well ventilated place, which is not given in normal indoor storages. Outdoor solutions were not possible because the building is listed to preservation. The small bottle in the fume hood was the only legal way to place the bottle. Inert Gas Supply: Nitrogen (99.996 %) and Argon (99.999 %) were supplied from a gas bottle cabinet. The pressure regulation valves are positioned outside the fume hood. Due to only one inert gas port, only Nitrogen or Argon could be used. If the reaction makes it necessary to handle both, one of them has to be connected to a normal handling port. Inside the fume hood there are two needle valves for inert gas. One is directly connecting the inert gas to the line; another is used to supply a gas flow controller. Vacuum: Vacuum is supplied by a Vacuubrand RZ6 rotary vane pump. Between pump and line a cold trap is installed. After a first needle valve a T-shaped connector with two further needle valves is added. This allows vacuum on the line and/or the soda lime tower. The on average achieved vacuum was around 2*10–2 mbar. Soda lime tower: It is a copper cylinder with a diameter of 12 cm and a height of 38 cm. ¼’’ monel tubes do connect it on one side to the main part of the line and on the other one also to the line, but to a, by three needle valves separated section next to the vacuum connection. The tower was 34 filled with granulated soda lime (Fisher Scientific). This substance is a combination of mainly calcium hydroxide and some potassium and/or sodium hydroxide. Its main application is the removal of carbon dioxide in anaesthetics and submarine breathing. It can also be used to quench fluorine and fluorides. The hydroxides react with fluorine to the corresponding fluorides and water (Equations 8 + 9) Ca(OH)2 + F2 → CaF2 + H2O + ½ O2 (8) Ca(OH)2 + 2 HF → CaF2 + 2 H2O (9) Line: The line was constructed of imperial sized electro polished monel tubes. All connections were sealed by Swagelok tube fittings. The port valves, the valves to the soda lime tower, to the inert gas supply and to the pressure gauges were ¼’’ Swagelok valves (Alloy 400 Integral Bonnet Needle Valve, 0.37 Cv, ¼’’. Swagelok Tube Fitting, Regulating Stem; Part Number M-1RS4). Barometers: Two MKS Baratron are attached to the line. A type 722A with a maximum of 10 mbar is used for low pressures, for high pressures up to 5000 mbar a type 750B was used. Both were using Inconel® and stainless steel for the exposed parts. They were controlled by a PR 4000B-F two channel digital power supply and readout for pressure gauges, pressure regulators, gas flow controllers and gas flow monitors.[120] Gas detectors: Two detectors were placed beneath the fume hood. They constantly measure the fluorine and hydrogen fluoride concentration within the laboratory. The system was installed by Dräger Safety AG & Co. KGaA. The controller is a Regard 2400 and the measuring cells are Polytron 3000. Two alert thresholds were set. The lower one was coupled to a red signal light. The higher one activates the red signal light and additionally a signal horn was activated. For fluorine this thresholds are set to 0.5 and 0.9, for hydrogen fluoride to 1.0 and 2.0. These detectors were checked by a Dräger service technician every half year. 35 3.1.2 Fluorination equipment: Additionally to the basic fume hood equipment, several more specialized vessels were needed for the fluorination reactions. In Figure 25 the set up for a minireactor reaction is shown. Mass flow controllers were used for the supply with fluorine and nitrogen and a peristaltic pump for the transport of the liquid substrate. To separate gas and liquid a gas liquid separator was used. All tubing were made of PFA, which exhibited the required chemical stability. A pulse dampener, placed in front of the reactor liquid inlet, was used to reduce the pulsing of the pump. For some experiments a syringe pump was used. This set up had no pulse dampener and instead of a gas liquid separator a PFA collecting flask. Behind all set ups, SodaLime filled PFA flasks were placed, to avoid any fluorine or hydrogen fluoride to be released into the atmosphere. mass flow controller peristaltic pump pump head reactor Tconnector gas liquid separator PFA tubing Figure 25 Picture of the set up for a minireactor reaction. A pulse dampener was normally used, but is not shown on this picture. Mass Flow Controller: Two MKS (MKS Instruments Deutschland GmbH) thermal mass flow controller (MFC) were used to handle fluorine and inert gas. Both were the M 330 type, with a flow range of 1-100 cm3 min–1. The fluorine correction factor compared to nitrogen is 0.98.[121] They are controlled by a PR 4000B-F two channel digital power supply and readout for pressure gauges, pressure regulators, gas flow controllers and gas flow monitors.[120] The control unit was placed outside the fume hood to control the flows without the need to open the fume hood. Pictures of the control unit and the mass flow controllers are shown in Figure 26. 36 Figure 26 Left: Picture of the used mass flow controller placed on a metal plate. Right: The control unit for the mass flow controller, in the flow rate mode (top) and set point mode. For the calibration one of the controllers was set to a certain gas flow. The valves of the second controller were closed. A water filled measuring cylinder was placed in a water filled beaker. The opening of the cylinder was placed under water to a have a water filled measuring cylinder in which nitrogen can be led in to measure the gas volume. In Table 5 the standard error for the fluorine gas flow controller was determined. The standard deviation was calculated with Equation x. ∑ (10) The standard deviation was found to be ±1.74 % of the target value. Table 5 Determination of the standard deviation of the fluorine gas flow controller. Gas flow was set to 4 ml min -1 Target [mL] Volume [mL] Difference [%] 53 52 -1.89 56 57 1.79 66 66 0.00 36 36.5 1.39 40 41 2.50 Standard deviation ±1.74 Syringe Pump: As syringe pump a Landgraf LA-30 laboratory syringe pump was used. To set the flow for this device needs to be programmed with the inner diameter of the syringe. To install a filled syringe it is advantageous to connect the syringe to a PFA tube before placing it into the syringe. Alternatively a Luer-Lock valve can be installed on the syringe. 37 Glas syringes are not suitable for conditions and systems with a higher backpressure, because some liquid can migrate along the stamp. This causes especially at low liquid flows and nonmeasurable loss, which makes reproducible conditions and quantitative analysis of a reaction impossible. Peristaltic Pump: The peristaltic pump was a Masterflex Variable Speed Drive (No. 07520-60) with a Masterflex PTFE Tubing Pump Head (No. 77390-00) and a PTFE tube (2 mm ID, 4 mm OD, No. R-77390-50). This Pump was chosen because of the chemical inertness of the PTFE to organic solvents, hydrogen fluoride and possible traces of fluorine. The speed control was analogue, an evaluation of flow speeds at the different speed levels are shown in Table 6. [122] The pump head worked with six steel rollers to handle the stiff PTFE tubing. Though six rollers are already more than most peristaltic pumps are using, pulsation of the liquid, especially at low speeds, was observed. Table 6 Flow measurement for the peristaltic pump. –1 [122] –1 Speed Level Flow [ml h ] Speed Level Flow [ml h ] 2 42.63 4 236.56 2 26.25 4 226.28 2 27.87 4 241.20 average 32.25 average 234.68 standard deviation 9.02 standard deviation 7.64 standard deviation [%] 27.98 standard deviation [%] 3.25 3 138.76 5 335.52 3 161.10 5 319.95 3 160.16 5 340.14 average 153.34 average 331.87 standard deviation 12.64 standard deviation 10.58 standard deviation [%] 8.24 standard deviation [%] 3.19 Gas-Liquid Separator: The product flow exiting the reactor consisted of a liquid and gas, which could contain nitrogen, hydrogen fluoride, fluorine, volatile liquids and decomposition products. Because no commercially solution was available, various self-made versions were constructed. All are based on PFA tubes. As most important feature, the inner diameter was identified. To reduce 38 the amount of minimum of liquid per experiment the volume of the gas liquid separator had to be minimal. Experiments with tubing with an inner diameter of ¼ inch OD and 1/16 inch OD capillaries as inlet for the products and outlet connection to the PFA soda lime trap all failed because of foam formation. To prevent this, a wider PFA tube (1/2’’ OD, ≈0.9 cm ID) were used. Foam formation still can occur, but within the normal flow rates, it did not reach the gas exit at the top of the gas liquid separator. Pictures of the finally used version is shown in Figure 27. to soda lime trap from reactor to reactor Figure 27 Final version of gas liquid separator. Left: Construction of gas liquid separator. Right: Gas liquid separator in use (some foam inside the tube). Pulsation Dampener: To reduce the pulsation typical for peristaltic pumps a small pulsation dampener was constructed of a PFA tube (1/2’’ OD, ≈0.9 cm ID) and a 5 mL plastic syringe. The one end of the PFA tube was connected to a 1/8 inch tube. Another 1/8 inch tube was attached to a hole in the middle of the tube. The end of the syringe was cut off and stuck into the still open end of the PFA tube. Using Teflon band and Parafilm the syringe was fixed. During operation it was ensured that the liquid has no contact to PFA tubes only. Using the plunger, the gas volume and the liquid level could be adjusted. The larger the gas volume the better was the pulsation dampening. But more liquid was needed to achieve the counter-pressure. In Figure 28 the principle structure of a pulsation dampener is shown. Additionally to separate pulsation dampener long tubings also are able to reduce pulsation, but this works better for soft tubings. The PTFE is too stiff to have a significant dampening effect. 39 Figure 28 The principle structure of a pulsation dampener. PFA Soda Lime trap: Behind all reaction set-ups a soda lime filed 120 mL PFA trap was placed (Savillex, 120 mL column component vessel, flat interior, 1-1/2", Figure 29). For the connection, two 1/8 inch PFA tubes were used. The tube connected to the reactor set-up was introduced close to the bottom of the flask. Figure 29 Picture of the soda lime trapped, which was connected to the gas exit of the gas liquid separator. 3.2 Microreactors The reactor construction and design was performed in the group of P. Woias et al.. Full details on the reactor design and construction were published.[73, 75] 3.2.1 Minireactor #1 The minireactor was based on three copper blocks (Figure 30). The bottom block was holding the reactor channels and it’s in and outlets. Those were cut into the copper with diameter of 1x1 mm. The for the liquid and gas supply as well as for the product outlet 1/8 ''. Swagelok® connectors were used. The middle block was the lid for the reactor channel and holding the cooling channels on its back. The reaction channels as well as its lid were nickel coated to improve the stability of the reactor against fluorine.[91] The cooling was made of a cut circle 40 deepening in which diamond shaped structures remained to lead to an even distribution of cooling liquid. The third block was used as lid for the cooling channels and for the connectors to the cooling medium. The channel length before the gas inlet was 6.6 cm. To protect the copper against the fluorine, a 10-20 µm nickel was galvanically deposited. Figure 30 Microreactor #1: Left before being assembled. Layers: 1. Reaction channel plus in- and outlets; 2. cooling channels; 3. connections fort the cooling medium. Middle: Assembled reactor. Right: connected reactor. Flow Characterisation: The minireactor was designed for slug flow. To get information about the flow regimes within the reactor flow experiments were carried out before the reactor was finally assembled. For this purpose 2-propanol and air was led through the reactor in a range from 5- 400 ml h–1. To achieve a shorter exposure time and to have a better visibility of the slug ends the channels were lighted by a green and a blue LED. Those LED’s were placed opposite each other in direction of the long channel sides. For each flow setting nine pictures were taken. In Figure 31 a typical picture is shown. Green LED Blue LED Liquid inlet Gas nozzle Figure 31 Example of a photo during the characterisation of the flow regimes. The nitrogen as well as the 2–1 propanol flow were set to 100 mL h . For the analysis of the bubble and slug lengths Adobe Illustrator® CS3 Version 13.0.2 was used. The lengths were calibrated by the known channel width. The results of this characterization are shown in Table 7 and Figure 32. Most value sets formed a slug flow, only for liquid flows of 5 and 10 mL h–1 and gas flows higher than 200 mL h–1 the flow changed to an annular type flow. With a 2-propanol flow of 300 and 400 mL h–1 and a gas flow of 10 mL h–1 the pressure achievable, was not sufficient to create a gas flow. 41 Table 7 Results of the flow characterization of the minireactor #1, calculations by P. Lang Liquid Gas flow flow [ml/h] [ml/h] Total flow [ml/h] Total velocity [m/s] 5 5 5 5 5 5 5 10 25 50 100 200 300 400 15 30 55 105 205 305 405 10 10 10 10 10 10 10 10 25 50 100 200 300 400 25 25 25 25 25 25 25 Bubble [mm] Surface [m2] 0.94 1.71 1.31 3.92 - bubble min. 2.49E-05 3.96E-05 4.84E-05 9.54E-05 - bubble max. 3.09E-05 4.96E-05 6.07E-05 1.21E-04 - 3.35 8.40 8.99 15.73 23.41 30.71 - 0.33 0.86 0.82 1.38 1.46 2.52 - 1.27E-05 2.86E-05 3.04E-05 5.16E-05 7.57E-05 9.87E-05 - 0.097 0.139 0.208 0.347 0.625 0.903 1.181 1.36 3.20 3.91 6.38 10.88 13.95 17.83 0.12 0.30 0.33 0.49 0.90 1.06 1.92 60 75 100 150 250 350 450 0.167 0.208 0.278 0.417 0.694 0.972 1.250 1.60 2.16 2.71 3.75 5.76 7.58 9.45 10 25 50 100 200 300 400 110 125 150 200 300 400 500 0.306 0.347 0.417 0.556 0.833 1.111 1.389 200 200 200 200 200 200 200 10 25 50 100 200 300 400 210 225 250 300 400 500 600 300 300 300 300 300 300 300 10 25 50 100 200 300 400 400 400 400 400 400 400 400 10 25 50 100 200 300 400 average Stabw 0.042 0.083 0.153 0.292 0.569 0.847 1.125 7.23 11.91 14.70 29.66 - 20 35 60 110 210 310 410 0.056 0.097 0.167 0.306 0.583 0.861 1.139 10 25 50 100 200 300 400 35 50 75 125 225 325 425 50 50 50 50 50 50 50 10 25 50 100 200 300 400 100 100 100 100 100 100 100 Liquid [mm] [75] Frequency [Hz] Surface [m2/m3] average Stabw average min 3.00 2.08 1.69 1.37 - 0.70 0.41 0.29 0.14 - 14 40 90 213 - 2.49E+05 1.14E+06 3.14E+06 1.46E+07 - 1.54E-05 3.56E-05 3.79E-05 6.49E-05 9.56E-05 1.25E-04 - 3.86 1.62 1.56 1.26 0.98 0.75 - 0.46 0.13 0.14 0.19 0.39 0.12 - 14 60 107 243 595 1153 - 6.60E+04 6.16E+05 1.17E+06 4.52E+06 1.62E+07 4.10E+07 - 6.47E-06 1.22E-05 1.45E-05 2.22E-05 3.64E-05 4.60E-05 5.82E-05 7.38E-06 1.47E-05 1.76E-05 2.75E-05 4.55E-05 5.77E-05 7.33E-05 15.93 2.45 1.92 1.26 0.99 0.93 0.80 3.17 0.16 0.21 0.13 0.12 0.16 0.18 6 57 108 275 632 971 1476 5.69E+03 1.00E+05 2.26E+05 8.81E+05 3.31E+06 6.43E+06 1.24E+07 0.07 0.08 0.15 0.21 0.29 0.66 0.49 7.24E-06 8.97E-06 1.07E-05 1.40E-05 2.03E-05 2.60E-05 3.19E-05 8.36E-06 1.06E-05 1.28E-05 1.69E-05 2.50E-05 3.23E-05 3.98E-05 7.86 3.26 2.49 1.46 0.98 0.88 0.74 1.03 0.33 0.17 0.19 0.07 0.06 0.08 21 64 111 286 709 1108 1682 1.11E+04 4.13E+04 8.61E+04 2.87E+05 1.04E+06 2.08E+06 3.86E+06 1.49 1.72 2.01 2.41 3.53 4.47 5.31 0.13 0.10 0.12 0.16 0.12 0.33 0.48 6.87E-06 7.60E-06 8.51E-06 9.76E-06 1.33E-05 1.62E-05 1.89E-05 7.88E-06 8.82E-06 9.97E-06 1.16E-05 1.60E-05 1.98E-05 2.32E-05 11.22 5.57 3.70 2.52 1.66 1.34 1.08 1.67 0.32 0.27 0.15 0.09 0.12 0.10 27 62 113 220 503 832 1282 6.73E+03 1.71E+04 3.45E+04 7.74E+04 2.41E+05 4.86E+05 8.72E+05 0.583 0.625 0.694 0.833 1.111 1.389 1.667 1.19 1.18 1.31 1.49 2.01 2.36 2.71 0.10 0.14 0.06 0.07 0.16 0.14 0.19 5.95E-06 5.91E-06 6.33E-06 6.87E-06 8.53E-06 9.62E-06 1.07E-05 6.71E-06 6.67E-06 7.20E-06 7.88E-06 1.00E-05 1.14E-05 1.28E-05 8.68 5.91 3.99 2.96 1.84 1.46 1.23 0.87 0.35 0.29 0.22 0.15 0.18 0.13 67 106 174 282 605 953 1354 7.20E+03 1.12E+04 1.98E+04 3.48E+04 9.28E+04 1.65E+05 2.61E+05 310 325 350 400 500 600 700 0.861 0.903 0.972 1.111 1.389 1.667 1.944 0.95 1.05 1.24 1.54 1.72 1.86 0.08 0.10 0.12 0.10 0.10 0.10 5.19E-06 5.49E-06 6.11E-06 7.04E-06 7.61E-06 8.03E-06 5.75E-06 6.13E-06 6.92E-06 8.10E-06 8.83E-06 9.36E-06 8.67 5.62 3.16 2.00 1.47 1.20 1.25 0.62 0.34 0.22 0.05 0.12 104 173 351 695 1130 1621 6.49E+03 1.14E+04 2.57E+04 5.87E+04 1.03E+05 1.56E+05 410 425 450 500 600 700 800 1.139 1.181 1.250 1.389 1.667 1.944 2.222 0.99 1.07 1.22 1.33 1.50 1.54 0.11 0.06 0.13 0.07 0.09 0.08 5.31E-06 5.57E-06 6.02E-06 6.37E-06 6.90E-06 7.05E-06 5.90E-06 6.23E-06 6.81E-06 7.25E-06 7.92E-06 8.12E-06 5.53 3.55 2.69 1.88 1.40 1.24 0.60 0.20 0.21 0.12 0.10 0.10 213 352 517 888 1392 1791 1.02E+04 1.76E+04 2.80E+04 5.09E+04 8.64E+04 1.14E+05 42 40 30 20 10 0 0 200 400 Flow 2-propanol [ml/h] Slug length at constant gas flow Bubble length 400 mL/h Bubble length 300 mL/h Bubble length 200 ml/h Bubble length 100 ml/h Bubble length 50 mL/h Slug length [mm] Bubble length [mm] Bubble length at constant gas flow 20 15 10 5 0 0 20 10 0 0 200 Gas flow [ml/h] 400 400 Slug length at constant 2-propanol flow Bubble length 400 mL/h Bubble length 300 mL/h Bubble length 200 ml/h Bubble length 100 ml/h Bubble length 50 mL/h Slug length [mm] Bubble length [mm] Bubble length at constant 2-propanol flow 40 30 200 Flow 2-propanol [ml/h] slug length 400 mL/h slug length 300 mL/h slug length 200 mL/h slug length 100 mL/h slug length 50 mL/h 20 15 10 5 0 0 200 Gas flow [ml/h] 400 slug length 400 mL/h slug lenght 300 mL/h slug length 200 mL/h slug length 100 mL/h slug length 50 mL/h Figure 32 Visual presentation of the results of the flow characterization of the minireactor #1. The experiments showed the possibility to vary the bubble length from 1.0 up to 30.7 mm and the liquid slugs from 0.8 to 15.9 mm. There was also a large difference in the frequency of bubble formation. For low total flows it was found to be 15-100 Hz. For the largest total flows values close to 1800 Hz were achieved. Design problems of minireactor #1: The design of the minireactor #1 had two very important problems, which became obvious during use. The inlets and the outlet were connected from the bottom of the reactor and the screw of the connector did not match to the hole drilled, and formed a dead volume. During use, liquid got into the nozzle and was trapped inside the dead volume. This trapped liquid was fully exposed to the fluorine and led to the formation of a blackish material, which caused clogging of the gas inlet. Pictures of the blackish material and a scheme showing the dead volume are shown in Figure 33. Figure 33 Left: Gas inlet after approx. 50 h of use, and directly after a clogging of the flow. Middle: scheme of the gas inlet showing the dead volume between screw and nozzle. Right: Blackish material also contains copper or nickel, it forms a blue solution if added to aqueous ammonia solution. 43 Additionally to the problem with the gas inlet the seal between the reactor plate and its lid plate was too thick. This formed a small slit in which solid organic material was found when the reactor was opened. It was only found close to the inlets, approximately within the first 15 cm of the reactor channel. In addition to the deposits the nickel coating was damaged. This damage only appeared were the organic material was found. A possible explanation for the damage of the protective layer can be a local hot spot, which was caused by organic material reacting within the dead space of the slit. A scheme of the reactor and pictures of the deposits and damages are shown in Figure x. Figure 34 Minireactor #1 after being opened. Top left: schematic view of the reactor showing the slit between the reactor plate and its lid plate caused by the seal ring; top right: close look on the reactor channels close to the inlets; bottom left: close lock on the channels in the middle section of the reactor channel; bottom right: reactor channel overview. 3.2.2 Minireactor #2 It was constructed out of four nickel coated copper blocks. All channels and holes were conventionally machined. The meandered channels measure 1x1 mm in their profile, with 90° corners. A top view of the channels is shown in Figure 35. The reactor was designed for slug flow. Figure 35 left: microreactor #2 before final assembly. Layers: 1 Liquid and Gas connectors; 2 Channels for temperature sensors and reaction channel; 3 heat sink; 4 coolant connectors In comparison to the minireactor #1 a modified heat sink was included and optimized for a maximum cooling at the point of the gas inlet. Five T-type thermocouple sensors were placed 44 inside the reactor block in 1 mm distance from the reaction channels themselves. They were placed 1, 2, 4, 29 and 53 cm behind the gas inlet. The minireactor had a reaction channel length of 53.7 cm from the junction of the gas inlet and the main reaction channel. The channel length before the gas inlet was 11 cm, which is approx. double the length, which the minireactor #1 came up with. This was made to ensure the liquid to be cooled or heated to the reactor temperature. The connection of liquids and gases was now set to the top of the reactor and the gas inlet. Figure 36 shows the reactor before assembly. Gasket Screws 1 2 3 4 Walls of cooling channels Cooling outlet Product outlet Liquid inlet Gas inlet Cooling inlet Temp. sensors Figure 36 left: top view on the reactor channel plate. On the left the two inlets, on the right the outlet. Middle: Scheme showing the position of the temperature sensors and the alignment of the cooling. Right: channels for temperature sensors. To determine, if a sufficient cooling in front of the inlet is provided, room temperature acetone was led into the liquid inlet and out of the gas inlet. The reactor was cooled to –12.9 °C and the temperature was measured directly, when leaving the reactor through the gas inlet. Acetone flows between 30 and 240 mL min-1 were tested. Even at the highest temperature the acetone temperature adapted to the reactor temperature within the 11 cm of reaction channel plus the in and outlets (Figure 37). The energy transferred at the maximum liquid flow was 13.03 kJ h–1. This value can be compared to the energy during the fluorination reaction. The maximal possible fluorine flow with the minireactor #2 was 240 mL h–1, with a reaction enthalpy of 473 kJ mol–1 this would be 10.01 kJ h–1. This shows that the energy during the reaction can be absorbed within the 11 cm for the idealized one phased case. The power for the maximum flow was 3.6 W. Together with the channel surface it leads to a heat transfer coefficient (Equation 11) of 245 W m–2K–1. (11) ∆ 45 Temperature [ C] Aceton flow [mL min-1] -2.00 32 153 235 Reactor temp. -7.00 Gas inlet temp. -12.00 -17.00 Figure 37 measured temperatures for the determination of the heat transfer. 3.2.3 Capillary reactor A common low cost approach to realize a microreactor is the use of a coiled capillary. A capillary reactor was build to one the one hand have a comparison to the much more complex minireactors, and on the other hand to have a backup system, when no other reactor is available, due to construction times. The total cost of the system build was around 100 €. The basic works were done as part the bachelor theses of P. Bendix [122]. As a mixer a stainless steel Swagelok T-connector with inner diameters of 1/16 inch was used. This is with a ID of 1.27 cm still too large to be micro scale by its definition. As capillary a 40 cm long PFA or stainless steel tubes were used. It inner diameter was 0.5 mm. A picture of the reactor is shown in Figure x. The coiled capillary was placed within a cooling bath. With the capillary reactor slug as well as tubular flows could be realized. Because of the pressure drop, slug flow only can be realized at short tube lengths (ca. 40 cm), when using modest pressures. Experiments have shown for a 100 cm tube an entry pressure of 500 mbar as not sufficient to move the plugs within the tube. For this experiment a liquid flow of 2 mL h–1 and a gas flow of 6 ml h–1 were used. At approx 9 ml h-1 a wavy annular type of flow begins for the 100 cm tube. This reactor could be used with the same set up as the minireactors described in previously. Figure 38 Left: tube reactors inlet feed set up. Right: tube reactor in use 1: 1/8‘‘ PFA tube for gas feed, 2: 1/8‘‘ PFA tube for liquid feed, 3: T-connector, 4: 1/16‘‘ PFA capillary, 5: PFA sample collecting flask, 6: copper coil in a cooling medium . 46 3.2.4 Characterisation of the used microreactors Calculation of film thicknesses: The film thickness of a slug is primarily dependent on the capillary number (Ca, Equation 12). It is the relation of viscous forces and the surface tension at a phase boundary. µ is the dynamic viscosity, u the fluids velocity and γ the surface tension. A typical propylene carbonate flow 2.5 mL min-1 and a gas flow of 5.8 mL min-1 were used to calculate the velocity. # !" (12) $ For a circular channel Aussillous et al. developed Equation X to calculate the film thickness.[123] %&'( )*+, 3 -./0124 (13) 3 567.77124 For a rectangular channel Kreutzer et. al. [74] proposed Equation x to calculated the film thickness in the diagonal direction: +*++',,9:*;<, =>;??,' 0.7 B 0.5 D.DE12F,GGH (14) The calculation of Reynolds- and Capillary numbers as well as the calculation of the film thickness is shown in Table 8. Table 8 Comparison of Reynolds number, capillary number and the slug flow film thickness for the minireactor, the tube reactor and the planned microreactor -1 Reactor v [m s ] Re Ca film thickness [µm] Minireactor 0.138 7 0.08 43 Tube reactor 0.138 3 0.08 39 Microreactor 0.138 2 0.08 13 Minireactor 0.138 7 0.08 43 Tube reactor 0.552 13 0.33 61 Microreactor 1.534 22 0.91 51 47 Calculation of pressure drop: With the law of Hagen–Poiseuille (Equation 15) the pressure difference (∆p) needed for a certain volume flow can be calculated, if length of the tube (l) and the viscosity of the liquid (η) is known. IJ KL G MN (15) O P Most important is the relation of the pressure difference and the tube radius. The radius is used by the power of four. Which makes it by far the dominating value. If a channel diameter is reduced to its half, the pressure difference needed will be increased by a factor of 16. To make this visible in Figure x are pressure differences for three different channel diameters comparable to the reactors used or planed (the law of Hagen-Poiseuille is optimized only for round tubes). As length the 53.7 cm of the minireactor #2 and as liquid propylene carbonate was chosen.[66] 10 Table 9 Calculated pressure differences for different -1 Channel diameter [mm] ∆ p [bar] 1.0 0.075 0.5 1.207 0.3 9.311 6 4 ∆p [bar] 8 channel diameters. With a total flow of 350 mL h . 2 350 ml h-1 0 0.9 0.7 0.5 0.3 Channel diameter [mm] 100 ml h-1 Figure 39 Calculated pressure differences in dependence of the channel diameter. With total flows of 350 mL h -1 -1 and 100 mL h . Bond and Weber numbers: This two numbers, like the Capillary number, describe the influences of different forces within a two-phased fluidic system. Bond number The Bond number Bo (Equation 16) describes the influence of the surface tension and it is defined as: QR ∆ S > (16) In the equation, Δρ is the density difference of the two fluids, g is the gravitation acceleration, dh the hydrodynamic diameter and σ the surface tension.[124] 48 For propylene carbonate at 25 °C the Bond number for the 1x1 mm channels of the minireactors is 0.28. For the planned microreactor the value would decrease to 0.025. For values below 0.1 the surface tension is clearly dominant over vicious forces.[125] The minireactors value is still close to this value. The surface tension can be neglected when the Bo value is >11. [124] Weber number The Weber number (Equation17) describes the influences of inertial and interfacial forces. This number can be used to estimate the flow regime of a multiphase flow. The influence of the gas velocity is superior to the liquid flow, that is the reason why often only the gas flows are used for the calculation. WeG is defined as: T U V > (17) ρ is the density of the liquid, uG the superficial gas velocity, dh the hydrodynamic diameter and σ the surface tension. Below a WeG ≈1 the surface tension dominates, which leads to slug and bubbly flows. Above a WeG of ≈20 the inertial forces are dominating and an annular typed flow. In between a transitional regime with a frothy slug-annular flow develops.[124] For the minireactor with propylene carbonate WeG values of 0.03 to 0.28 can be found for gas flows of 2-6 mL min-1. These values indicate a surface tension dominated flow. 3.2.5 Planned microreactor As next generation of microreactors for direct fluorination a silicon based reactor was being developed in the Woias group by P. Lang [75] and K. Cobry. As basic components 2x2 cm silicon chips were used. The reaction channels were planned to be 300x300 µm in size and having a meandered layout to induce additional flows inside the liquid. They could be formed by well known silicon etching techniques. To prevent the silicon to be etched by contact with hydrogen fluoride or fluorine a protective metal layer was necessary. The metals could be placed on the silicon using sputtering or galvanic techniques. The results of the material tests for the protective layer are described in section 4.1. A picture of a test microreactor as well as its construction scheme is shown in Figure 40. 49 Figure 40 Left: Construction scheme of the reactor plate for a 2x2 cm microreactor silicon chip (by P. Lang [75] ) Right: SEM picture of the reaction channels (chip and picture pictu by K. Cobry, IMTEK Freiburg). The key feature of the microreactor was its design to be operated in a bubbly flow regime. This was achieved by an optimized nozzle geometry by P. Lang.[75] The nozzle was etched using potassium hydroxide, which creates a pyramidal hole hole into the silicon, which can reach the opposite side of the chip and form a nozzle (Figure 41). By changing the geometry of the etching mask the nozzle geometries and sizes can be modified. Intensive studies have been carried out at the IMTEK to create smaller smaller bubbles. By using micro sized bubbles on the one hand the reaction is planned to be increased in speed to have an improved space time yield. On the other hand the dividing of the fluorine into very small bubbles decreases the heat formation on a liquid uid segment. Figure 41 Left: Connective points of a prototype microreactor chip. Middle: close up of the gas nozzle. Showing the pyramidal form. 3.2.5 Set up for silicon chip coating tests Silicon based microreactorss for the direct direct fluorinations do need a protective layer to separate the silicon from the fluorine. To develop a long time stable coating P. Lang used different methods to form metal layers of copper, gold and nickel. To prevent the fluorine gas two chip holders were developed. The first one was not equipped with PTFE seals. This led to a significant fluorination of the unprotected back. To prevent this influence the second one was equipped with three PTFE O-rings. O rings. Photos and schemes of the holders are shown in Figure 42.. Both holder were developed and constructed at the IMTEK. 50 Figure 42 Scheme and pictures of chip holder #1 and #2 and the monel autoclave. In the scheme red squares show PTFE O-rings. 3.3 Fluorination Of Metal Salts A method for the fluorination of 1 to 5 g metal salts was established. As reaction vessel a 130 mL PFA flask was used. It was connected to the fluorine line by a ¼ inch PFA tube. The finely powdered salt was placed together with a magnetic stirrer bar into the flask. The flask was tightly closed and high vacuum was applied till the pressure was constant. Then the connection to the line was closed. And a certain fluorine pressure was applied to the line. Then the stirrer was activated in a manner, which made the stirrer bar hit the flask walls regularly to improve the agitation of the powder and preventing powder to be left unstirred in the corners of the flask. Leaving the flask some space for moving increases the effect. After the reaction the fluorine was removed. This technique can be applied for air sensitive substances by inserting a needle valve into the connecting tube. This way the flask can be filled inside a glove box and then being connected to the line. A picture of the set up is shown in Figure 43. Figure 43 Set up for the fluorination of metal salts by using a PFA flask connected to the fluorine line and a stirrer bar. 51 4 Results Most results described in this thesis were direct fluorination reactions using the minireactor #1, the minireactor #2 or the capillary reactor. Additionally, the results of the silicon chip coating tests and a batch direct fluorination of closo-Na2[B12H12], which was carried out together with M. Rühle,[126] are described. The fluorination experiments can be separated in two mayor phases. In the first one a syringe pump (see also section 3.1.2) was used. This allowed only a single fluorination per run and a fluorine to substrate ratio of approx. 5 %. The later experiments were carried out using a peristaltic pump system (see also section 3.1.2), in which the product mixture is separated from the waste gases and nitrogen and led back to the reactor. This led to a repeated fluorination and higher fluorine to substrate ratios were possible. 4.1 Silicon Chip Coating Tests To test if the surface of a metal coated silicon chip is fluorine resistant it was exposed to a static fluorine (+ nitrogen) atmosphere for a certain time. The main result of these experiments could easily judged by optical investigation. In case of a leak in the coating, in the worst cases large areas of the protecting metal were lifted off. In less worse cases a hole could be found on the back side of the chip. The fluorination time was up to 22 hours, in which even small punctures developed to holes. As materials for the coatings gold, copper and nickel were used. In Figure 44 two chips, covered with a nickel layer, are shown after being fluorinated. They well illustrate the effect of an insufficient coating. Figure 44 Two nickel protected silicon chips after fluorination with 1 bar 100 % fluorine for 4 h. Left: 16 µm nickel coating. Right: 6 µm nickel coating. 52 4.1.1 Gold coating Galvanically gold coated silicon chips were fluorinated for 18 h with a 30 % fluorine/ nitrogen mixture. At the outer edges of the chip the gold started to roll up and through leaks the silicon was attacked on a. This showed on the one hand leaks in the gold layer, on the other hand a lack of adhesion between the layers. The gold itself showed no sign of reaction. A currentless gold coated silicon chip was coated with a 0.109 µm gold layer on a palladium start layer. This chip was fluorinated for 22 h with a 50 % fluorine/nitrogen mixture. Especially on the edges of the channels the gold layer broke and rolled up. The fluorine could react with the silicon. This showed again the leaks and a lack of adhesion. Pictures of fluorinated gold protected chips are shown in Figure 45. Figure 45 Silicon chip coated with 0.109 µm gold, after being fluorinated. 4.1.2 Copper coating A current less copper coated (0.53 µm) silicon chip was fluorinated for 3 h with a 50 % fluorine/ nitrogen mixture. Especially on the edges of the channels the copper layer broke and rolled up. The fluorine could react with the silicon. This showed on the one hand leaks in the copper layer, on the other hand a lack of adhesion between the layers. The copper itself showed no reaction. A picture of the chip is shown in Figure 46. ´ Figure 46 Silicon chip coated with 0.53 µm copper after being fluorinated. Front and back side. 4.1.3 Nickel coating Evaporated nickel A silicon chip was coated with nickel by sputtering. This formed a 200 nm thin nickel layer. This coating was able to protect the silicon against a 50 % fluorine/nitrogen mixture for 1.5 h. 53 A picture is of the chip is shown in Figure 47. But due to the complex coating process using evaporated nickel, a coating by galvanization techniques was w preferred. Figure 47 Silicon chip coated with 200 nm nickel by evaporation after being fluorinated. Front and back side. Galvanized nickel A test series of chips with 6, 12 and 18 µm galvanized nickel coatings on a palladium start layer was carried out. The chips were fluorinated with 1 bar 100 % fluorine for 6 h. To measure the time until til the protecting layer broke, the fluorine consumption was measured by checking the drop of fluorine pressure in the system. The results are shown in Figure 48. The 6 µm nickel chips showed leaks at the channel edges after 11 3.5 h. One of the 12 µm nickel chips started to show small fluorine consumption after 3.5 h. This leak appeared only at a the IMTEK logo (Figure 49,, right picture). This logo had structures of 50x50 µm, which were more difficult to coat than the planed 300 µm reaction channels. All 16 µm nickel coated chips showed no sign of any leak within the 6 h of testing. With this result, res at least under static fluorine conditions, galvanically applied nickel seems to be well suited as protection layer for the microreactors, s, when it is used in a sufficient thickness. One of the stable 16 µm chips and the one with a 12 µm coating with thee attacked silicon at the IMTEK logo are shown in Figure 49. Fluorine Consumption [%] 25.0 6µm/40mA/Ni-Galv 20.0 6µm/50mA/Ni-Galv 15.0 12µm/50mA/Ni-Galv Galv 10.0 12µm/40mA/Ni-Galv Galv 5.0 18µm/30mA/Ni-Galv Galv 0.0 0 -5.0 200 400 Time [min] Figure 48 Testing of nickel galvanized silicon chips. 54 18µm/40mA/Ni-Galv Galv Figure 49 Pictures of nickel galvanized silicon chips. Left+middle: 16 µm nickel, front and back. Right: 12 µm nickel, leak at the IMTEK logo. Gold coated galvanized nickel Two chips which were protected by first 10 µm nickel and then by 10 µm gold on a tempered 200 nm palladium start layer. The additional gold layer on the already tested nickel layer was used as a possible connector between two reactor layers. Unlike nickel, gold can be easily used in bonding techniques in which the connections of two metal layers is created by heat and pressure. The soft gold readily diffuses into the gold of the second chip at relatively mild conditions. The formation of metal silicon alloys can be suppressed. Both chips were exposed to 1 bar 100 % fluorine for four hours. Both did not show any change on their front. The pressure lowered slightly during the fluorination, which was caused by a small leak on the back side Teflon O-ring. At the place of the leak the silicon was locally etched. Pictures of a chip before and after the fluorination are shown in Figure 50. Figure 50 Nickel-gold coated chip after 4 h 100 % fluorine (1 bar) exposure. 4.2 Direct Fluorination Experiments Mainly small organic molecules were directly fluorinated, e.g. toluene or cyclic carbonates. Additionally, the direct fluorination of closo-[B12H12]2- was investigated and tetraalkyl ammonium salts were tested for the direct fluorination. 4.2.1 Choice of substrates Acetonitrile: Acetonitrile is the most often used solvent in direct fluorination reactions in microreactors.[21, 23, 24] Its positive effect, as an aprotic, polar solvent, on the direct fluorination was calculated and discussed by H. Fukaya und K. Morokuma (see section 2.4.1).[53] In the direct fluorination of toluene it was found to be better suited as the protic, polar methanol.[23] 55 Acetonitrile also can be fluorinated itself. Possible isomers are CH2F-CN (b.p. 81-80 °C [127]), HF2-CN (b.p. 22-23 °C [127] ) and CF3-CN (b.p. -64 °C). In electrochemical studies also the formation of C2F5NF2, C2F6, C2F5H, NF3, CF4 and CF3H was observed.[128] When being used as solvent in a direct fluorination the formation of polymeric fluorination products and the incorporation of nitrogen in aromatic substrates were reported.[129] Acetonitrile is an organic, polar and aprotic solvent with a boiling point of 81.6 °C.[130] It has a relatively high dielectric constant is εr= 35.09 (25 °C) [131] and a low dynamic viscosity of 0.35 mPa s. Toluene: The fluorination of toluene was chosen, because it was the standard substance reported in the prior literature on direct fluorination in mini/microreactors.[23, 24] The broad use of fluorinated aromatics in the literature [95] and the wide availability of the substance makes toluene a good choice. Typically, four mono fluorinated toluene isomers, o-, m- and p-fluorinated toluene as well as benzyl fluoride are formed. Measurements of the ratio between the isomers can be used to determine the selectivity of the reaction. However, bad reported fluorination yields, below 20 %, were largely caused by low selectivities with concomitant appearance of many side products, like the formation of high molecular species.[24, 132] This potentially brings into question the suitability of toluene as a model system. Toluene has a relatively high boiling point of 110.06 °C and the regioselectivity of the reaction can be easily observed by four different hydrogen sites for mono fluorination. Its dynamic viscosity is 0.56 mPa s and its dielectric constant εr= 2.43. These values are typical for low polar solvents. Methyl tert-butyl ether (MTBE): The fluorination of this ether was of interest because Chambers and Sandford et al. had published on the feasibility to fluorinate primary ethers.[22] The simple MTBE offers only two chemically different hydrogen atoms, and thus appeared as an interesting model system to investigate the selectivity of O-CH3 vs. C-CH3 fluorination. Its boiling point is with 55.1 °C relatively low and it had with 0.32 mPa s the lowest viscosity of the substrates tried. The dielectric constant was with 2.6 similar to toluene. 56 iso-Propylacetate Alternatively to the tert-butyl group of MTBE the direct fluorination iso-propylacetate with its iso-propyl group was tried in the direct fluorination. It was expected to be more stable compared to MTBE, because of its reduced ability to stabilise radical intermediates. Its boiling point is 88 °C and its viscosity 0.00525 g cm-1s-1, which is a typical value for an organic solvent. The dielectric constant is 4.7, which makes it a low to medium polar substance. The density is 0.8691 g mL-1. Fluorination of n-butyl trifluoroacetate This substance was chosen, because it was on the one hand a possible precursor to perfluorobutanol. On the other hand it was carried out to learn about the fluorination of a nbutyl chain, which was also present in the fluorination of [(n-Bu)4N]+ BF4-. A primary ester should be most stable against radical cleaving and it has a relatively high boiling point of 104 °C, which avoids evaporation. Ethyl acetoacetate Ethyl acetoacetate was chosen as substrate because there were relatively good literature results for the direct fluorination in microreactors and its property to be quite selectively fluorinated at the acidic CO-CH2-COO- position. It can easily tautomerize (Figure 51). At room temperature the enol form is present with 7.6 %. This value is depending on the solvent environment. For 10 % in acetonitrile it is 4.9 % and for the same concentration in hexane 39 %. [133] H O O O O O O Figure 51 Keto and enol form of ethyl acetoacetate. The Chambers group fluorinated several 1,3-ketoesters using a microreactor. For the fluorination of ethyl acetoacetate they obtained as main product the mono fluorination at the 2 position. This was also the only monofluorinated isomer they reported. Additional to this one they found double fluorination at the 2 and 4 position and the 2 position was in part twice fluorinated (Figure 52). 57 O O 3 1 O O O 10% F2 in N2 4 O O 2 HCO2H, 5°C 5 3% 71 % O F O O F O F F 6 O 12 % F Figure 52 The three main products obtained by Chambers et al., 2-fluoro-3-oxo-butyric acid ethyl ester (71 %), 2,4-difluoro-3-oxo-butyric acid ethyl ester (12 %) and 2,2-difluoro-3-oxo-butyric acid ethyl ester (12 %). They found a tendency in which substrates with a high initial enol concentration showed higher monofluorination selectivity. They also report the necessity to do a rapid work up because decarboxylation reactions can occur. Those reactions lead to the formation of fluoroketones.[114] The melting point of ethyl acetoacetate is -44 °C and its boiling point is 180 °C. The dielectric constant is with εr=16.6 in the range of a medium polar substance. Its viscosity is 0.015393 g cm-1s-1. Ethylene carbonate (EC) EC was chosen because of three factors, which made it very interesting for the direct fluorination experiments. The first and second points were practical ones: First, the substitution of one of the four symmetry-equivalent hydrogen atoms in EC led to a single possible product. Double fluorination only leads to three new products. This improves the signal intensity in the analysis, which was necessary especially for experiments using the syringe pump. Second, its structure is very stable against fluorine and aHF concerning fragmentation. Fluorine concentrations of 30 % during direct fluorination without solvent were the standard in classical batch reactions.[15, 134, 135] This allowed higher fluorine flows and shorter reaction times. The third reason was concerning a possible application. Monofluorinated ethylene carbonate (F1-EC) is commercially used as solvent additive in lithium ion batteries and is produced on a few ton a–1 scale, which would be in the desired range for commercial mini-/micro-reactor production. Because pure ethylene carbonate has a melting point of 36 °C, heating or a solvent is necessary. As co-solvent the product F1-EC was used, cf.[134]. Its melting point of 17 °C allowed experiments at room temperature. The physical properties of EC and F1-EC were investigated by the Nanbu and the Kobayashi group. The values they found are not fully matching (Table 10). Especially the tendency of the important dielectric constant from the 58 non fluorinated EC to the F1-EC is different: The Nanbu group reported an increase of this value, the Kobayashi group a decrease. Table 10 Density, viscosity and dielectric constant for EC and F1EC. Values [136] measured by the Nanbu and Kobayashi [135] groups. EC Property F1-EC Nanbu Kobayashi Nanbu Kobayashi ρ [g mL ] 1.36 (40 °C) 1.32 (40 °C) 1.48 (40 °C) 1.5 (23 °C) η [mPa s] 1.6 (40 °C) 1.9 (40 °C) 2.4 (40 °C) 4.1 (40 °C) εr 90 (40 °C) 90 (40 °C) 100 (40 °C) 78.4 (23 °C) -1 Propylene carbonate (PC) By having an additional methyl group compared to ethylene carbonate, the number of possible fluorination can be determined. Propylene carbonate, like ethylene carbonate, is remarkably stable against fluorine or hydrogen fluoride, and is a liquid with a melting point of –55 °C. This permits experiments without solvent and at lower temperatures. To the best of my knowledge only one group published a direct fluorination of propylene carbonate: Nanbu et al. published a batch reaction, but did not give any details on yield or optimization.[137] However, they investigated the influence of the mono fluorination on relative permittivity, density, refractive index and dynamic viscosity. As a conclusion they expect the monofluorinated propylene carbonate to form better conducting salt solutions than non fluorinated propylene carbonate. Table 11 Density, viscosity and dielectric constant for EC and F1EC 20 °C Property PC F1PC ρ [g mL ] 1.21 1.41 η [mPa S] 0.025 0.092 εr 64 95 -1 a [137] a for EC their value of the dielectric constant differed strongly from the results of another group. Because of this, the value might possibly be erroneous. 59 Closo-K2[B12H12] To my knowledge, no ionic substances and no boron clusters have thus far been investigated for direct fluorination in mini-/micro-reactors. However, 2009 Strauss et al. published a batch method for the direct fluorination of closo-K2[B12H12] giving closo-K2[B12F12].[138] The substrate salt K2[B12F12] and the products are soluble in acetonitrile, which is helpful for fluorination in mini/micro structured reactors. Moreover, similar related batch reactions were reported to be delicate against complete degradation and formation of the thermodynamic sinks BF3 gas and [BF4]–.[138] Since the halogenated dodecaborates can be used to stabilize e.g. reactive cations or electrolyte salts in lithium ion batteries,[139] it appeared interesting to investigate the fluorination of such an inorganic substrate in our mini-reactor-system. To date very little results on the direct fluorination of inorganic substances in mini-/microreactors have been published. Tetraalkylammonium salts: After the successful fluorination of the closo-K2[B12H12] first initial experiments with other ionic substances were carried out. For those experiments tetraalkylammonium salts were chosen. To my knowledge there is no example of a direct fluorination of a tetraalkylammoinium compound so far. But there are two examples of electrochemical fluorination by Dimitriov et al.[140] and the Seppelt group.[141] Dimitriov et al. fluorinated [(nBu)4N]+ X- (X=Br, BF4). They report the corresponding perfluroalkylamines as main products in 22 % yield for the BF4– salt and 38 % for the Br- salt. Additionally they mention the formation of hydrogen, CF2 and NF3 during the process. The Seppelt group fluorinated [N(CH3)4]Cl and obtained as main products N(CF3)3 and (CF3)2NCF2CF3, both with a yield of 4 %. Only one fluorinated ammonium salt they reported qualitatively as NMR signal. In the spectra of the HF phase used for the electro fluorination, they found [(CH3)3NCH2F]+. [(n-Bu)4N]+ BF4- has a melting point at 160 °C and starts decomposing at 325 °C. The conductivity is 6.2 mS cm–1 measured in THF at 25 °C.[142] The investigations in this substance class were thought as a first step to the synthesis of perfluorinated ionic liquids. Some direct fluorination of ionic liquids were already carried out by Hirschberg et al.. For example they fluorinated 1-butyl-1-methylpyrrolidinium, which was preferably fluorinated at the terminal position of its alkyl chain.[143] An review article about “Ionic Liquids with Fluorine-Containing Cations” was published by Xue et al.. in 2005.[144] 60 4.2.2 Fluorination of acetonitrile In this work acetonitrile was mainly used as solvent for various direct fluorination reactions. In the first reactions, in which a syringe pump was used for the liquid feed, only very little fluorine could be used in the fluorinations. The average maximal conversion was ≈5 %. These early reactions were direct fluorinations of toluene (see also chapter 4.2.3). Only the normal mono- and difluorinated acetonitriles could be identified in reaction mixture. There is no trace of the perfluorinated acetonitrile observable, which is also explainable by its very low boiling point of -64 °C. An exemplarily selected 19 F-NMR spectrum of a low conversion reaction is shown in Figure 53. Figure 53 19 F-NMR spectrum of a direct fluorination experiment. A toluene:acetonitrile (1:9) mixture was used. -1 -1 -1 Flow F2= 2 mL min , flow N2= 4 mL min , liquid flow = 20 ml h , temperature = 15 °C. The signal at -117.1 ppm is the 19 F-NMR signal of CF2HCN and at -232.4 ppm CH2FCN. The amount of the monofluorinated toluenes in a low conversion reaction, is about 3.5 times larger compared to the amount of fluorinated acetonitriles. This shows acetonitrile to be a suitable solvent regarding to the preference for fluorination. In Figure 54 two 19 F-NMR spectra are shown. The bottom spectrum was obtained from a batch direct fluorination of ethylene carbonate dissolved in acetonitrile (10 g EC, 70 mL acetonitrile). The reaction was later distilled into two fractions: The first fraction was collected at a temperature of 82 °C, which is the boiling point of acetonitrile and monofluorinated acetonitrile. The spectrum also shows this compound to be the main compound in this fraction. Additionally to the mono- and twice fluorinated acetonitrile some other signals can be observed. For better identification of acetonitrile caused signals, the upper spectrum was added. This upper spectrum shows a direct fluorination of pure EC. 61 Figure 54 Top: Direct fluorination of pure ethylene carbonate. Bottom: 10 g EC, 70 mL MeCN, 7 g NaF, batch in -1 -1 PFA flask, T=0 °C, flow F2=5 ml min , flow N2= 20 mL min , distillation. First fraction: 1 bar, 82 °C; second fraction 20 mbar, 113 °C. 4.2.3 Fluorination of toluene oluene For the reactions the minireactor #1 and a syringe pump were used. In Figure 55. the possible monofluorinated nofluorinated isomers are shown. CH3 CH3 MeCN M +F F2 F + -H HF Tol CH2F CH3 CH3 + + F ortho-F1Tol meta-F1Tol F para-F1Tol CH2F-F1Tol Figure 55 Direct fluorination of toluene in acetonitrile. The four possible monofluorinated isomers are shown. The fluorination of toluene in acetonitrile was planned planned as a standard reaction for reactor performance evaluation. Similar to the literature, literature it was possible to fluorinate toluene using acetonitrile as solvent. The mono-fluorinated mono fluorinated molecules were obtained as main products. A typical 19 F spectrum is shown in i Figure 56 and 57. The acetonitrile was already removed in this spectrum. Next ext to the four main products, products a great variety of signals attributable to side products was observed. Many of these signals could only be explained by fragmentation of the parent molecule. Figure 56 19 F NMR spectrum of the reaction mixture after direct fluorination of toluene and washing with water. -1 –1 Conditions: T= 0 °C. 12.5 vol% toluene in acetonitrile. liquid flow= 90 ml h . gas flow= 480 ml h . 25 vol% fluorine. a: meta-F-toluene, b: ortho-F F-toluene, c: para-F-toluene. 62 Figure 57 Magnification of the spectrum in Figure. 56, showing the signals of the three ring monofluorinated toluenes. They are compared to the pure substances. substance There were two different methods used for the removal of the hydrogen fluoride. In the first one sodium fluoride was used in the second one the product solution was diluted with acetonitrile and washed with an aqueous sodium hydroxide solution to neutralize neutr the acid. Further work up by washing with water three times, allowed the removal of most acetonitrile and fluorinated acetonitrile components. In Figure 58 a 19F-NMR spectrum after the washing procedure is shown. The signal of the monofluorinated acetonitile acetonitile was reduced to traces. The signal of the twice fluorinated acetonitrile was fully removed. Figure 58 19 F-NMR NMR spectrum of a direct fluorination of toluene with acetonitrile as solvent, after quenching and washing with water. In Table 12 results of some direct fluorination reactions are listed. The fluorine to substrate ratio was around 5 % for all shown reactions, because only experiments with a syringe pump for the liquid feed were carried out. Because of this the ratios of formation for the four isomers could be measured with only small influence of the formation of twice fluorinated products. The results are compared to two literature values, published by the Jähnisch et al. [23] and De Mas et al. [24]. 63 Table 12: Results of the direct fluorination of toluene. The reaction parameters and the relative appearances for the four monofluorinated toluene isomers are shown. They were standardized on the amount of m-FToluene. Literature results: a: Jähnisch et al. No. Flow F2/N2 -1 Liquid flow [23] ; b: De Mas et al. Toluene in [mL min ] [mL h ] MeCN [%] -1 [24] T [°C] m-F- o-F- p-F- CH2F- Toluene Toluene Toluene Toluene 1 a 2/4 20 10 15 1.0 2.9 2.0 0.6 2 a 2.7/2.7 15 10 15 1.0 3.2 2.2 0.9 a 2/4 20 100 no cooling 1.0 3.4 1.7 2.1 b 2/4 20 100 -10 1.0 2.8 1.3 0.7 c 6/2 90 12.5 0 1.0 3.9 2.3 0.9 1 5 3 1.0 3.3 1.9 3 a 4 b 10 a 3 : Work up with NaF b 4 : Work up by washing with NaOH solution and water The results show, as expectable, the ortho-isomer to be the main product. Corrected by the number possible hydrogens for substitution the values of the para-isomer in average are the same as the values of the ortho-isomer. This indicates an only weak differentiation in the direct fluorination caused by electronic or steric reasons between the two isomers. Clearly less favoured are the meta-positions, which is typically in the electrophile aromatic substitution of toluene. Even with the direct fluorination, this position is fluorinated about three times less than the ortho- and para-positions. The least favoured reaction site is the methyl group. Though there are three hydrogen atoms it is less often fluorinated than the meta-positions. There is one outlier for this value in a reaction without cooling. The increased temperature certainly can increase the fluorination at the side chain, but a more than doubled increase was not expectable. However, it is in agreement with a more radical like mechanism. Additionally to the impurities visible in the NMR spectrum, the reactor reaction channels were blocked after an average of five hours of use. A mass spectrometer analysis (EI. 200 °C. 70 eV. 500 µA) of the dark organic material collected inside the reactor showed masses up to 532 m/z (Figure x). 64 D:\icis \kraca75b Source:200',70eV,500uA ,m /z 41-800 04/27/10 04:28:04 PM MH-Tol.11 EI-direkt s 120, kraca75b #1 RT: 3.61 AV: 1 NL: 9.62E5 T: + c EI Full m s [ 41.00-799.99] 91.0 950000 900000 850000 181.1 800000 750000 700000 650000 109.0 272.2 600000 199.1 550000 Intensity 165.1 500000 450000 290.2 127.0 141.0 400000 257.2 350000 161.1 201.1 330.2 300000 308.2 348.2 250000 203.1 239.1 366.1 312.2 65.0 200000 374.2 150000 384.2 386.2 59.1 406.2 100000 424.2 442.2 50000 460.1 480.2 498.2 532.1 0 50 100 150 200 250 300 m /z 350 400 450 500 Figure 59 ESI-MS spectrum of dark organic material occurred during the direct fluorination of toluene. Table 13 Possible structures of fragments in an EI-mass-spcetrum after a direct fluorination of toluene. -1 Mass [m·z ] Formula 59.1 -1 Possible structure Mass [m·z ] Formula Possible structure ? 165.1 C7H4F4 F4-Toluene C14H13 65.04 C5H5 + Cyclopentadiene 181.10 77.04 C6H5 + Benzene 199.09 91.05 C7H7 + Toluene 109.04 C7H6F + 127.04 C7H5F2 + 127.05 C10H7 141.07 C11H9 161.1 ? + Toluene dimer C14H12F Monofluorinated toluene + dimer 201.1 ? ? F1-Toluene 203.09 C16H11 F2-Toluene 239.1 ? + Naphthalene 257.13 C20H17 + Dimethyliated benzene-trimer + Methylnaphthalene 271.15 C21H19 + Toluene trimer* ? 289.37 C21H18F + Phenylnaphthalene ? + Monofluorinated toluene trimer * * Mass is one below theoretical value. Very evident were masses, which were a multiple of the mass of toluene minus two. This is an indication for the formation of less soluble or even solid toluene oligomers. Especially the formation of such less soluble materials, which led to blockages inside the reactor, makes the 65 fluorination of toluene not suitable as a standard reaction. The occurrence of heavier substances when fluorinating toluene was also reported by Jensen / Jähnisch [23, 24] . In Table 13 possible fragments are shown. 4.2.4 Fluorination of methyl tert-butyl ether: F2 + N2 O + MeCN F B A H3C OF C This experiment did not lead to fluorinated MTBE. Even with low fluorine concentrations of 16 % the main product found was tert-butyl-fluoride (Figure 60. p-fluoro-toluene as internal standard). It was produced with a selectivity of around 75 %. Only traces of mono fluorinated methyl groups were observed. This method might be an alternative route towards tert-butyl fluoride: its selectivity is similar to good reactions in the literature. e.g. Barthazy et al. obtained 84 % yield using a catalyst and thaliumfluoride on tbutyl-iodide [145] and Koroniak et al. obtained 80 % yield using piperidine and (2H)-pentafluoropropene on tert-butanol [146]. CH3 H 3C F CH3 Figure 60 19 –1 F NMR spectrum of the products of direct fluorination of MTBE. Conditions: flow = 40 ml h . F2: -1 –1 3 mL min . N2: 3 mL min . T= –20 °C. 19 F = –131.8 ppm: decet of tert-butyl-fluoride. 4.2.5 Fluorination of iso-propylacetate: The experiments for the fluorination of iso-propylacetate were carried out as part of the bachelor thesis of P. Bendix [122] All experiments with this substrate were carried out in the capillary reactor using a syringe pump. The influences of the substrate concentration, of the temperature and of the fluorine concentration were investigated. There are four main products found in the direct fluorination experiments. These compounds were 1-methyl-1-fluoroethylacetate, 1-fluoromethyl-2fluoroethylacetate, 1-methylethylfluoroacetate substances are shown in Figure 61. 66 and 1-fluoromethylethylacetate. This B C O O 3 O 4 5 O O O F2 + N2 2 1 (MeCN) F O A F O O O D F F F E Figure 61 Direct fluorination of iso-propylacetate. Observed substances: A: 1-methylethylacetate, B: 1fluoromethylethylacetate, C:1-methylethylfluoroacetate, D: 1-fluoromethyl-2-fluoroethylacetate, E: 1-Methyl-1fluoroethylacetate. Definitions for yield and conversion are shown in Equation 18 and 19. The yield is defined as the amount of the mono-fluorinated isomers (nF1MeEtAc) divided by the start amount of 1methylethylacetate (nMeEtAc). The conversion is defined as the total amount of fluorinated isomers divided by the start amount of 1-methylethylacetate (nMeEtAc). W XYZ [\]^_` ! ∑ XYb [\]^_` X[\]^_` X[\]^_` 100 (18) 100 (19) Influence of temperature and fluorine concentration To find a possible dependence of yield and conversion with the cooling temperature and the fluorine concentration a series of experiments were carried out. For the temperatures values -11, 0 and 26 °C and for the fluorine concentrations 10, 20 and 50 % fluorine were chosen. By using the syringe pump the grade of fluorination was only controlled by the fluorine concentration for these experiments. The results show (Table 14) the decrease of temperature to be favourable (Figure 62, left graph). The yield can be increased by approx. 25 % for 10 and 20 % fluorine concentration and even by almost 70 % when using 50 % fluorine. As mentioned above is the conversion and yield depending on the fluorine concentration because of the constant liquid flow. When the yield is divided by the fluorine concentration, adjusted values for the influence of the fluorine concentration are obtained (Figure 62, right graph). These values show an increase of yield when higher fluorine concentrations are used. This trend is unexpected, but can be observed for all three temperatures. Most significant is the effect for the reaction at -11 °C. 67 Table 14 Yield and conversion for different fluorine concentrations and temperatures 10 % F2/N2 20 % F2/N2 50 % F2/N2 Temp. [°C] Y C Y C Y C 26 0.6 2.2 1.5 5.0 4.1 5.4 0 0.7 2.8 1.5 4.6 4.2 5.5 -11 0.74 3.2 1.9 2.4 6.9 9.1 Figure 62 Left: Graph showing the yield in dependence of the temperature for different fluorine concentrations. Right: Graph showing the yield per 10 % fluorine in dependence of the fluorine concentration. Influence of fluorine concentration and substrate concentration on isomer distribution To investigate these dependences, reactions were carried out in which pure and diluted MeEtAc was fluorinated with fluorine concentrations of 10, 20 and 50 %. The reaction using 50 % fluorine and pure substrate was repeated several times, but always led to a clogging of the reactor. The composition of the three mono-fluorinated isomers showed a strong dependence on the fluorine concentration and on the substrate concentration (Table 15). The formation of 1fluoromethylethylacetate was favored for the harsher conditions using undiluted isopropylacetate and 50 % fluorine. When using milder conditions, first the amount of 1methylethylfluoroacetate increases, then, for the mildest conditions, 1-methyl-1- fluoroethylacetate was the main product. Table 15 Dependence of fluorine and substrate concentrations on the monofluorinated isomer composition. VMeEtAc / 1-fluoromethylethylacetate 1-methyl-1-fluoroethylacetate 1-methylethylfluoroacetate VMeCN 10 % F2 20 % F2 50 % F2 10 % F2 20 % F2 50 % F2 10 % F2 20 % F2 50 % F2 9 14 77 100 81 0 0 44.4 22.9 0 100 72 82 - 12.5 1.6 - 14.7 15.9 - [%] 68 Comparison of tube reactor and a batch reaction The tube reactor was compared with a classic approach for the direct fluorination of iPrOAc in acetonitrile. The fluorination was carried at with fluorine concentrations of 10, 20 and 50 % at 0 °C. The results are shown in Table 16. Comparing the results the advantage of a micro sized reactor becomes obvious. At a fluorine concentration of 10 %, the tube reactor shows 43 % higher yield of 1-fluoromethylethylacetate. This value strongly increases if higher fluorine concentrations are used. For 20 % the yield is already 69 % larger and at 50 % it is even a yield increase of 211 %. Table 16 Results of the direct fluorination of iPrOAc (33 vol%) in acetonitrile.(66 vol%) at 0 °C. Y= Ytube/Ybatch Concentration F2 [%] Y 10 1.43 20 1.69 50 2.11 4.2.6 Fluorination of n-butyl trifluoroacetate In this experiment the realisation of perfluorination was examined. 1 g n-Butyl trifluoroacetate was directly fluorinated using the minireactor #2 and the peristaltic pump cyclisation system. As solvent 2 g PC was used. To achieve perfluorination the total fluorination time was 30.5 h with a slowly increasing fluorine concentration from 42 % (4 mL min-1 N2/3 mL min-1 F2) to 75 % (2 mL min-1 N2/6 mL min-1 F2). Finally six time more fluorine was used as needed for the perfluorination of the substrate. During the reaction the liquid amount decreased, it was necessary to add 1 g PC after 25 h. For the quenching of the hydrogen fluoride silica gel was used. After the reaction the NMR spectra showed still a large amount of signals (Figure 63). Because of too strong overlapping of the signals the 1H-NMR spectrum could not used for the analysis of different species. The 19 F-NMR spectrum showed only partial overlapping of signals, but was not sufficient for the identification of a single molecule. Nevertheless some larger fluorine coupling systems were found in the sample. The largest were two 5 fluorineand one 4 fluorine-atom systems. Unfortunately the signal ranges of fluorinated n-butyl trifluoroacetate and propylene carbonate were very similar in the 19 F and 13 C spectra it was not possible to securely determine which of them belong to the fluorinated substance. 69 Only a small signal of the initial trifluoromethyl group was observable after the reaction. This indicates a loss of substrate or a hydrolysis of the group during the reaction or the work up. Figure 63 19 F-NMR spectrum of the direct fluorination of n-butyl trifluoroacetate. The two main signal areas are enlarged. For future experiments the use of ethylene carbonate as solvent is preferable, if possible. Though the temperature range is limited, the significant reduced amount of signals and the reduced of fluorinationable sites limited to four. 4.2.7 Fluorination of ethyl acetoacetate For the fluorinations of ethyl acetoacetate the minireactor #1 and a syringe pump were used. Four experiments were carried out with a 33 % F2/N2 mixture at 25 °C. The liquid flow was set to 4 mL min-1 which led to a maximum conversion of 1.7 %. The ethyl acetoacetate was used without any solvent. The 19F-NMR spectra showed several signals which were, except of the main signal, not identifiable with certainty because of the low intensity (Figure 64). The main signal at -194.5 ppm was identified as the ethyl 2-fluoroacetoacetate. In the 19 F-NMR spectrum, its integral was 41 ± 3 % of the total integral over all signals. Comparing this value to literature results, it is low. The Chambers group for example published a yield of 69 % in a microreactor, at 5-10 °C, with a 10 % fluorine mixture and dilution of the ethyl acetoacetate by formic acid.[147] Because of the low maximum conversion, a repeated fluorination of the same solution was tried. This led to an improvement of the signal intensity, but also to a large increase of signals. The amount of possible mono fluorinated isomers is four, the amount for twice fluorinated is 70 10 and those 14 isomers can cause 20 different NMR signals in the 19 F spectrum. For intensive studies with 2D NMR studies the concentrations of the minor products still was too small. Because of this, the fluorination of ethyl acetoacetate was not further investigated. Figure 64 19 F-NMR spectrum a repeated direct fluorination of ethyl acetoacetate using the minireactor #1 and a syringe pump. The main signal areas are shown enlarged. The two large signals are p-fluorotoluene (standard, 119 pmm) and ethyl 2-fluoroacetoacetate (-195 ppm). 4.2.8 Fluorination of ethylene carbonate: The aim at fluorinating ethylene carbonate was to achieve a high yield of the commercially interesting F1EC. Therefore, a high conversion and a limited generation of difluorinated ethylene carbonates were desirable. For all reactions the minireactor #2 and the peristaltic pump system were used. As solvents acetonitrile or F1EC were used. Fluorination with acetonitrile as solvent The first experiments of the direct fluorination of ethylene carbonate were carried out using acetonitrile as solvent. In Figure 65, an exemplary 19 F-spectrum is shown. In this spectrum, 71 % of the fluorinated material found was F1EC. The main component of the rest was 7 % mono fluorinated acetonitrile. On the one hand, this shows acetonitrile being a suitable solvent for this reaction, but the acetonitrile also led regularly to jammed reactor channels. 71 Figure 65 19 -NMR spectrum of a direct fluorination of EC dissolved in acetonitrile. 30 % EC in acetonitrile, 33 % F2, nF2/nEC= 0.16, T= 8°C, minireactor #2 Fluorination with F1EC as solvent To reach the desired high conversion, the fluorine should react completely with ethylene carbonate without causing fragmentation. Since ethylene carbonate is a solid at room temperature, liquid mono fluorinated ethylene carbonate was added as a solvent.[134] Adding the product as the solvent has the advantage of reducing the number of possible active reagents and allows for a straightforward separation of the products. However, the selectivity with respect to doubly fluorinated products (F2EC) is reduced, yet all F1EC-concentrations below 25 wt% led to ethylene carbonate-crystallization at room temperature. In our experimental setup, the PFA tubing could not be heated, and thus we had to tolerate this relatively high content of F1EC and the connected losses in selectivity. It can certainly be improved, by using a heated tubing system. In all fluorination experiments, ethylene carbonate was shown to be very robust with respect to decomposition. Other than the expected difluorinated isomers, only a very small amount of side products, below < 2 mol% with respect to the mono fluorinated ethylene carbonate were observed. In Figure 66 two 19F spectra of direct fluorinations using 45 and 88 % fluorine are shown. Even with such harsh reaction conditions, good results were obtained. 72 4,4-F2EC O O O cis-F2EC trans-F2EC F1EC O O O O O O O O O F -80 19 F -120 -100 F F BF4 Standard ab Figure 66. Two F F F -140 ppm F-NMR traces of EC after the fluorination reaction. Conditions: 30 wt% F1EC as solvent. a: 0.5 equiv. F2, T = 22 °C, liquid flow = 2.5 mL min –1 –1 gas flow = 6.6 mL min , 45 vol% F2. b: 0.5 equiv. F2, T = 22 °C, –1 –1 – liquid flow = 2.5 mL min , gas flow = 3.4 mL min , 88 vol% F2, minireactor. The [BF4] signal is formed from residual HF. Conversion, yield and selectivtiy referring to the fluorinated ethylene carbonate In this section of the discussion, fluorine is considered to be the only reagent for the fluorination of EC. This view focuses on the production of fluorinated material. The conversion, C, the yield, Y, and the selectivity, S, with respect to the formation of monofluorinated ethylene carbonate (F1EC) are described by equations (20) to (y). ! XcdF Xcd] XcdF 100 (20) W XYZ cd] XYZ cdF XcdF 100 (21) e XYZ cd] XYZ cdF XcdF Xcd] 100 (22) e "f g hij g Wi kl XYZ cd] XYZ cdF m]no`p[qr `sorrnt u W FD X)<;?9Y3 cd 6X=&9Y3 cd 6wG,GY3 cd ! FD XYZ cd] 6w)<;?9Y3 cd 6X=&9Y3 cd 6XG,GY3 cd XY3 XY3 100 100 (23) (24) 100 (25) In the Equations (20) to (23) nEC0 is the molar amount of ethylene carbonate before the reaction, nECr the amount after the reaction. nF1EC0 is the amount of F1EC added as solvent and nF1ECr is the amount of F1EC found in the product mixture. In Table 17, the results of our direct fluorination experiments (Entries 1-5) and those from the literature (Entries 6-8) are collected. Entries 1 and 2 show two results at different conversion levels. The selectivity for F1EC was, as expected, reduced at higher conversion and due to the formation of doubly fluorinated species (see also Figure 67). The yield increased up to a 73 maximum of 49 %, which is lower than already known methods, but this low value is also caused by the use of F1EC as a solvent. The conversion with respect to the use of fluorine is very good, and values around 74 to 80 % were achieved. Entries 3 and 4 compare experiments at 22 and 40 °C: In this case the conversion kept constant, and a rise of the temperature seemed to be advantageous. Entry number 5 shows an experiment using almost undiluted F2. The results of this experiment were slightly inferior to those using diluted fluorine. This demonstrates the robustness of EC as well as the reactor system. The results of three more classical approaches by Woo,[15] Böse [134] and Kobayashi et al.,[135] are also included with Table 17 as entries 6-8. All three are batch reactions, in which diluted fluorine was bubbled through pure EC or EC diluted with F1EC.[15, 134, 135] Our results on yields regarding the fluorine use are comparable with those of Woo and Böse, both of which were patented as optimized processes. Entry 6 in Table 17 of Kobayashi et al. was a relatively simple lab approach using a PFA vessel. Here our method is clearly superior in the fluorine use. The conversion of valuable fluorine with a fluorine use of up to > 80%, while using a significantly higher fluorine concentration at the same time is certainly the main advantage of using the [%] mini-structured reactor. 70 Y 45% F2 22°C 50 S 45% F2 22°C Y 97% F2 30 45 65 Conversion [%] 85 S 97% F2 Figure 67 Yield and selectivity for the direct fluorination of ethylene carbonate. For an overall evaluation, the space-time yield was calculated. The values of our system included with Table 17are typical for microreactors (e.g. compared to Jähnish et. al. [23]), and range from 6300 to 10500 mol m-3 h-1, which is two orders magnitude higher than the laboratory batch approach of Kobayashi et al.[135] However, this value was only calculated for the active reaction volume. It certainly would be reduced, when the walls of the reaction volume are taken into account. Nevertheless, for a better comparability this value was chosen. The relative ratios of the doubly fluorinated ethylene carbonate isomers were very constant through all experiments. However, they differ strongly from the results Kobayashi et al.[135] obtained when fluorinating pure F1EC. This indicates a reduced kinetic influence, which could be caused either by an increased temperature (unlikely) or by a faster mixing of the biphasic gas-liquid reaction (likely). 74 Table 17 Results of the direct fluorination of ethylene carbonate. The Entries 6-8 are literature results. 6: laboratory batch experiment of Kobayashi et al. batch reaction by Solvay Fluor No Time [h] 1 Flow F2 -1 [134] ; 8: patented batch reaction by Ulsan Glas Flow N2 -1 [135] ; 7: patented [15] T [°C] C [%] Y [%] S [%] Space-Time-Y. - -1 [mol m ³ h ] a Y F2 [%] C F2 [%] Y F2EC [%] Relative ratios (trans/cis/4,4-F1EC) trans/cis/4,4-F1EC [mL min ] [mL min ] 4.0 3.0 3.6 22 80 49 59 6348 52 74 20.9 (11.4/6.7/2.8) 4.0/2.4/1.0 2 1.8 3.0 3.6 22 45 32 70 10585 76 83 6.1 (3.1/2.1/0.9) 3.5/2.4/1.0 3 3.3 3.0 3.6 40 68 46 68 8174 59 77 15.5 (8.6/4.8/2.1) 4.0/2.3/1.0 4 3.1 3.0 3.6 22 69 39 58 8173 57 79 14.2 (7.5/4.6/1.9) 3.8/2.4/1.0 5 2.0 3.0 0.1 22 47 31 66 9706 80 82 5.1 (2.5/1.9/0.7) 3.5/2.6/1.0 6 41 105 245 50 - 70 - 102 39 - 7 - - - 35 - 64 - - 69 - - - 8 5.3 6533 26133 55 - 76 93 - 57 61 - - a) calculated by using the volume of the reaction containing segment (tube volume, flask volume) b) separate reaction, started from pure F1EC 75 75 (59/11/5) b 11.8/2.2/1.0 b Trice Fluorinated Ethylene carbonate In the previous experiments, the focus was laid on the mono fluorinated ethylene carbonate. Because of this it was always the main product, and the twice fluorinated were found as side products. The trice fluorinated isomer was only found in traces. In an experiment with 1.3 equivalents fluorine, it was found in a yield of 1.1 %. This low concentration allowed no further investigations, but even the 19 F-NMR data have not been published so far. The 19 F- NMR spectrum is showed in Figure 68. Figure 68 19 F NMR-spectrum magnifying the three signals caused by the F3EC. Conversion, Yield And Selectivtiy Referring to the Fluorine Use In this section of the discussion, the focus is laid on the effectiveness of the reactor. The fluorine conversion is focused on. FD EC 5-- ∑ di|luorinated ethyl carbonate isomers ŒF5 EC B ∑ ŒFD EC ! X‹3 W X‹3 e X‹Z •Ž6∑ X‹3 •Ž 5-- (26) ŒF5 EC 5-- (25) (27) nF5 EC (28) In Table 18, the results of our direct fluorination experiments and those in the literature are shown. Entries 5 and 6 in the Table include the outcome of the fluorinations shown in Figure 4. Upon decreasing the fluorine content to substoichiometric levels, the relative yield, Y, and selectivity, S, increased. Most noticeable is the high conversion in the range of 83 to 76 95 %. This implies that almost every fluorine molecule introduced into the system was used to fluorinate the ethylene carbonate. This is a remarkable finding, as competing batch approaches usually are required to introduce a large excess of expensive fluorine.[135] The results of three more classical approaches by Woo,[15] Böse [134] and Kobayashi et al.,[135] are also included with Table 18 as entries a, b, and c. All three are batch reactions, in which diluted fluorine was bubbled through pure EC or EC diluted with F1EC.[15, 134, 135] Our results on yields are comparable with those of Woo and Böse, both of which were patented as optimized processes. Entry b in Table x of Kobayashi et al. was a relatively simple lab approach using a PFA vessel. Here our method is clearly superior. The conversion of valuable fluorine with a fluorine use of up to 95%, while using a significantly higher fluorine concentration at the same time is certainly the main advantage of using the mini-structured reactor. Table 18 Results of EC direct fluorination experiments using the minireactor.* n(F2)/n(EC) F2 conc. T Y S C [equiv.] [%] [°C] [%] [%] [%] 1 1.0 45 30 59 71 84 2 1.0 45 22 59 71 83 3 1.0 30 22 61 74 83 4 0.75 45 22 68 79 86 5 0.5 45 22 77 80 95 6 0.5 88 22 76 84 91 7 0.3 45 22 78 83 95 a 1.2 30 45-50 57 93 59 b 1.8 30 50 38 - - c 0.89 5 30-35 63 - - Nr. -1 * All experiments were carried out with a mixture of 30 %wt of F1EC in EC. The liquid flow = 2.5 mL min for all experiments except no. 7. In this experiment a liquid flow of 0.5 mL min –1 was used. The F1EC added as solvent was subtracted before any calculation. Entries a, b and, c are literature values of a: Woo et al. Kobayashi et al. [135] , c: Böse et al. [134] [15] , b: , respectively. See Equations 25-28 for Y(-ield), S(-electivity) and C(- onversion) 77 Quantumchemical calculations The following quantum chemical calculations were carried out by Mr. Sascha Goll within the group of Prof. I. Krossing. For a better understanding of the reaction, the underlying thermochemistry of the reaction was calculated using the reliable Gaussian 3 (G3) compound method as level of theory.[148, 149] The results of these calculations are shown in Figure 69. There is no obvious trend of the ∆rG° values going from the monofluorination to the perfluorination of ethylene carbonates. They are between -468 and -508 kJ mol-1. Especially high values were calculated for the monofluorination and when a fluorine atom was added to a carbon already bearing a fluorine atom. If comparing the difluorinated isomers, the 4,4-difluoro-EC is the energetically most preferred isomer, which contrasts to the experimental results. This shows that the reaction is still kinetically controlled. However, as expected for the efficient heat dissipation in our minireactor system, our results are already closer to the thermodynamic product than those in the batch experiment by Kobayashi et al., i.e. the amount of 4,4-difluoro-EC is much higher in our experiments than that in Ref. [135] (cf. Table 17). EC F1EC -496.0 cis-F2EC 37.6 trans-F2EC 4,4-F2EC 22.8 F4EC -503.3 0.0 F3EC -469,9 -488.7 -1 Figure 69. ∆rG° [kJ mol ] values for the direct fluorination of ethylene carbonate (CO3HnFm+F2 → CO3Hn-1Fn+1+HF). For the cis-F2EC and the trans-F2EC the relative energies compared to the global minimum 4,4F2EC. Calculations were done with the Gaussian 3 compound method. 78 Table 19: Calculations by Sascha Goll: Calculated total energies [(RI-)BP86/def-SV(P) level], enthalpies (G3 level) and Gibbs energies (G3 level) of all investigated particles. The italic values at the top are used for the energetics of the reactions at the bottom BP86/SV(P) G3-H G3-G [Hartree] [Hartree] [Hartree] F2 -199.36336 -199.422856 -199.445729 HF -100.344859 -100.397801 -100.417488 Molecule symmetry EC C2 -342.167175 -342.187532 -342.221335 F-1-EC C1 -441.343816 -441.402213 -441.438479 relative Gibbs energies [kJ/mol] 4,4-F-2-EC Cs -540.52543 -540.620217 -540.659176 cis-F-2-EC Cs -540.511157 -540.606719 -540.643145 cis-F-2-EC C1 -540.511139 -540.605928 -540.644843 37.6 trans-F-2-EC C2 -540.516766 -540.612971 -540.65048 22.8 F-3-EC C1 -639.694569 -639.826466 -639.866384 F-4-EC C2v -738.871784 -739.038662 -739.080778 BP86/SV(P) G3-H G3-G [in kJ/mol] [in kJ/mol] [in kJ/mol] -415.2 -497.9 -496.0 EC + F2 -> 1-F-EC + HF 0.0 Temperature measurement For some of the ethylene carbonate direct fluorination experiments, the reactor temperatures were measured with the embedded temperature sensors. These experiments demonstrated the excellent temperature control enabled by the active cooling. No difference in temperature between the five sensors could be detected with the cooling being active. Experiments performed without active cooling and the reactor being placed in a Styrofoam box, yielded nearly adiabatic conditions and the reproducible measurement of an independent temperature at the five sensor positions (Figure 70). 79 Figure 70 Minireactor placed in a Styrofoam box for temperature measurements. The box was covered with a Styrofoam lid before the experiments were started. Even though the sensors were placed only 1 mm vertically above the reaction channel, the thermal conductivity of the reactor copper block blurred the differences within 0.25 °C. Still, according to the temperature profile shown in Figure x, the reaction mainly takes place within the first four centimeters of the reactor. The temperature increase measured for the reactor block, liquid and air within the box during the reaction was equal to an uptake of 105 % of the energy expected for full fluorine conversion. For the calculations, the Equations 29-32 and assumptions were used. Assumptions: - A linear increase of the reactor temperature, because of the low temperature difference of 2.68 °C. - The reactor body was assumed to be 100 % copper, neglecting the nickel coating and screws. ••>,‘<’ •“,;9*<,” 100 — Œœ3 ˜™šL› —Ÿ™2 ¡šL —¢£¤ £ ••>,‘<’ ••,;=)‘< 6•–&:*&” (29) ∆H•Žž‹5•Ž f1šNN™L jŸ™2 ¥¦ 100 X¥¦ ∆ D B (30) ¡šL “,¦§ ∆h (31) X“,¦§ ∆ (32) D c is the specific heat capacity of copper and ΔT the temperature difference of before and after 55.3 minutes of reaction. The energy-release measured is thus within the uncertainty of the calculation and the experiment. This indicates an almost complete reaction. Therefore, the system could be used 80 in an optimized manner to experimentally determine the reaction enthalpies of direct fluorinations. 25 26 1 cm T [°C] T [°C] 24.9 24.8 24.7 2 cm 25 4.1 cm 29.9 cm 24 53.1 cm 24.6 Styrofoam box 0 20 40 60 23 09:36:00 reaction channel position [cm] 10:04:48 10:33:36 11:02:24 Time Figure 71. Top: Schematic top view e of the reaction and cooling channel layers. Bottom: Temperature profile of –1 the direct fluorination of EC in acetonitrile without active cooling, liquid flow = 0.67 mL min , gas flow = 6 mL -1 min , 33 vol% F2, minireactor. 4.2.9 Fluorination of propylene carbonate: The direct fluorination of propylene carbonate (PC) allowed the use of a broader range of temperatures due to the much lower melting point compared to ethylene carbonate, without the need of a solvent. No direct fluorination of PC was published in literature so far. For propylene carbonate the number of isomers is larger than for ethylene carbonate. There are four monofluorinated isomers (F1PC, Figure 72) and a large amount of polyfluorinated compounds. To reduce the amount of polyfluorinated products, only 0.44 equivalents of fluorine were used in most experiments. O CH 2F O F O O O cisCHF CF O O OO transCHF O F F O O F Figure 72. The four possible monofluorinated propylene carbonates. CFH2: 4-fluoromethyl-1,3-dioxolan-2-one, CF: 4-fluoro-4-methyl-1,3-dioxolan-2-one, transCHF: 4-trans-fluoro-5-methyl-1,3-dioxolan-2-one, cisCHF: 4-cisfluoro-5-methyl-1,3-dioxolan-2-one. Eight experiments were investigated with a simple 23 factorial design DOE approach, using temperatures of –10 and +40 °C and nitrogen and fluorine flows of 2 and 4 mL min–1. The maximal possible conversion was 40 %.The results shown in Tables 20-21, the formulas for C(-onversion), Y(-ield) and S(-electivity) are given in the Equations 33-35. nPC0 is the initial PC amount, nPCr the amount after the reaction. 81 X\dF X\d] ! X\dF x100 (33) Xd©3 Y YZ \d 6Xd©Y YZ \d 6X=&9d©Y YZ \d 6X)<;?9d©Y YZ \d W X\dF XYZ \d e X\dF X\d] x100 (34) x100 (35) Table 20 Results of an eight experiments matrix of the direct fluorination of propylene carbonate. The N2 flow and the F2 flow were set to 2 and 4 mL min –1 and the temperature to –10 and 40 °C. 3.75 g (36.7 mmol) PC was used. 0.44 equivalents of F2 were added in each experiment. The reaction time was 180 min for a F2 flow of 2.0 and 90 min for a flow 4.0 mL min Nr. a –1 -1 -1 Temp [°C] Flow F2 [mL min ] Flow N2 [mL min ] C [%] Y [%] S [%] CH2F:cisCHF: transCHF:CH [%] 1 -10 4.0 2.0 32 23 72 32:21:19:27 2 40 2.0 2.0 26 19 72 31:22:20:27 3 -10 2.0 4.0 33 26 77 32:21:19:27 4 40 4.0 4.0 26 21 80 31:21:20:27 5 40 4.0 2.0 26 20 79 31:21:21:27 6 -10 2.0 2.0 30 22 73 32:21:19:28 7 -10 4.0 4.0 29 22 75 32:21:19:28 8 40 2.0 4.0 29 20 68 31:22:20:28 a -10 4.0 0.0 24 17 71 32:21:19:27 a Relative abundance, normalized by the number of hydrogen atoms at the position of propylene carbonate and to a total sum of 100 %. Table 21 Influence of the temperature and the gas flows of F2 and N2 on conversion yield and selectivity based on the experiments shown in table 4. - = low value, + = high value Temperature Flow F2 -10 / +40 °C 2 / 4 mL min 2 / 4 mL min - 31 29 29 + 27 29 29 - 23 21 21 + 20 22 22 - 74 73 74 + 75 76 75 +/- Flow N2 -1 -1 C [%] Y [%] S [%] 82 Unexpectedly, the conversion and the yield did not show any significant changes when the fluorine concentration was varied between 33 and 66 % and the total gas flow between 4 and 8 mL min–1. This stability of the system with respect to the fluorine concentration was unexpected. The largest influence had the change in temperature. When the cooling temperature was dropped from +40 to -10 the conversion raised from 27 to 31 % and the yield from 20 to 23 %. The temperature had no influence on the selectivity. Flow F2 Temperature Flow N2 29.4 29.4 28.5 28.5 Y [%] C [%] Flow F2 76.3 23.1 31.1 26.8 26.5 - Temperature Flow N2 Flow F2 + 22.0 21.6 21.4 22.0 21.0 19.9 19.5 - + S [%] Temperature Flow N2 31.5 75.0 74.5 74.6 74.2 73.8 72.0 72.5 - + Figure 73 Graphs showing graphically the influence of temperature, F2 flow and N2 flow on conversion, yield and selectivity on the direct fluorination of PC. To see how far this can be pushed, a reaction using 100 % fluorine was performed (Entry a, Table 20). In this case the lowest yield of all experiments was obtained. This indicates an at least a small relation between the fluorine concentration and the yield. When the reaction was run with 100 % fluorine and the nitrogen gas flow was switched of still some small gas bubbles were observed in the product stream. Figure 74 two pictures of the exiting gas liquid flow. Left: nitrogen diluted fluorine, leading to bubbles of nitrogen in the flow. Right: undiluted fluorine was used. There are still small bubbles within the liquid flow. This might be a sign for some decomposition of the propylene carbonate. In contrast to the fluorine concentration and the total gas flow, lowering the temperature showed a small effect: The yield and conversion both were slightly improved. 83 Isomer Distribution: In all eight experiments, at a conversion of around 30 % (Table 22), the relative abundance of the four possible isomers did not change at all. The distribution of the fluorinated isomers showed the methyl group to be mainly fluorinated. Next favoured, the tertiary carbon is fluorinated, even though it is just one hydrogen atom and it is sterically shielded by the methyl group. The secondary carbon hydrogen atoms are the least favoured. Here the cis isomer was fluorinated slightly more often. Figure 75 shows the NMR-distributions of the F1PC-isomers in the mixture after quenching the HF. They are plotted against different equivalents of fluorine used in the fluorination reaction. The increase of the F1PC concentration lowers with increasing amount of fluorine used due to the formation of doubly fluorinated propylene carbonates. The amounts of CF-F1PC as well as the two CHF-F1PC isomers were found to reach around 7 % concentration at most. For those isomers a more conventional synthesis should be developed. The concentration for CH2F-F1PC was found to reach around 24 % at best; it seems to be the best accessible isomer by direct fluorination. Table 22 Conversion + yields and selectivities of the F1PC isomers (sum and individual) plotted against equivalents fluorine used. Values after quenching the HF. T= –20 °C, 240 mL h –1 F2, 240 mL h –1 N2, 150 mL h –1 PC n F2/n PC 0.4 1.0 1.5 C 50 72 76 Y F1PC 21 37 36 Y CH2F 12 23 23 Y CHF 5 8 7 Y CF 4 6 6 S F1PC 76 52 49 S CH2F 43 32 30 S CHF 18 11 10 S CF 14 8 9 84 80 76 72 76 70 C 60 50 52 49 [%] 50 Y F1PC Y CH2F 37 40 36 Y CHF Y CF 30 21 S F1PC 20 S CH2F 10 S CHF S CF 0 0.4 0.9 n F2 / n PC 1.4 Figure 75 Conversion + yields and selectivities of the F1PC isomers (sum and individual) plotted against –1 equivalents fluorine used. Values after quenching the HF. T= –20 °C, 240 mL h F2, 240 mL h –1 N2, 150 mL h –1 PC. Quantum chemical calculations using the Gaussian 3 method (Figure 76 + Table 23) revealed the CF-F1PC to be the thermodynamically preferred monofluorinated PC-isomer. This thermodynamic stability of the CF-F1PC isomer can be an explanation, why in experiments this position was preferred over the fluorination of the CH2 position, although it is sterically shielded by the methyl group. The fluorination of the kinetically most available methyl group on the other hand was computed to be thermodynamically the least favoured, less stable by 56 kJ mol-1 if compared to CH-F1PC. Nevertheless it was the major product isomer found experimentally and clearly shows that also the reaction in the minireactor is controlled by kinetics and not by thermodynamics. equatorial-PC axial-PC 0.0 0.1 -503.3 CF-F1PC trans-CHF-F1PC cis-CHF-F1PC CH2F-F1PC 0.0 10.4 12.3 56.4 -1 Figure 76. ∆rG° [kJ mol ] values for the direct monofluorination of propylene carbonate (CO3H6+F2 → CO3H5F+HF). Additionally the energy difference for the axial and equatorial propylene carbonate is shown, as well as the relative energies of the F1PC isomers with respect to the minimum isomer CF-F1PC. Calculations were done using the Gaussian 3 compound method. 85 Table 23 Calculations by Sascha Goll: Calculated total energies [(RI-)BP86/def-SV(P) level], enthalpies (G3 level) and Gibbs energies (G3 level) of all investigated particles.The italic values at the top are used for the energetics of the reactions at the bottom. BP86/SV(P) G3-H G3-G relative Gibbs [Hartree] [Hartree] [Hartree] energies [kJ/mol F2 -199.36336 -199.422856 -199.445729 HF -100.344859 -100.397801 -100.417488 Molecule symmetry axial-PC C1 -381.455014 -381.466034 -381.503923 0.1 equatorial-PC C1 -381.455508 -381.466174 -381.50396 0.0 CH2F-F-1-PC C1 -480.614138 -480.663155 -480.70329 56.4 CF-F-1-PC C1 -480.634839 -480.685394 -480.724772 0.0 cis-CHF-F-1-PC C1 -480.632627 -480.680614 -480.720078 12.3 trans-CHF-F-1-PC C1 -480.632401 -480.681184 -480.720826 10.4 BP86/SV(P) G3-H G3-G [in kJ/mol] [in kJ/mol] [in kJ/mol] -422.3 -509.8 -505.6 PC + F2 -> 1-F-PC + HF Fluorine loss In all reactions, the amount of fluorinated isomers found in the product mixture, was lower than the amount of fluorine used in the reaction. The fluorine use was defined as shown in Equitation 36. Quantitatively it was only possible to measure the monofluorinated PC isomers, which were clearly the main compounds. ªk«R¬iŒ «- X®3 ∑ X® ¯¦´9 Z *100 (36) The analysis of the eight experiments (see above) for the fluorine use shows values between 49 and 58 %. The rest of the signals showed on average about 13 % fluorine use. The dependence between the fluorine use and the temperature and the N2 and F2 flows are shown in Figure 77. As for conversion and yield, the only significant way to change the fluorine use was the reduction of the temperature. The influence of the gas flows was low and not significant. 86 Fluorine use [%] 58.0 57.7 56.0 54.0 52.0 Temperature 53.5 51.3 50.8 48.7 52.6 50.0 48.0 - Flow F2 Flow N2 + Figure 77 Graph showing graphically the influence of temperature, F2 flow and N2 flow on conversion, yield and selectivity on the fluorine use. The reason why the fluorine was not found in the products could have had different reasons. Two possible reasons were the non-reacting of the fluorine with the substrate within the reactor or the decomposition of the substrates into gaseous fragments. Both is possible and was supported by the gas bubbles appearing at a 100 % fluorine flow (see above). Further possible reasons and experiments to investigate their influence are described in the following. Assignment of error bars of the quantification method: After quenching of the reaction the standard was added to the mixture and stirred to reach an even distribution of the standard liquid. Then the NMR sample was taken within a few minutes. Possible errors could have been an uneven distribution of the standard or evaporation during the mixing time. For testing purposes a known amount of propylene carbonate was dissolved in acetonitrile and sodium fluoride was added. Then the normal standard adding procedure was used. This was done twice. The deviations for the analysis of the 1H and 19 F NMR signals were < 0.8 %. Possible evaporation: The gas flow can increase the surface of the liquid and lead to a faster evaporation. This possibility was eliminated by the fluorination with 100 % fluorine: In this case, the gas flow at the gas liquid separator was reduced close to zero. Nevertheless, the fluorine use in the mixture was only 43 %, which was even worse than in the experiments with diluted fluorine. Impure substrate: Most impurities would have led to fluorinated signals within the NMR spectra. But water or methanol would have led to gaseous products. However, there was no sign of impurities and the substrate was freshly dried and distilled. This led to no differences in the results. 87 Reaction with reactor metal: The minireactor had a large metal surface, which possibly could be fluorinated and absorb the fluorine. The fluorination of propylene carbonate was therefore carried out in the tube reactor, which except of the T-mixer consists of already perfluorinated PFA. In these experiments, the average fluorine use was 49 %, which means no change to the minireactor experiments (see also below) Tube reactor To compare the results for the fluorination in the minireactor to another reactor system, fluorinations in the PFA tube reactor were carried out. The tube length was 100 cm and the temperature was set to 0 °C. The results are shown in Table 24 and Figure 78. At a nitrogen flow of 5 mL min-1 a slug flow was formed, but it stopped after approx half of the capillary, because of insufficient pressure. Because of this the nitrogen flow was increased to reach a wavy annular – annular flow, which needed a lower backpressure. Table 24 Results of the fluorination of PC in the PFA tube reactor. Conditions: t= 0°C, tube length= 100 cm, liquid flow= 1.372 mL h -1 Entry F2/PC Conversion of fluorine 1 9.50 3.00 0.46 49.8 2 12.00 3.00 0.46 50.9 3 15.00 3.00 0.45 31.6 4 15.00 4.50 0.43 40.8 CH2F-PC:cisCHF-PC:transCHF-PC:CF-PC Conversion of F2 a -1 N2 [mL min ] F2 [mL min ] -1 50 3 mL min-1 F2 40 4.5 mL min-1 F2 30 9.00 11.00 13.00 15.00 17.00 Flow N2 Figure 78. Conversion of fluorine in dependence of the nitrogen flow. The results show a decrease of fluorine conversion with increasing nitrogen flow. This can be caused by non reacting fluorine, which can pass because of the annular flow, or the evaporation of volatile products. The increase of the fluorine flow and connected to it the liquid flow (because of the constant F2/PC ratio), leads to an increase of fluorine conversion. 88 4.3 Ethylene And Propylene Carbonate As Solvent For The Fluorination Of Ionic Substrates 4.3.1 Comparison of EC and PC Ethylene and propylene carbonate are both potential solvents for the fluorination of ionic substances. Ethylene carbonate seems to be advantageous because of the lower content of exchangeable hydrogen atoms. To verify this assumption an equimolar solution of ethylene carbonate and propylene carbonate was fluorinated using the minireactor #2. It was fluorinated for 2 hours, with a 33 % fluorine nitrogen mixture (nF2/nEC-PC=0.36), at 20 °C and without the use of a further solvent. In the 19F- NMR Spectrum the integrals of all significant and identified signals were integrated. There was 1.55 times more fluorinated propylene carbonate in the mixture than fluorinated ethylene carbonate. If this number is corrected by the number of hydrogen atoms the ratio is reduced to 1.04. This show the two carbonates were fluorinated with the same rate of reaction. In most cases ethylene carbonate is the substance of choice, if it forms a liquid when mixed with the ionic substrate or is heated. If this is not possible a partial or full replacement with propylene carbonate would be necessary. 4.3.2 Fluorination of closo-K2[B12H12] As a test, the salt closo-K2[B12H12] was almost completely dissolved to a formal concentration of 2.5 wt% in a 4:1 EC/PC mixture (a light clouding of the liquid remained). The finely suspended non-dissolved salt went into solution within the first 30 min of the reaction. Figure 79 19 F NMR spectrum of the closo-K2[B12H12] after the reaction. Conditions: 2.5 wt% in a mixture of EC –1 –1 and PC (3.3:1), T = 20 °C, liquid flow 5 mL min , gas flow 6.6 mL min , 30 % F2. 89 The NMR yield of the perfluorinated material was found to be 58 %. Only 12 % of just polyfluorinated material was left (Figure 79). The main side reactions occurring were the formation of [BF4]– and BF3. Strauss et al. published an isolated yield of 74 % [138] using a classical batch approach with acetonitrile as solvent. In their approach, the HF formed was constantly removed by addition of sodium fluoride and a work-up in the middle of the reaction is necessary. The fluorination using the minireactor and our EC/PC solvent combination already shows a promising result though still the yield is lower than in the literature. However, the experiment was carried out only on a 100 mg scale and was not optimized in terms of temperature and gas/liquid flow rates. Preliminary tests indicate that the use of pure PC as a solvent is also possible, but more fluorine would be required. Thus, the cyclic carbonates were shown to be suitable solvents for direct fluorination of closoK2[B12H12]. Acetonitrile has also been tested as possible alternative solvent that also leads to the fluorinated product, but tended to cause an undesirable clogging of the reactor. Similar to the experiments with toluene, it forms solid organic material within the reactor. Comparison to a batch approach in acetonitrile Following results were part of the teachers studies thesis of M. Rühle in Freiburg [126] , the direct fluorination reaction itself was carried out under my supervision. The fluorination was carried out following the synthesis published by the Strauss group.[138] The closo-Na2[B12H12] placed in a glass flask and dissolved in acetonitrile (Figure 80). Sodium fluoride was added to the solution as hydrogen fluoride scavenger. After approximately half of the reaction time the reaction was worked up and continued the following day, as published by the Strauss group. The isolated yield was 44.6 %. This is significantly lower as the literature published yield of 76.1 %. Figure 80 Set up for the direct fluorination of closo-Na2[B12H12]. The yellow liquid on the right side is a sodium iodide solution to indicate free fluorine. 90 4.3.3 Fluorination of tetraalkylammonium salts To investigate the potential of the direct fluorination of ionic substances, for the fluorination of tetraalkylammonium salts some initial experiments were carried out. Reaction of halides with the reactor: The direct fluorination of a tetramethylammonium halide (TMA-halide) was thought to be the first system for testing. For reasons of availability TMA-chloride was initially tested and dissolved in methanol. When this solution was passing the reactor (no fluorine yet), it changed immediately from colourless to a bright yellow colour (Figure 81). This was suspected to be caused by elemental chlorine. To support this theory sodium bromide was added, and the colour changed to brown. This indicated an oxidation of the bromide to elemental bromine. When acetonitrile was added a bright red solution was yielded. This can be caused by a solvent–halide charge transfer complex. The used and fluorinated reactor seemed to have obtained some remaining oxidative potential from the fluorination reactions, and to be able to oxidize the chloride into chlorine. Most probable is the formation of some higher fluorinated metal fluorides. This might be similar to the fluorine releasing agent K2NiF6.[150] Further investigations on this have not been carried out as part of this in this thesis. Figure 81 Left: Tetramethylammonoium chloride in methanol after purging it through the used minireactor. Middle: After addition of sodium bromide to the left solution. Right: after the addition of acetonitrile. Fluorination of tetra-n-butylammonium tetrafluoroborate: Solubility in ethylene carbonate It was tried, if the addition of tetra-n-butylammonium tetrafluoroborate can reduce the melting temperature of ethylene carbonate and form a liquid. When 2.00 g EC were added to 2.02 g [N(Bu)4]BF4 at the interface a liquid was forming. To accelerate the process the mixture was heated to 40 °C, which melted the EC and a solution was formed. This solution was still stable, when it was cooled to room temperature. 91 Direct fluorination in ethylene carbonate The above mentioned solution of ethylene carbonate and [N(Bu)4]BF4 was fluorinated using the minireactor #2 and the peristaltic pump. 7.4 equivalents fluorine were added, regarding the amount of the salt. This led to several broad signals in the 19F-NMR spectrum (Figure 82). From 2D-NMR experiments the signals are fluorine atoms connected to aliphatic chains. But because of the large number of signals and possible combinations further analysis was not possible. The 1H-NMR spectrum showed two regions of very broad signals. First from 1.5 to 2.5 ppm and second from 3.0 to 4.0 ppm. Though no identifiable product was obtained, this reaction showed the feasibility of the direct fluorination of ammonium salts with ethylene carbonate as solvent. This was the first step to planned fluorinations of ionic liquid compounds. Figure 82 19 F NMR Spectrum of the fluorination of [N(Bu)4]BF4 in ethylene carbonate. 92 5. Summary The work and results of this thesis could be split in two parts. The first part was the reactor development in cooperation with the IMTEK in Freiburg i. Br., and the second the investigations of direct fluorinations using these reactors. Reactor development Two gas liquid slug reactors were developed together with the IMTEK in the group of Prof. P. Woias and an additional low budget capillary reactor was realized. Tests for a Microreactor: The main targeted reactor for the direct fluorination was a silicon chip microreactor, based on a 2x2 cm-silicon chip (Figure I). The chip was designed with 300 µm meandering channels and was optimized for bubbly flow. Because of the spontaneous reaction of fluorine with silicon to silicontetrafluoride, the silicon needed to be protected by an inert metal coating on a sputtered palladium start layer. Possible coatings to be evaluated were statically fluorinated by first diluted fluorine and later pure fluorine to check the suitability of the coating as a protective layer. Gold and copper layers showed leaks in the coverage and an insufficient adhesion to the chip’s surface. Nickel was found to be a suitable material to protect the silicon. Galvanic nickel needed a thickness of 16 µm to form a leaktight layer on a test chip with narrower structures than on the planned reactor itself. The surface was fluorinated with 1 bar 100 % fluorine for at least 4 h and no change of the coating was observed. To bond silicon chips to form closed channels, the nickel layer should be extended by an additional gold layer, which showed better bonding properties, without changing the protective properties. Figure I Left: Construction scheme of the reactor plate for a 2x2 cm microreactor silicon chip (by P. Lang [75] ) Right: SEM picture of the reaction channels (chip and picture by K. Cobry, IMTEK Freiburg). Minireactors: For the direct fluorination reactions two minireactors were developed with the IMTEK. They are based on nickel coated copper plates with drilled reaction and cooling channels. The reaction channel has a length of 53 cm and a 1x1 mm rectangular profile. The 93 channel is meandering with rectangular corners. Two generations of minireactors were realized. The first generation showed in fluorination experiments weaknesses at the gas inlet, which were solved in the second generation minireactor. Figure II left: microreactor #2 before final assembly. Layers: 1 Liquid and gas connectors; 2 Channels for temperature sensors and reaction channel; 3 heat sink; 4 coolant connectors; middle: photo of reaction channel plate; right: scheme of the alignment of reactor and cooling channels and the position of the temperature sensors. Capillary Reactor: The capillary reactor was based on a 1/16 inch OD PFA capillary and a Swagelok T-connector. The capillary had an inner diameter of 0.5 mm, which is in the range of a typical microreactor. The T-connector’s inner diameter was with 1.27 mm more in the range of a minireactor. The total cost for the reactor were around 100°€. Figure III Left: tube reactors inlet feed set up. Right: tube reactor in use 1: 1/8‘‘ PFA tube for gas feed, 2: 1/8‘‘ PFA tube for liquid feed, 3: T-connector, 4: 1/16‘‘ PFA capillary, 5: PFA sample collecting flask, 6: copper coil in a cooling medium . Additional to the development of the reactors, their periphery was set up and continuously improved. For the gas feed two gas flow controller were installed, which were controllable from outside of the fume hood. A syringe pump was first used for the direct fluorination, to investigate different substrates in their behavior for the direct fluorination reaction. Using a syringe pump limited the maximum fluorine to substrates ratios (≈ 5 %). To improve this, a recyclisation of the liquid product stream was realized and optimized. For this system a peristaltic pump was installed, and a gas liquid separator as well as a pulsation dampener was self build from PFA tubes. 94 Fluorination Reactions Direct fluorinations were carried out using the minireactor and the capillary reactor. Acetonitrile, iso-propylacetate, propylacetate, n-butyl butyl trifluoroacetate, ethyl acetoacetate, toluene t and methyl tert-butyl ether were fluorinated only using the syringe pump set up. Fluorination of Toluene: Toluene in acetonitrile was the most often used test reaction for the direct fluorination in microreactors microreactors in literature. The main product of the direct fluorinaton in the minireactor showed the four mono fluorinated isomers as the main products, besides a larger number of side products. The isomer distribution was close to a distribution found in a microreactor with much smaller channels. The reactions of toluene leaded to several clogging of the reactor. Investigations of the the blackish solid material found in the reactor showed the formation of toluene oligomers and other higher molecular weight side products. Figure IV 19 F NMR spectrum of the reaction mixture after direct direct fluorination of toluene after washing with water. -1 –1 Conditions: T= 0 °C. 12.5 vol% toluene in acetonitrile. Liquid flow= 90 ml h . gas flow= 480 ml h . 25 vol% fluorine. a: meta-fluorotoluene, b: ortho-fluorotoluene, ortho c: para flurotoluene, d:fluoromethylbenzene. Fluorination of Organic Acetates: Acetates The fluorinations of iso-propylacetate, propylacetate, n-butyl trifluoroacetate and ethyl acetoacetate were carried out successfully, but because of the high number of possible isomers and the low fluorine to substrate rate this substrates were too complex for reactor evaluating experiments. ex Fluorination of Methyl-tert--Butyl Ether: When methyl-tert-Butyl utyl ether was fluorinated a decomposition of the molecule took place. As main fluorinated product tert-butyl-fluoride tert was found in the product mixture. Only traces of fluorinated methyl-groups methy groups were observed. CH3 H 3C F CH3 Figure V 19 -1 –1 F NMR spectrum of the products of direct fluorination of MTBE. Conditions: Conditions flow = 40 ml h . F2: –1 3 mL min . N2: 3 mL min . T= –20 20 °C. 19 F = –131.8 ppm: decet of tert-butyl-fluoride. 95 Fluorinations with Recyclisation: Ethylene carbonate, propylene carbonate, closoK2[B12H12] and tetraalkylammonium salts as substrates were fluorinated with the recyclisation of the product liquid system. Ethylene Carbonate: Ethylene carbonate, a solid cyclic carbonate broadly used as solvent in lithium ion batteries, was directly fluorinated using its mono fluorinated isomer as solvent. It was shown to be very stable regarding decomposition in the direct fluorination reaction. Practically no side products were found when ethylene carbonated was fluorinated even when using almost undiluted fluorine. For a conversion of 80 % the mono fluorinated isomer was synthesized with 49 % yield. Taking only the reaction volume into account a space-time-yield of averagely 8600 mol m-3 h-1 were achieved with the minireactor. The experimental obtained distribution of the three possible difluorinated isomers, was thermodynamic data calculated on G3 level. With ethylene carbonate also measurements using the integrated temperature sensors if the minireactor were carried out. They showed no increase of temperature with activated active cooling. Without the cooling, an increase of temperature within the first four centimeter was observable. In a simple calorimetric measurement of the reaction, with the reactor placed within a Styrofoam box, the measured energy found was about 105 % of calculated value. 4,4-F2EC O O O cis-F2EC trans-F2EC F1EC O O O O O O O O O F -80 19 F -100 F F BF4 Standard ab Figure VI. Two F F F -120 -140 ppm F-NMR traces of EC after the fluorination reaction. Conditions: 30 wt% F1EC as solvent. a: 0.5 equiv. F2, T = 22 °C, liquid flow = 2.5 mL min –1 –1 –1 gas flow = 6.6 mL min , 45 vol% F2. b: 0.5 equiv. F2, T = 22 °C, –1 – liquid flow = 2.5 mL min , gas flow = 3.4 mL min , 88 vol% F2, minireactor. The [BF4] signal is formed from residual HF. Propylene Carbonate: Propylene carbonate had a lower melting point than ethylene carbonate, which allowed investigations on the influence of reactions parameters including the temperature. In these experiments, the temperature was the only parameter investigated with a low significant influence. The gas flows, nitrogen as well as fluorine, had only a very 96 small effect. The development of the four possible mono fluorinated isomers in dependence of the conversion showed a maximal mono fluorinated isomers yield of 37 %. With 23 % the methyl group is the most favoured isomer. On G3 calculated thermodynamic values showed this group to be thermodynamically least favoured. Closo-[B12H12]2-: The cyclic carbonates were tested as possible solvents for the direct fluorination of ionic substances. As first system the fluorination of closo-[B12H12]2- clusters were investigated. Before the cluster was fluorinated using the minireactor, the reaction was carried out in a classical glass flask approach in acetonitrile. The reaction in the minireactor, with a mixture of ethylene and propylene carbonates as solvents, showed a yield of 58 % of the perfluorinated cluster and 12 % of polyfluorinated clusters. This value is larger as our batch approatch, but still smaller than a published batch of Strauss et al., who published a yield of 74 %.[138] Initial Fluorinations of [N(Bu)4]BF4: The second ionic system was [N(Bu)4]BF4. Here only initial experiments were carried out. The tetrafluoroborate salt was chosen, because when bromide and chloride ammonium salts where led through the reactor, even before any fluorine was present, the halogen atoms were oxidized to the halogens. This indicates a possible active role of the reactors walls in the direct fluorination. Using [N(Bu)4]BF4 this was not observable. It was possible to dissolve the salt in ethylene carbonate at room temperature due to reduction of the melting points. The NMR spectra of this reaction showed numerous signals of fluorinated alkyl groups, but the perfluorination was not reached and remains a future task. Conclusion In this thesis the direct fluorinations of a series of organic and inorganic compounds have been investigated in a novel minireactor. According to the most often used test reaction in the literature, the direct fluorination of toluene in acetonitrile was investigated. However, it was found to be unsuitable for many reasons. By contrast, the direct fluorination of cyclic carbonates was found to be very promising as a benchmark test reaction. It was possible to safely fluorinate those carbonates at fluorine concentrations up to 100 %. In general, the fact that the reactions were almost independent from a large range of parameters, temperature as well as gas flows, was unexpected, but allows increasing the efficiency of the reaction. The possible fluorination with 100 % fluorine can play an important role in the future design of the silicon chip microreactor allowing multiple gas inlets in one reaction channel, without the cumulative addition of the diluting nitrogen flows. On our way to this microreactor a lot has 97 been learned from the performed direct fluorinations and we built up expertise, which requirements a successful microreactor needs to fulfill. Also the finding that especially the organic carbonates should be used as alternative solvents to acetonitrile in the direct fluorination of ionic substances is worth to be further investigated in future work. 6. Experimental Section 6.1 Used Materials: 99.9 % fluorine was donated by Solvay Fluor. Germany. Commercially available ethylene carbonate (99%) from Alfa Aesar. acetonitrile (HPLC grade) form VWR. sodium fluoride (pure) from Merck and tbutyl-methyl-ether (> 99 %) from Riedel de Häen were used 4-fluoroethylencarbonate was donated by the Solvay Fluor. Germany. Closo-Na2[B12H12] and closoK2[B12H12] were prepared according to the method of Knapp et al.[151]. 6.2. Line Operations To operate the line safely special operations are used, which are not common in normal Schlenk lines techniques. Some of the most important techniques are presented following. 6.2.1 Leak testing When the vacuum pressure in the line is less good than normal, there can be two reasons. First, a substance was absorbed to the lines wall and is now increasing the pressure by slow evaporation. This problem can be identified by slow improvement of the vacuum over time. It is also possible to heat the line and an increase of pressure would show absorbed material. If there is a leak in a joint it also can behave this way, so it is best to heat where plain tubing is. The second alternative is a real leak. The best way to locate such leak is to first shut as many valves as possible. Then they are opened one after another to first locate the section.[152] To locate a large leak, set an increased inert gas pressure (approx. 1.2 bar) to the line and use a leak detection spray. Smaller leaks can be found by setting vacuum to the line and apply some liquid (e.g. acetone) to the suspected joints. Observe the pressure while doing so. If a leak is present, the gas flow is reduced for a short time, and an improvement of the vacuum can be observed. 98 6.2.2 Measuring lines volume To measure the volume of a metal line, which is essential to calculate the mass of containing gases by known pressure, SO2 can be used. First the line and a glass flask are well evacuated. The weight of the evacuated flask is to be measured exactly. Then a certain measured pressure of SO2 is expanded to the part of the line which is to be measured. Then the glass flask is connected to the line and the SO2 condensed into it. The flask is then weighted again and by using the ideal gas law the volume of the line can be calculated. V mµ¶3 RT Mµ¶3 p nRT p This procedure is to be repeated three times to reduce the error. A change of the lines volume over time is possible. A later repetition is recommended.[152] 6.2.3 Purging of the line When fluorine or a fluoride gas is present its need to be quenched. Most important: the direct connection of the lime to the vacuum stays closed to the very end of this procedure! First the soda lime tower (tower) is well evacuated. Then the valve of tower-vacuum is closed and the valve line-tower is slowly opened. When there is no pressure drop in the line anymore, the connection line-tower is closed instantly. This fast closing is to prevent water, which is in the tower to get into the line. After this the line is purged with nitrogen. If the vacuum is still sufficient in the tower the nitrogen is also transferred by opening tower-line valve into the tower. If the pressure is not sufficient, after a few minutes of quenching time, the vacuum of the tower can be renewed. At least purge three times with nitrogen. Then open the vacuumtower valve before opening the line-tower valve. Now vacuum is directly connected to the line, but to remove residue traces the tower is still in between. By doing so, the maximum vacuum is around 5 mbar. When the pressure does not lower anymore first close the linetower valve then the tower-vacuum line. Finally the main line-vacuum valve is opened and waited till the vacuum is at maximum. Then close the line-vacuum valve and expand a slight overpressure (approx. 1.1-1.2 bar) to the line, to finish the procedure. 6.2.4 Connecting of PFA tubes To connect two PFA tubes, in one a solid, lengthy object must be introduced till it comes out on the opposite side for at least approximately 3 cm. The object can be various things. Most suitable would be a metal staff or wire. But especially for small diameters canulas can be used. The use of glass pipettes is possible, but removing it after the reaction can lead to a 99 breaking of the glass. In case both tubes have the same diameters the open objects end is introduced into the second tube till it touches the first one. In case of two different diameters the object is introduced into the smaller one and the larger tube is, if possible, placed interleaving for about 5 cm. The set up is now relatively stable. Holding it on both tubes or the on tube and the object on the other side, the connecting position or where the tubes are interleaving can be heated by a heat gun (600 °C). It is important not to heat any part in which the object is not placed. When the PFA is melting, it changes from a milky look to clear. The tubes are now pushed slightly against each other and a carefully twisting of one tube helps to form an even connection. The movements must be carried out carefully to avoid a rupturing of the molten tube. To connect a smaller tube to the wall of a larger one at first a hole must be created into the larger tubes wall, which is thinner as the outer diameter of the smaller tube. The stabilizing object is now placed into the tubes in the connection zone. When heating the tubes it is important to selectively heat the smaller tube, which normally melts earlier, due to smaller wall strength. The larger tube needs to be heated, but must not be liquid. Now the small tube is gently pushed against the larger one in a twisting movement to connect the tubes. If the tubes were connected successfully, can be checked using a liquid. This is strongly recommended, because leaks often are very hard to detect visually. 6.2.5 Preparation of a hydrogen fluoride containing NMR sample Hydrogen fluoride can react with the glass of a NMR tube. If only traces of HF are present it will only lead to some surface etching, this will not endanger the safety. But in the 19F-NMR spectrum a signal of BF3 appears. To protect the glass it is possible to use PFA tubes with a OD of 3.5 mm. First a piece of tube is prepared which is around 3 cm longer than the NMR tube. To seal the bottom end, the PFA tube is placed inside a NMR tube, and with a heat gun (600 °C) the very end of the NMR tube is heated. Meanwhile the PFA tube is carefully rotated and some pressure toward the NMR tubes bottom is applied on the PFA tube. When the end begins to melt, which is indicated by the PFA getting transparent, it now forms a closed cap at the end of the PFA tube. The melted zone must not be too large, because otherwise the solid PFA will be in the active detection zone of the NMR spectrometer. After the seal is formed, remove the heat gun, wait for 20- 30 seconds and pull out the tube. 100 6.3 General Comments On Working With Elemental Fluorine Fluorine is a highly reactive gas, which forms upon reaction highly toxic hydrogen fluoride. For working with those two substances special precautions need to be respected. 6.3.1 Risks, toxicity and treatment Fluorine: Its hazard statements are H270 (intensify fire, oxidizer), H330 (fatal if inhaled), H314 (causes severe skin burns and eye damage).[153] The effects of fluorine itself are caused by the massive oxidation potential causing burns when in contract to organic material. These effects are limited to the body’s surfaces, which is directly only dangerous if inhaled. Especially the formation of oedemas can be fatal. Hydrogen fluoride: Working with this substance requires increased attention. Hydrogen fluoride is dangerous in all of its states. As a gas is can easily inhaled, as liquid it is highly concentrated and as aqueous solution it can easily mistaken as water. Unlike other acids (e.g. HCl) it can penetrate skin and damage deeper tissue. Additionally it can cause severe systemic problems. Diluted acid can lead to a prolonged initial time in which no or only light symptoms appear. This inital time can be between 1 and 24 hours. A list of systemic symptons of acute hydrogen fluoride intoxication is shown in Table 25. Table 25 Signs of acute systemic fluoride toxicity [154] Body system Symptoms General Malaise, weakness, pallor Cardiopulmonary Tachycardia, hypotension, prolonged QT interval, ventricular fibrillation, pulmonary oedema Neurologic / Respiratory depression and paralysis, CNS neuromuscular depression, carpopedal spasm, tetany, seizures Metabolic Hypocalcemia, hyperkalemia, hypomagnesemia To prevent accidents with hydrogen fluoride it is crucial to stick to the usual precautions when working with hazardous substances. It is important to work in a well ventilated fume hood, because of the volatility of HF. Additionally to a lab coat, when working with high concentrated solutions, an additional leather apron is recommendable. Also the use of a protective visor in addition to the lab glasses should be considered. It is crucial to wear 101 sufficient hand protection. Gloves are imperative. A good choice would be a pair of thicker nitrile gloves and one-use gloves on top. The one-use gloves and its contamination can be instantly removed upon contact. Wearing thin gloves only is not sufficient, because of some permeability of hydrogen fluoride. If there is a spilling of aqueous hydrogen fluoride or it is just suspected it is advisable to check if the suspicious puddle is containing HF. To do so use a pH sensitive paper or some type of carbonate or hydrogen carbonate. This will indicet the present of acid by colour or bubbles. If carbonates or hydrogen carbonates are used the acid will be quenched. Attention! Many fluorides, which are the products of a quenching, are still toxic, though not volatile and less skin permeable. Calcium and magnesium fluoride are classified as not toxic. It is advisable to use salts of this metal to quench HF. Table 26 Effects, critical amounts and treatments of different HF exposures Exposure Skin Effect Critical amount Serious painful burns > 160 cm ; also cause in layers, deep tissue whitening of skin, blistering, edema Coughing, 2 Treatment can serious Lung pulmonary edema, 2.calcium gluconate gel or injection, 3. Systemic observation First Aid: 1. Oxygen, 2. 2.5 % calcium gluconate by Nebulizer > 30 ppm Medical 1. 2.5 % calcium respiratory distress and edemas airway Eye treatment: gluconate by Nebulizer, 2. Observation for swelling of the upper Severe Medical treatment: 1. debride (if necessary), intoxicaiton choking, cyanoses, First aid: 1. water, 2. calcium gluconate gel systemic chest tightness, chills, fever, [155] burns destruction opacification of with < 8% causes or damage which heal the > 20 % opacification cornea, blindness and necrosis [154] First aid: 1. Wash Medical treatment: topical tetracaine hydrochlorid, consult eye specialist First aid: 1. milk or water, do not induce Severe burns in mouth, Ingestion esophagus stomach, and systemic vomiting. -1 [154] 20 mg kg Medical treatment: 1. lavage with calcium gluconate effects or calcium chloride 2. treat systemic effects In Table 26 effects, critical amounts of acid and its treatment, separated by type of exposure are shown. In all cases is important not to spread the contamination. Before first aid the affected person is to moved away from the hydrofluoric acid source. When removing contaminated clothes it is to be insured nobody touches parts where still HF can be present. 102 Generally a treatment with calcium gluconate solution/ gel is always advisable. Always, even when a contact is just suspected, go for a professional medical examination. For this case always have some official paper ready, in which the treatment of hydrofluoric acid is described in details. Most medical staff is not accustomed to such contaminations! 6.3.2 Passivation of tubing and containments All metal tubing and containments were passivated prior use to form a protective layer of metal fluoride on the inner surfaces. After everything was leak tested fluorine pressure was slowly increased to 0.25 bar. After two hours the pressure was increased to 1.0 bar. This pressure was left over night. Next day all parts were heated for a few minutes. Then the fluorine was removed. 6.3.3 Use of iodides for fluorine detection An easy technique to detect the presents of fluorine it to use iodide salts, e.g. sodium iodide. When in contact to fluorine the iodide gets oxidised to elemental iodine (Equation 36). 2 NaI + F2 → I2 + 2 NaF (36) There are two main applications for this technique. On the one hand the product gas stream and be led through an aqueous iodide solution. A yellow brown colouring of the solution indicates fluorine and defending on the type of reaction can show the completion of the fluorination. The colour can be enhanced by the adding of a starch solution, which would change the colour to blue. On the other hand filter paper stripes can be wetted with an aqueous iodide solution. They can be used to indicate leaks at the fluorine bottle or the pressure gauge, especially after changing the fluorine bottle. A slight yellow colouring, especially at the edges of the paper can occur without the present of fluorine. 6.3.4 General safety advises A) Always pay attention to the line pressure. The pressure is the only way to find leaks. B) Never open valves to fast when handling fluorine. The high speed of the sudden fluorine flow can start unwanted reactions e.g. ignite Teflon seals, or your substrate react more vigilantly than you expected. 103 6.4 Chip Coating Tests The chipholder, its PTFE seals and the chip were degreased using soft non fuzzing tissues and acetone. After the chip was placed in the holder and the lid closed, the screws of the chip holder were tightened carefully, by tightening them in several steps and always with one opposite of the screw before. The assembled holder was placed into a vacuum tight, fluorine resisting vessel, which had been either a 120 mL PFA flask or a monel autoclave. Vacuum was applied, till end vacuum (1*10-2 mbar) was reached, before a certain pressure of fluorine/nitrogen mixture or pure fluorine was led into the line and flask. The pressure drop was observed during the exposure time. In case of rapid or critical pressure loss (e.g. 1 bar→ < 0.8°bar) the experiment was aborted, to avoid to heavy damage at the silicon. In these cases or after reaching the intended exposure time the fluorine was removed and the system evacuated until the pressure was constant (1*10-2 mbar). For experiments under inert conditions the flask now was transferred into a glove box. Normally this was not necessary. 6.5 NMR Conditions For the analysis of the experiments an Advance II+ 400 (Bruker. Germany) with a 5 mm broad band fluorine observation (BBFO) head was used. To enable a precise integration of the NMR signals for the ethylene carbonate reactions, the spectra were recorded using the following optimized settings: 19 F-NMR spectra at 377 MHz were set to a spectral width of 41.667 Hz. an acquisition time of 3.15 s and a relaxation D1 delay of 30 s. The size of the spectra was 262.144 data points. For the 1H-NMR spectra at 400 MHz a spectral width of 6410 Hz. an acquisition time of 5.11 s and a D1 delay of 30 s was used. The spectral size was 65.536 data points. For both nuclei, spectra with 32 scans were recorded. An exponential function with a value of 0.5 was applied to the spectra before processing. The phase and baseline corrections as well as the integration were carried out very precisely. Using this procedure, the error bar of the measurements was found to be around 1%. 6.6 Fluorinations 6.6.1 Fluorination of toluene and TBME First the nitrogen flow was started and the substrate was diluted with acetonitrile. The liquid was filled into a glass syringe equipped with a Luer-lock valve. The syringe was placed into 104 the syringe pump and the pump was started at the desired flow level. When the liquid started to exit the reactor, the fluorine flow was started and turned off when the syringe had another two minutes before being empty. During the reaction, the toluene (1) solution showed a tendency to turn to a light brown colour after the reaction. The liquid flow was switched off when the syringe was empty. After a further 20 min of purging the PFA vessel and the reactor, the nitrogen flow was switched off. Work up 1: The mixture was transferred to a different vessel equipped with excess sodium to scrub the hydrogen fluoride. For the separation of acetonitrile the mixture was washed three times with water. Remaining sodium fluoride and bifluoride were removed by filtration. Work up 2: The product solution was diluted with acetonitrile and washed with an aqueous sodium hydroxide solution to neutralize the acid. Then it was washed with water until the liquid level was almost constant, which indicated the acetonitrile to be removed. The remaining liquid was dried over dry magnesium sulfate and filtered. Because of the low maximal conversion, caused by the syringe pump, the investigated analytics were only qualitative. Monofluoro-acetonitrile 19 1 F (200MHz, CDCl3): -232 (1F, t, 1JFH= 46.1) [Lit: -232 ppm][127] H (200MHz, CDCl3): 4.88 (2H, d, 1JHF= 46.1 Hz) [Lit: 4.59 ppm][127] 13 C (100.6MHz): 67.3 (CH2F-CN) [Lit:67.1 ppm][156], 114.5 (CH2F-CN) [Lit 114.08 ppm][156] Difluoro-acetonitrile 19 F (200MHz, CDCl3): -117 (2F, d, 1JFH= 50.8 Hz) [Lit: -119 ppm][127] tert-Butyl fluoride: 19 1 F (200MHz, CDCl3): -131.84 (1F, decet, 3JFH= 20.6 Hz) [Lit: -131ppm][157] H (200MHz, CDCl3): 1.38 (9H, d, 3JHF= 20.6 Hz) [Lit: 1.39 ppm] [157] 105 6.5.2 Fluorination of iso-propylacetate Capillary reactor reactions The iso-propylacetate was pure or diluted with acetonitrile placed into a syringe pump and connected with a 1/4 '' PFA tube. The tubes diameter was reduced to the reactors 1/16 '' using stainless steel connectors. First the nitrogen flow then the liquid flow was started. The fluorine valve was opened shortly after the liquid started to leave the capillary. After sufficient material was collected, first the fluorine then the liquid flow was switched off. The reaction time was planned to be long enough to compensate the error caused by the nonpresent fluorine flow in the first seconds and when switching of the fluorine. The reaction mixture was led into a collector vessel, containing a mixture of acetonitrile and sodium fluoride. Batch reaction The substrate was placed in a PFA flask. A PFA tube led the waste gases from the reaction vessel into a soda lime trap. The flask was cooled from the outside using a cooling bath. The nitrogen-fluorine mixture was then bubbled into the stirred solution of pure or with acetonitrile diluted iso-propylacetate. For bubbling in, a 1/8'' PFA tube was used. The opening of the tube was made smaller by manipulation of the PFA with a heat gun to reduce the bubble size (inner diameter approximately 0.1 mm). After the reaction was completed, the reaction flask was well purged with nitrogen, and the remaining hydrogen fluoride was quenched with sodium fluoride. 3 O 1 4 5 O 2 1-Fluoromethylethylacetate 1 H (CDCl3, 400 MHz, 298°C): δ= 1.26 (3H, 3JHH=6.6 Hz, 4JHF=1.4 Hz, 3), 2.04 (3H, s, 5) 4.29-4.52 (2H, m, 2), 5.04-5.17 (1H, m, 1) 13 C (CDCl3, 101 MHz, 298°C): δ= 14.4 (3), 20.1 (5), 68.3 (1), 83.7 (2), 169.5 (4) 19 F (CDCl3, 376 MHz, 298°C): δ= -229.6 (1H, tdq, 2J=47.6 Hz, 3J=20.2 Hz, 4J=1.4 Hz, 2) 1-Methylethylfluoroacetate 1 H (CDCl3, 400 MHz, 298°C): δ= 1.29 (6H, d, 3JHH=6.2 Hz, 2+3), 4.82 (2H, d, 2JFH=47.1 Hz, 5) 5.12 (1H, sept, 3JHH=6.2 Hz, 1) 106 13 C (CDCl3, 101 MHz, 298°C): δ= 20.9 (2+3), 77.0 (5) 65.8 (1), 166.8 (4) 19 F (CDCl3, 376 MHz, 298°C): δ= -239.6 (1F, t, 2JFH=47.1 Hz, 5) 1-Fluoromethyl-2-fluoroethylacetate 1 H (CDCl3, 400 MHz, 298°C): δ= 2.09 (3H, s, 5), 4.60 (4H, dd, 2JHF=47.3, 3JHH=4.5, 2+3), 4.86 (1H, m, 1) 13 C (CDCl3, 101 MHz, 298°C): δ= 19.8 (5), 69.9(1), 79.8 (2+3), 169.4 (4) 19 F (CDCl3, 376 MHz, 298°C): δ= -234.6 (1F, dd, 2JFH=47.3 Hz, 3JFH=20.1 Hz, 2+3) 1-Methyl-1-fluoroethylacetate 19 F (CDCl3, 376 MHz, 298°C): δ= -96.5 (1F, sept, 3JFH= 19.1 Hz) 5.6.3 Fluorination of ethyl acetoacetate The ethyl acetoacetate was filled into a glass syringe equipped with a Luer-lock valve. The syringe was placed into the syringe pump. To start the reaction first the nitrogen flow was set to 4 mL min-1, and then the pump was started at a flow level of 4 mL min-1. When the liquid started to exit the reactor, the fluorine flow was started and turned off when the syringe had another two minutes before being empty. The liquid flow was switched off when the syringe was empty. The product was collected in a PFA vessel, which was well purged, before the hydrogen fluoride was purged by a mixture of acetonitrile and sodium fluoride. O 1 2 O 4 3 O 5 6 CH3COCHFCOOC2H 19 F (CDCl3, 376 MHz, 298°C): δ= -194.56 (1F, dq, 2JFH= 48.5 Hz, 4JFH= 3.5 Hz) 5.6.4 Fluorination of ethylene carbonate For the fluorination of ethylene carbonate it was dissolved in F1EC at 40 °C prior to the experiment. First the nitrogen flow was started and the liquid was filled into the separator vessel. The circulation pump was started with a low flow of 0.5 mL min–1. When stable gas slugs were observed at the outlet, the liquid flow was carefully increased to the desired level between 0.5 mL min–1 and 2.5 mL min–1. Only when the system was running stable, the fluorine flow was started and turned off after the desired reaction time. The liquid flow was switched off 5 min after the fluorine flow was turned off. After a further 20 min of purging 107 the PFA vessel and the reactor, the nitrogen flow was switched off. Two different work up methods were used: For low HF amounts the mixture was transferred to a different vessel, equipped with excess sodium fluoride (3 g per 3 g substrate) and acetonitrile (4 mL per 3 g substrate) to scrub the hydrogen fluoride. The mixture was stirred for two minutes, and then the solution was filtered. For high HF amounts the mixture was transferred into a PFA beaker containing an excess of silica gel and acetonitrile. After stirring for 3 hours, the mixture was filtered and washed with acetonitrile. p-Fluoro-toluene was used as an internal standard for NMR analysis (typically 0.3-0.7 g per 3 g substrate). O 3 2 O O 1 4 5 Monofluorinated ethylene carbonate 1 H NMR (400 MHz, CD3CN): 4.45 (1H, ddd, 3JHF=21.6 Hz, 2JHH= 11.2 Hz, 3JHH=1.1 Hz, 5), 4.60 (1H, ddd, 3JHF=34.0 Hz, 2JHH= 11.2 Hz, 3JHH=4.2 Hz, 5), 6.31 (1H, ddd, 2JHF=64.3 Hz, 3 JHH= 4.2 Hz, 3JHH=1.1 Hz, 4) 13 C NMR (100.6 MHz, CD3CN): 71 (5), 106 (4), 153 (2) 19 F NMR (376.5 MHz, CD3CN): –123.7 (1F, ddd, 2JFH=64.3 Hz, 3JFH= 34.0 Hz, 3JFH=21.6 Hz, 4) Difluorinated ethylene carbonate 4,4-Difluoro-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN): 4.79- 4.85 (2H, m, 5+5) 13 C NMR (100.6 MHz, CD3CN): 71 (5), 125 (4), 148 (2) 19 F NMR (376.5 MHz, CD3CN): –74.10- –74.16 (1H, m, 4+4) The NMR signals for 4,4-Difluoro-1,3-dioxolan-2-one have been previously published by Kobayashi et al.[135] and Ishii et al.[158], but both group groups misinterpreted the signals in the 1 H and 19F spectra. The signals formed by the –CF2-CH2- group is, though it appears to be a triplet with a coupling constant of 12.0 Hz, a signal of higher order. It is an AA’XX’ system in which a dominating 19F-19F coupling is responsible for the formation of a pseudo triplet. 108 Trifluorinated ethylene carbonate 4,4,5-Trifluoro-1,3-dioxolan-2-one: 19 F NMR (376.5 MHz, CD3CN): –81.3 (1F, ddd, 2JFF=142.4 Hz, 3JFF=10.1 Hz, 3JFH=2.5 Hz, 4), -94.5 (1F, ddd, 2 JFF=142.4 Hz, 3 JFF=5.2 Hz, 3 JFH=0.5 Hz, 4), –141.6 (1F, ddd, 3 JFH=59.2 Hz, 3JFF=10.1 Hz, 3JFF=5.2 Hz, 5) 6.6.5 Fluorination von 4-methyl-1,3-dioxolan-2-on (PC) Fluorination in the minireactor #2 First the nitrogen flow was started and the PC was filled into the separator vessel. The circulation pump was started with a low flow of 0.5 mL min–1. When stable gas slugs were observed at the outlet, the liquid flow was carefully increased to the desired level between 0.5 mL min–1 and 2.5 mL min–1. Only when the system was running stable, the fluorine flow was started and turned off after the desired reaction time. The liquid flow was switched off 5 min after the fluorine flow was turned off. After a further 20 min of purging the PFA vessel and the reactor, the nitrogen flow was switched off. Two different work up methods were used. For low HF amounts the mixture was transferred to a different vessel, equipped with excess sodium fluoride (3 g per 3 g substrate) and acetonitrile (4 mL per 3 g substrate) to scrub the hydrogen fluoride. The mixture was stirred for two minutes, and then the solution was filtered. For high HF amounts the mixture was transferred into a PFA beaker containing an excess of silica gel and acetonitrile. After stirring for 3 hours the mixture was filtered and washed with acetonitrile. p-Fluoro-toluene was used as an internal standard for NMR analysis (typically 0.3-0.7 g per 3 g substrate). Fluorination in the capillary reactor The PC was filled into a glass syringe equipped with a Luer-lock valve. The syringe was placed into the syringe pump. To start the reaction first the nitrogen flow was set to 12 mL min-1, and then the pump was started at a flow level of 1.372 mL min-1. When the liquid started to exit the reactor the fluorine flow was started and turned off when the syringe had another two minutes before being empty. The liquid flow was switched off when the syringe was empty. The product was collected in a PFA vessel, which was well purged, before the hydrogen fluoride was purged by a mixture of acetonitrile and sodium fluoride. 109 O 3 2 O O 6 4 1 5 Monofluorinated isomers of propylene carbonate: All assignments are based on 1H and 19 F 1D-NMR and HMBC, HSQC, NOESY 2D-NMR experiments of isomer mixtures. 4-Fluoromethyl-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN):4.35 (dd, 1H, 2JHH= 8.8 Hz, 3JHH= 6.1 Hz, 5), 4.58 (ddd, 1H, 2 JHF=64.9 Hz, 2JHH=11 Hz. 2JHF=3.7 Hz, 6), 4.58 (ddd, 2JHF= 8.8, 2JHF= 8.8, 4JHF=1.4, 5), 4.71 (ddd, 1H, 2JHF=64.9, 2JHH=11.3 Hz, 3JHF=2.0 Hz, 6), 4.91- 5.02 (m, 1H, 4) 13 C NMR (100.6 MHz, CD3CN): 65.0 (5), 75.2 (4) 82.2 (1JCF = 155 Hz, 6), 154.9 (2) 19 F NMR (376.5 MHz, CD3CN): δ –236.4- -237.6 (t, 1F, 2JFH= 46.9 Hz, 3JFH=25.3 Hz, 2JFH= 1.4 HZ, 6) Trans-4-Fluoro-5-methyl-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN): δ 1.45 (dd, 3H, 3JHH= 6.8 Hz, 4JHF=0.7 Hz, 6), 4.86 (dqd, 1H, 3 JHF= 19.6 Hz, 3JHH= 6.8 Hz, 3JHH= 0.9 Hz, 5), 6.07 (1H, dd, 2JHF= 63.4 Hz, 3JHH= 0.9 Hz, 4) 13 C NMR (100.6 MHz, CD3CN): δ 15.7 (6), 79.9 (4), 109.2 (5), 152.4 (2) 19 F NMR (376.5 MHz, CD3CN): δ -122.4 (ddq, 1F, 3JFH= 19.6 Hz, 2JFH= 63.4 Hz, 4JHF=0.7 Hz, 5) Cis-4-Fluoro-5-methyl-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN): δ 1.46 (dd, 3H, 3JHH= 6.6 Hz, 4JHF=2.4 Hz, 6), 4.95 (dqd, 1H, 3 JHF= 25.6 Hz, 3JHH= 6.6 Hz, 3JHH= 4.0 Hz, 4), 6.25 (dd, 1H, 2JHF= 64.2 Hz, 3JHH= 4.0 Hz, 5) 13 C NMR (100.6 MHz, CD3CN): δ 11.3(6), 77.8 (4) 106.1(5) 152.4 (2) 19 F NMR (376.5 MHz, CD3CN): δ –141.2 (ddq, 1F, 3JFH= 64.2 Hz, 2JFH= 25.6 Hz, 4 JHF=2.4 Hz, 5) 4-Fluoro-4-methyl-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN): δ 1.82 (d, 3H, 3JHF= 18.0 Hz, 6), 4.46 (1H, 3JHF= 32.4 Hz, 2 JHH= 10.6 Hz, 5), 4.60 (dd, 1H, 3JHF= 17.6 Hz, 2JHH= 10.6 Hz, 5) 110 13 C NMR (100.6 MHz, CD3CN): δ 19.9 (2JCF= 34 Hz, 6), 74.9 (5), 115.2 (4), 152.7 (2) 19 F NMR (376.5 MHz, CD3CN): δ –92.1- –93.8 (m,1F, 4) Difluorinated isomers of propylene carbonate: All assignments are based on 1H and 19 F 1D-NMR and HMBC, HSQC, NOESY 2D-NMR experiments of isomer mixtures. 4-Difluoromethyl-1,3-dioxolan-2-one: 1 H NMR (400 MHz, CD3CN): δ 4.51 (dd, 1H, 2JHH= 9.36 Hz, 3JHH= 5.08 Hz, 5), 4.60-4.62 (m, 1H, 5), 4.95-5.00 (m, 1H, 4), 6.05 (td, 1H, 2JHF= 26.9 Hz, 2JHH= 2.6 Hz, 6) 13 C NMR (100.6 MHz, CD3CN): 64 (5), 73 (4), 113 (6), 155 (2) 19 F NMR (376.5 MHz, CD3CN): δ –134.0- –136.2 (m, 2F, 6) Cis-4-Fluoromethyl-5-fluoro-1,3-dioxlan-2-one: 1 H NMR (400 MHz, CD3CN): δ 4.77 (ddd, 1H, 2JHF=47.3 Hz, 2JHH= 11.3 Hz, 3JHH= 6.6 Hz, 5), 4.87 (ddd, 1H, 2JHF= 45.3 Hz, 2JHH= 11.3 Hz, 3JHH= 3.6 Hz, 5) 5.02-5.18 (m, 1H, 4), 6.43 (dd, 1H, 2JHF= 63.8 Hz, 3JHH= 4.5 Hz, 5) 13 C NMR (100.6 MHz, CD3CN): δ 79 (4) 79 (6), 105 (5), 152 (2) 19 F NMR (376.5 MHz, CD3CN): δ –140.8- –140.6 (m, 1F, 5), –234.1 (dddd, 1F, 2 JHF= 47.3 Hz, 2JHF= 45.3 Hz, 3JHF= 17.7 Hz, 4JFF= 4.56 Hz, 6) Trans-4-Fluoromethyl-5-fluoro-1,3-dioxlan-2-one: 1 H NMR (400 MHz, CD3CN): δ 4.83 (ddd, 1H, 2JHF= 45.4 Hz, 2JHH= 12 Hz, 3JHH= 2.4 Hz, 6), 4.77 (ddd, 1H, 2JHF= 47.2 Hz, 2JHH= 12 Hz, 3JHF= 4.6 Hz, 6), 4.92-5.07 (m, 1H, 4), 6.39 (dd,1H, 2JHF=62.5 Hz, 3JHH=1.0 Hz, 5) 13 C NMR (100.6 MHz, CD3CN): δ 81 (6), 81 (4), 106 (5), 152 (2) 19 F NMR (376.5 MHz, CD3CN): δ -125.0 (ddd, 1F, 2JHF= 62.5 Hz, 2JHF= 20.1 Hz, 5), –240.2 (ddd, 1F, 2JHF= 47.2 Hz, 2JHF= 45.4 Hz, 3JHF= 29.0 Hz, 6) 4-Fluoromethyl-4-fluoro-1,3-dioxolan-2-one 1 H NMR (400 MHz, CD3CN): δ 4.78 (ddd, 2H, 2JHF= 45 Hz, 3JHF= 10 Hz, 4JHH= 2.5 Hz, 6), 4.63-4.72 (m, 2H, 5) 13 C NMR (100.6 MHz, CD3CN): δ 70.5 (5), 79.5 (6), 112.2 (4), 151.7 (2); 111 19 F NMR (376.5 MHz, CD3CN): δ –112.79- –113.00 (m, 1F, 4), –236.36- 236.66 (m, 1F, 6) 6.6.6 Fluorination of closo-K2[B12H12]: Minireactor 0.0907 g (0.44 mmol) well dried closo-K2[B12H12] was dissolved in 3.0 g EC and 0.9 g PC. First the nitrogen flow of 400 mL h–1 was started and the mixture was filled into the separator vessel. The circulation pump was started at a low flow of 30 mL h–1. When stable gas slugs were observed at the outlet, the liquid flow was slowly increased to 300 mL h–1. The reaction gas flow was set to a total gas flow of 400 mL h-1. During the reaction, the fluorine concentration was increased from 22 % to 45 %. The average fluorine concentration was 33 %. After 5.6 h the fluorine was switched of and the reactor purged with nitrogen. After quenching the HF with calcium carbonate, the standard was added and a NMR sample prepared. Batch 4.99 g Na2[B12H12] (26.6 mmol) were placed in a 250 mL three-necked flask and dissolved in 285 mL acetonitrile. The flask was connected to the nitrogen and fluorine gas flow controller. 9.49 h NaF (225.9 mmol) and 6 mL water were added. The mixture was cooled to 0 °C. Then 5.4 L fluorine (220 mmol, 8.3 equiv., 15 mL min-1) diluted by 21.6 L nitrogen (60 mL min-1 ) were bubbled into the solution over 6 h. After completion of the reaction, the flask was purged with nitrogen for 60 min. The mixture was filtered, the solvent removed under reduced pressure and the solid dried over night. For a second fluorination step the dried solid was dissolved in 300 mL dry acetonitrile and 11.79 g NaF (280.7 mmol) were added. The mixture was cooled to 0 °C. Over two hour 1.8 L of fluorine (74 mmol, 2.8 equiv.) were bubbled into the mixture. The used flows were 55 mL min-1 nitrogen and 15 mL min-1 fluorine. The reaction mixture was neutralized with NaHCO3 solution and filtered. The solvent was removed under reduced pressure. To the brownish solid 25 mL of a 6 % hydrogen peroxide solution was added. It was then heated for 3 h at 80 °C, which made the colour change to a slight yellow. Further 20 mL of the hydrogen peroxide solution were added and heated for further 4 h. The water was removed by heating it to dryness. 11 B NMR (CD3CN, 298 K): δ = -1.2 (s) [BF4]-, -16.8 (s, br, weak, [B12F12-xHx]-2), -18.0 (s.) [B12F12]-2 112 19 F NMR (CD3CN, 298K): δ = -152.0 (s) [BF3], -152.1 (s, [BF4]-), -269.5 (s, br, [B12F12]-2) [126] 6.6.7 Fluorination of tetra-n-Butylammonium Tetrafluoroborate Synthesis of tetra-n-Butylammonium Tetrafluoroborate 19.45 g (0.061 mol) [N(nBu)4]Br was dissolved in 90 mL water. The solution was cooled to 0 °C. 10.7 mL (0.072 mol) HBF4 (50 % in water) were added drop wise to the stirred solution over 25 min. After the addition, the mixture heated to room temperature and stirred for further 40 min. The white solid was filtered and washed five times with water (30 mL) and two times with diethyl ether. It was then dried under reduced pressure. Fluorination of tetra-n-Butylamine Tetrafluoroborate 2.02 g (6.1 mmol) [N(nBu)4]BF4 were added to 2.00 g (22.7 mmol) EC and were heated to 40 °C. A 4 mL min-1 nitrogen flow was set for the minireactor #2 and the solution was started flow with a speed of 150 mL h-1. The temperature of the reactor was set to 40 °C. For a fluorination time of 8.5 h a fluorine flow of 2 mL min-1 was used. After the reaction and purging with nitrogen over night, the solution was transferred into a PFA flask. The hydrogen fluoride was removed using first vacuum and then silica gel. After filtration NMR spectra were measured. 113 7 Literature [1] P. Kirsch, Modern Fluoroorganic Chemistry, Wiley-VCH, Weinheim, 2004. [2] D. O'Hagan, J. Fluorine Chem. 2010, 131, 1071. [3] H.-J. Böhm, D. Banner, S. Bendels, M. Kansy, B. Kuhn, K. Müller, U. Obst-Sander, M. Stahl, ChemBioChem 2004, 5, 637. [4] M. Hird, Chem. Soc. Rev. 2007, 36, 2070. [5] J. M. Slattery, A. Higelin, T. Bayer, I. Krossing, Angewandte Chemie International Edition 2010, 49, 3228. [6] R. McMillan, H. Slegr, Z. X. Shu, W. Wang, J. Power Sources 1999, 81-82, 20. [7] F. Diederich, K. Mueller, C. Faeh, Science 2007, 317, 1881. [8] F. Drakesmith, R. Chambers, Electrofluorination of Organic Compounds, Organofluorine Chemistry, Vol. 193, Springer Berlin / Heidelberg, 1997. [9] L. Conte, G. Gambaretto, J. Fluorine Chem. 2004, 125, 139. [10] N. V. Ignat'ev, H. Willner, P. Sartori, J. Fluorine Chem. 2009, 130, 1183. [11] J. L. Adcock, R. J. Lagow, J. Am. Chem. Soc. 1974, 96, 7588. [12] S. Rozen, Acc. Chem. Res. 1988, 21, 307. [13] G. Sandford, J. Fluorine Chem. 2007, 128, 90. [14] G. Sandford, J. Hutchinson, Elemental Fluorine in Organic Chemistry: Topics in Current Chemistry, Vol. 193, Oxford University Press, Inc. + The American Chemical Society, Berlin, Heidelberg, 2006. [15] Byung Won Woo, Seung Woo Yoon, Junh Ho Lee, Soon Hong Park, Nak Joon Jang, H. J. Yoon, in U.S. Pat. Appl.: US 2006/0167279 A1, Vol. US 2006/0167279 A1, Ulsan Chemical co., Ltd., 2006. [16] P. Watts, C. Wiles, Chemical Communications 2007, 443. [17] V. Hessel, D. Kralisch, U. Krtschil, Energy & Environmental Science 2008, 1, 467. [18] F. Jensen Klavs, in Microreactor Technology and Process Intensification, Vol. 914, American Chemical Society, 2005, pp. 2. [19] A. Gunther, K. F. Jensen, Lab on a Chip 2006, 6, 1487. [20] R. D. Chambers, T. Nakano, T. Okazoe, G. Sandford, J. Fluorine Chem. 2009, 130, 792. [21] R. D. Chambers, M. A. Fox, G. Sandford, J. Trmcic, A. Goeta, J. Fluorine Chem. 2007, 128, 29. 114 [22] R. D. Chambers, T. Okazoe, G. Sandford, E. Thomas, J. Trmcic, J. Fluorine Chem. 2010, 131, 933. [23] K. Jähnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Löwe, C. Wille, A. Guber, J. Fluorine Chem. 2000, 105, 117. [24] N. de Mas, A. Guenther, M. A. Schmidt, K. F. Jensen, Industrial & Engineering Chemistry Research 2003, 42, 698. [25] A. F. Holleman, E. Wiberg, N. Wiberg, G. Fischer, Lehrbuch der Anorganischen Chemie, 102 ed., Walter de Gruyter, Berlin/New York, 2007. [26] M.-C. Lasne, C. Perrio, J. Rouden, L. Barré, D. Roeda, F. Dolle, C. Crouzel, W. Krause, Vol. 222, Springer Berlin / Heidelberg, 2002, pp. 201. [27] J. E. Huheey, E. A. Keiter, R. L. Keiter, Vol. 3, de Gruyter, New York, 1993. [28] N. N. Greenwood, A. Earnshaw, Chemie der Elemente, VCH, Weinheim, 1988. [29] C. E. Housecroft, A. G. Sharpe, A. Rompel, Anoganische Chemie, 2006. [30] T. Jackson, The Elements - Fluorine, Benchmark Books, New York, 2004. [31] W. D. Armstrong, P. J. Bekhus, Jounal of Dental Research 1938, 17, 393. [32] T. Aoba, Crit Rev Oral Biol Med 1997, 8, 136. [33] R. O'Dell, R. A. Sunde, Handook of Nutritionally Essential Mineral Elements, Marcel Dekker, Inc. , New York, 1997. [34] J. Schmedt auf der Günne, M. Mangstl, F. Kraus, Angewandte Chemie International Edition 2012, 51, 7847. [35] A. Tressaud, Angewandte Chemie International Edition 2006, 45, 6792. [36] K. O. Christe, Inorg. Chem. 1986, 25, 3721. [37] I. Krossing, I. Raabe, Angewandte Chemie International Edition 2004, 43, 2066. [38] B. H. Minor, D. Herrmann, R. Gravell, Process Safety Progress 2010, 29, 150. [39] A. Kharitonov, B. Loginov, Russian Journal of General Chemistry 2009, 79, 635. [40] T. Köchner, T. A. Engesser, H. Scherer, D. A. Plattner, A. Steffani, I. Krossing, Angewandte Chemie International Edition 2012, 51, 6529. [41] J. B. Goodenough, Y. Kim, Chem. Mater. 2009, 22, 587. [42] http://www.drugs.com/stats/top100/2012/sales, U.S. Pharmaceutical Sales - 2012, 2012. [43] G. Theodoridis, T. Alain, in Advances in Fluorine Science, Vol. Volume 2, Elsevier, 2006, pp. 121. [44] J. Flahaut, J. Fluorine Chem. 2004, 27. 115 [45] C. E. Houscroft, A. G. Sharpe, Inorganic Chemisty, 2. ed., Harlow, 2005. [46] B. d. Darwent, Bond Dissociation Energies in Simple Molecules, Vol. 31, United States Department of Commerce, 1970. [47] A. Dale, C. Lythall, J. Aucott, C. Sayer, Thermochim. Acta 2002, 382, 47. [48] R. J. Lagow, J. L. Margrave, Prog. Ing. Chem. 1986, 161. [49] C. Rüchardt, H.-D. Beckhaus, Angewandte Chemie International Edition in English 1980, 19, 429. [50] D. H. R. Barton, R. H. Hesse, R. E. Markwell, M. M. Pechet, S. Rozen, J. Am. Chem. Soc. 1976, 98, 3036. [51] A. L. D. W. T. Miller, J. Am. Chem. Soc 1956, 2793. [52] P. S. Frederiks, J. Tedder, J. Chem. Soc. 1960, 144. [53] H. Fukaya, K. Morokuma, The Journal of Organic Chemistry 2003, 68, 8170. [54] M. L. C. J. L. Adcock, Ind. Eng. Chem. Res. 1987, 26, 208. [55] R. J. L. L. A. Shimp, J. Org. Chem. 1977, 42, 3437. [56] N. J. Maraschin, R. L. Lagow, J. Am. Chem. Soc. 1972, 94, 8601. [57] T. R. Bierschenk, T. J. Juhlke, H. Kawa, R. J. Lagow, Vol. US 258,708 (Ed.: U. S. Patent), Exfluor Research Corporation, 1994. [58] T. Okazoe, K. Watanabe, M. Itoh, D. Shirakawa, H. Takagi, K. Kawahara, S. Tatematsu, Bull. Chem. Soc. Jpn. 2007, 80, 1611. [59] R. D. Chambers, R. C. H. Spink, Chemical Communications 1999, 883. [60] T. Furuya, C. A. Kuttruff, T. Ritter, Current Opinion in Drug Discovery & Development 2008, 11, 803. [61] P. Kirsch, M. Lenges, A. Ruhl, F. Huber, R. D. Chambers, G. Sandford, J. Fluorine Chem. 2007, 128, 1221. [62] D. D. Moldavskii, T. A. Bispen, G. I. Kaurova, G. G. Furin, J. Fluorine Chem. 1999, 94, 157. [63] M. Noel, V. Suryanarayanan, S. Chellammal, J. Fluorine Chem. 1997, 83, 31. [64] N. V. Ignat'ev, U. Welz-Biermann, U. Heider, A. Kucheryna, S. von Ahsen, W. Habel, P. Sartori, H. Willner, J. Fluorine Chem. 2003, 124, 21. [65] N. Kockmann, in Advanced Micro & Nanosystems, Vol. 5, Wiley-VCH, 2006. [66] J. Ducrée, P. Koltay, R. Zengerle, Microfludics, Springer-Verlag, Berlin, 2005. [67] K. Jähnisch, V. Hessel, H. Löwe, M. Baerns, Angewandte Chemie International Edition 2004, 43, 406. 116 [68] H. Löwe, W. Ehrfeld, Electrochim. Acta 1999, 44, 3679. [69] T. Wirth, Microreactors in Organic Synthesis and Catalysis, Wiley-VCH, Weinheim, 2008. [70] N. Kim, E. Evans, D. Park, S. Soper, M. Murphy, D. Nikitopoulos, Experiments in Fluids 2011, 51, 373. [71] J. R. Burns, C. Ramshaw, Lab on a Chip 2001, 1, 10. [72] P. Sobieszuk, J. Aubin, R. Pohorecki, Chemical Engineering & Technology, 35, 1346. [73] P. Lang, M. Hill, I. Krossing, P. Woias, Chemical Engineering Journal 2012, 179, 330. [74] M. T. Kreutzer, F. Kapteijn, J. A. Moulijn, J. J. Heiszwolf, Chem. Eng. Sci. 2005, 60, 5895. [75] P. Lang, Albert-Ludwigs-Universität (Freiburg i. Br. ), 2012. [76] V. Hessel, S. Hardt, H. Löwe, Chemical Micro Process Enginieering, Wiley-VCH, Weinheim, 2004. [77] A. Jahn, J. Reiner, W. Vreeland, D. DeVoe, L. Locascio, M. Gaitan, Journal of Nanoparticle Research 2008, 10, 925. [78] S. J. Haswell, R. J. Middleton, B. O'Sullivan, V. Skelton, P. Watts, P. Styring, Chemical Communications 2001, 391. [79] J. Wiley, Processing and Finishing of Polymeric Materials, Vol. 2, John Wiley & Sons, Inc., New Jersey, 2011. [80] S. Franssila, Introduction Into Micro Fabrication, 2 ed., John Wiley & Sons Ltd., Chichester, 2010. [81] M. Elwenspoekl, U. Lindberg, H. Kokl, L. Smith, IEEE 1994, 223. [82] J. A. McGeough, M. C. Leu, K. P. Rajurkar, A. K. M. De Silva, Q. Liu, CIRP Annals Manufacturing Technology 2001, 50, 499. [83] C. Jeggy, Micro-Injection Moulding, UCL Presses Universitaires De Louvain, Louvain, 2004. [84] S. Kang, Micro/Nano Repication (Processes and Applicaitons), Wiley, New Jersey, 2012. [85] M. Worgull, Hot Embossing, William Andrew, 2009. [86] B. Stefan Schlautmann and Henk Wensink and Richard Schasfoort and Miko Elwenspoek and Albert van den, Vol. 11, 2001, p. 386. 117 [87] N. Kockmann, Micro Process Engineering (Advanced Micro and Nanosystems), John Wiley & Sons, Weinheim, 2008. [88] Corrosion Resitance of NICKEL- CONTAINING ALLOYS in HYDROFLUORIC ACID, HYDROGEN FLUORIDE and FLUORINE, Vol. 443, NiDI. [89] K. J. W. Atkinson, Imperial College of Science, Technology and Medicine (London), 2002. [90] H. M. Haendler, W. L. Patterson, W. J. Bernard, J. Am. Chem. Soc. 1952, 74, 3167. [91] K. Hauffe, Z. Werkstofftech 1984, 15, 427. [92] B. J. Garrison, W. A. Goddard, Phys. Rev. B: Condens. Matter 1987, 36, 9805. [93] R. A. a. N. d. M. a. R. S. a. A. J. F. a. M. A. S. a. K. F. J. Leonel, Journal of Micromechanics and Microengineering 2007, 17, 384. [94] J. Kobayashi, Y. Mori, K. Okamoto, R. Akiyama, M. Ueno, T. Kitamori, S. Kobayashi, Vol. 304, 2004, pp. 1305. [95] L. Conte, G. P. Gambaretto, M. Napoli, C. Fraccaro, E. Legnaro, J. Fluorine Chem. 1995, 70, 175. [96] R. D. Chambers, M. P. Greenhall, J. Hutchinson, Tetrahedron 1996, 52, 1. [97] C. Wiles, P. Watts, Chemical Communications 2011, 47, 6512. [98] M. Brivio, R. H. Fokkens, W. Verboom, D. N. Reinhoudt, N. R. Tas, M. Goedbloed, A. van den Berg, Anal. Chem. 2002, 74, 3972. [99] M. E. Kopach, M. M. Murray, T. M. Braden, M. E. Kobierski, O. L. Williams, Organic Process Research & Development 2009, 13, 152. [100] D. M. Ratner, E. R. Murphy, M. Jhunjhunwala, D. A. Snyder, K. F. Jensen, P. H. Seeberger, Chemical Communications 2005, 578. [101] T. Dietrich, A. Freitag, U. Schlecht, Chemical Engineering Journal 2010, 160, 823. [102] R. Böhling, A. Drochner, M. Fehlings, D. König, H. Vogel, Chemie Ingenieur Technik 1999, 71, 226. [103] Y. Tanaka, M. Terao, T. Akutsu, K. Isozaki, Yokogawa Technical Report English Edition 2009, 47, 37. [104] Andreas Freitag, Thomas R. Dietrich, Ralf Scholz, S. Link, SENSOR MAGAZIN 2009, 2, 62. [105] M. Baumann, I. R. Baxendale, L. J. Martin, S. V. Ley, Tetrahedron 2009, 65, 6611. [106] T. Gustafsson, R. Gilmour, P. H. Seeberger, Chemical Communications 2008, 3022. 118 [107] R. D. Chambers, D. Holling, R. C. H. Spink, G. Sandford, Lab on a Chip 2001, 1, 132. [108] R. D. Chambers, W. K. Gray, G. Sandford, J. F. S. Vaughan, J. Fluorine Chem. 1999, 94, 213. [109] R. D. Chambers, D. Holling, A. J. Rees, G. Sandford, J. Fluorine Chem. 2003, 119, 81. [110] R. D. Chambers, J. Hutchinson, J. Thomson, J. Fluorine Chem. 1996, 78, 165. [111] R. D. Chambers, A. M. Kenwright, M. Parsons, G. Sandford, J. S. Moilliet, Journal of the Chemical Society, Perkin Transactions 1 2002, 2190. [112] R. D. Chambers, M. Parsons, G. Sandford, E. Thomas, J. Trmcic, J. S. Moilliet, Tetrahedron 2006, 62, 7162. [113] R. D. Chambers, M. A. Fox, D. Holling, T. Nakano, T. Okazoe, G. Sandford, Chemical Engineering & Technology 2005, 28, 344. [114] R. D. Chambers, M. A. Fox, G. Sandford, Lab on a Chip 2005, 5, 1132. [115] C. B. McPake, G. Sandford, Organic Process Research & Development 2012, 16, 844. [116] O. Dunkelberg, A. Haas, M. F. Klapdor, D. Mootz, W. Poll, E. H. Appelman, Chem. Ber. 1994, 127, 1871. [117] R. D. Chambers, J. Hutchinson, G. Sandford, A. Shah, J. F. S. Vaughan, Tetrahedron 1997, 53, 15833. [118] J. Frühauf, B. Hannemann, Journal of Micromechanics and Microengineering 1997, 7, 137. [119] N. de Mas, A. Guenther, M. A. Schmidt, K. F. Jensen, Industrial & Engineering Chemistry Research 2008, 48, 1428. [120] MKS, PR 400B-F, Vol. Y-19640003, MKS Instruments Deutschland GmbH, 2008. [121] MKS, MKS Type 1479A Mass-Flo® Controller, Vol. 119474-P1, 1997. [122] P. Bendix, Universtität Freiburg (Freiburg i.Br..), 2011. [123] P. Aussillous, D. Quere, Physics of Fluids 2000, 12, 2367. [124] M. M. Awad, in An Overview of Heat Transfer Phenomena (Ed.: S. N. Kazi), INTECH, Online, 2012, pp. 251. [125] W. Li, Z. Wu, Experimental Thermal and Fluid Science 2011, 35, 866. [126] M. Rühle, University of Freiburg (Freiburg i. Br.), 2010. [127] J. Burdon, J. R. Knights, I. W. Parsons, J. C. Tatlow, Journal of the Chemical Society, Perkin Transactions 1 1976, 0, 1930. [128] M. Haruta, N. Watanabe, J. Fluorine Chem. 1976, 7, 159. 119 [129] V. Grakauskas, The Journal of Organic Chemistry 1970, 35, 723. [130] S. Di Cave, B. Mazzarotta, Journal of Chemical & Engineering Data 1991, 36, 293. [131] C. V. V. Ramana, A. B. V. K. Kumar, M. A. Kumar, M. K. Moodley, Journal of Chemistry 2013, 2013, 4. [132] P. Löb, H. Löwe, V. Hessel, J. Fluorine Chem. 2004, 125, 1677. [133] M. T. Rogers, J. L. Burdett, Can. J. Chem. 1965, 43, 1516. [134] O. Böse, D. Seffer, K. Peterkord, in Deutsches Patent- und Markenamt: DE10308149A1, Vol. DE10308149A1, Solvay Fluor, Germany, 2004. [135] M. Kobayashi, T. Inoguchi, T. Iida, T. Tanioka, H. Kumase, Y. Fukai, J. Fluorine Chem. 2003, 120, 105. [136] M. Takehara, S. Watanabe, N. Nanbu, M. Ue, Y. Sasaki, Synth. Commun. 2004, 34, 1367. [137] N. Nanbu, K. Takimoto, K. Suzuki, M. Ohtake, K. Hagiyama, M. Takehara, M. Ue, Y. Sasaki, Chem. Lett. 2008, 37, 476. [138] D. V. Peryshkov, A. A. Popov, S. H. Strauss, J. Am. Chem. Soc. 2009, 131, 18393. [139] Vaughey John, Jansen Andrew N, Dees Dennis W, in US 2010040955, Vol. US 2010040955, US 2010040955 ed., 2010. [140] A. Dimitrov, W. Radeck, S. Rädiger, O. Bechstein, J. Fluorine Chem. 1993, 60, 57. [141] A. Dimitrov, D. Pfeifer, U. Jonethal, S. Rädiger, K. Seppelt, J. Fluorine Chem. 1997, 82, 143. [142] B. Aakermark, G. Aakermark, L. S. Hegedus, K. Zetterberg, J. Am. Chem. Soc. 1981, 103, 3037. [143] M. E. Hirschberg, N. V. Ignat'ev, A. Wenda, H. Willner, J. Fluorine Chem. 2012, 137, 50. [144] H. Xue, J. n. M. Shreeve, European Journal of Inorganic Chemistry 2005, 2005, 2573. [145] P. Barthazy, A. Togni, A. Mezzetti, Organometallics 2001, 20, 3472. [146] H. Koroniak, J. Walkowiak, K. Grys, A. Rajchel, A. Alty, R. Du Boisson, J. Fluorine Chem. 2006, 127, 1245. [147] R. D. Chambers, M. A. Fox, D. Holling, T. Nakano, T. Okazoe, G. Sandford, Lab on a Chip 2005, 5, 191. [148] L. A. Curtiss, K. Raghavach Ari, P. C. Redfern, V. Rassolov, J. A. Pople, J. Chem. Phys. 1998, 109, 7764. 120 [149] G. W. T. M. J. Frisch, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Revision C.02, Gaussian Inc., Wallingford CT, 2004. [150] L. B. Asprey, J. Fluorine Chem. 1976, 7, 359. [151] V. Geis, K. Guttsche, C. Knapp, H. Scherer, R. Uzun, Dalton Transactions 2009, 15, 2687. [152] G. Schatte, J. Passmore, S. Brownridge, X. Sun, T. Way, D. Wood, Guide For Fluorine Chemists, 2 ed., New Brunswick, 1995. [153] Airgas, Material Safety Data Sheet: Fluorine, Radnor, 2012. [154] E. M. Caravati, The American Journal of Emergency Medicine 1988, 6, 143. [155] HoneywellHydrofluoricAcid, Recommended Medical Treatment for Hydrofluoric Acid Exposure, Honeywell Specialty Materials, 2006. [156] G. Butt, J. Cilmi, P. M. Hoobin, R. D. Topsom, Spectrochimica Acta Part A: Molecular Spectroscopy 1980, 36, 521. [157] P. Barthazy, L. Hintermann, R. M. Stoop, M. Wörle, A. Mezzetti, A. Togni, Helv. Chim. Acta 1999, 82, 2448. [158] H. Ishii, N. Yamada, T. Fuchigami, Tetrahedron 2001, 57, 9067. 121 8 Danksagung An dieser Stelle möchte ich eine Reihe von wunderbaren Menschen danken, die mich während meiner Doktorarbeitszeit in Freiburg begleitet und unterstützt haben. ♦ Professor Dr. Ingo Krossing für die Möglichkeit diesem spannenden Thema zu Arbeiten und der mit seiner positiven und optimistischen Art immer wieder aufs Neue motivieren konnte. Zudem möchte ich ihn dafür danken, dass er mir die Teilnahme an einer Reihe von spannenden Konferenzen im In- und Ausland ermöglicht hat. ♦ Professor Dr. Carsten Knapp und Dr. Sebastian Riedel für all die Unterstützung und all das was ich von Ihnen über die Fluorchemie lernen durfte. ♦ Professor Dr. Peter Woias, Dr. Philipp Lang und Dr. Keith Cobry für die gute und fruchtbare Zusammenarbeit. ♦ Allen im Institut für Anorganische und Analytische Chemie für die angenehme und hilfsbereite Atmosphäre. Mein besonderer Dank gilt hierbei Maribel, Sascha und Mahdi für ihre Freundschaft und die schöne Zeit im Raum 137. ♦ Meinen Freunden, die nicht hier Wohnen, aber mich immer fleißig Besucht haben. ♦ Pia, für all die Kraft und Hilfe mit der Sie mich in den letzen Jahren immer Unterstützt hat. Sie hat immer die richtigen Worte gefunden um mich wieder aufzubauen, wenn mal etwas nicht geklappt hat. ♦ Und abschließend noch meinen Tanten, meinen Geschwistern und ganz besonders meinen Eltern für den Rückhalt den ich während meines ganzen Leben von Ihnen erhalten habe. Danke euch Allen!!! 122