The present invention relates to processes for purifying a tetrafluoropropene, particularly 1,3,3,3-tetrafluoropropene and 2,3,3,3-tetrafluoropropene.

1,3,3,3-Tetrafluoropropene is also known as HFO-1234ze, HFC-1234ze, R-1234ze or simply 1234ze, and 2,3,3,3-tetrafluoropropene is also known as HFO-1234yf, HFC- 1234yf, R-1234yf or simply 1234yf. Hereinafter, unless otherwise stated, 1 ,3,3,3- tetrafluoropropene will be referred to as R-1234ze and 2,3,3,3-tetrafluoropropene will be referred to as R-1234yf. The hydrofluorocarbon R-1234ze exists in E and Z isomeric forms, each of which will be referred to as R-1234ze(E) and R-1234ze(Z), respectively, where appropriate. The term R-1234 is used to designate R-1234ze or R-1234yf, or mixtures thereof. (Hydro)fluoroalkenes are increasingly being considered as working fluids in applications such as refrigeration, heat pumping, foam blowing, fire extinguishers/retardants, propellants and solvency (e.g. plasma cleaning and etching). The processes used to make (hydro)fluoroalkenes can lead to the generation of toxic and/or otherwise undesirable by-products. The dehydrohalogenation of some (hydro)fluoroalkanes is described in International patent application no. WO 2009/137656 (which is incorporated herein by reference).

The presence of small quantities of impurities may not be detrimental to the bulk physical properties of the (hydro)fluoroalkene product and for some applications their removal is unnecessary. However, some applications require very low levels of impurities and/or the presence (or absence) of certain physical properties.

The listing or discussion of a prior-published document in this specification should not necessarily be taken as an acknowledgement that the document is part of the state of the art or is common general knowledge.

Trifluoromethylacetylene (TFMA) is a hydrofluorocarbon that may be produced as a byproduct of the synthesis of R-1234, ie R-1234ze and/or R-1234yf and other similar substances. The presence of small amounts of TFMA (e.g. at approximately 1000 ppm w/w or above) in R-1234, for example in R-1234ze, can result in the product being classified as flammable at a test temperature of 23°C using the ASHRAE-34 methodology. This is unexpected as R-1234 compositions such as R-1234ze l compositions containing highly flammable species such as propylene at substantially higher concentrations (e.g. in excess of 2%) may still be classified as non-flammable at a test temperature of 23°C using the ASHRAE-34 methodology. The applicant has found that the processes described herein may allow the amount of TFMA in mixtures containing TFMA and R-1234, such as R-1234ze to be reduced so that the resulting mixture is no longer classified as flammable at 23°C.

The present invention provides a process for reducing the flammability of R-1234, particularly R-1234ze, or a composition comprising R-1234, particularly R-1234ze, which process comprises reducing the amount of TFMA present in the R-1234 or a composition comprising R-1234.

TFMA and R-1234 may be separated from a mixture comprising those components using techniques such as distillative separation, adsorption and/or membrane separation.

The distillative separation of TFMA from R-1234, particularly R-1234ze, more particularly R-1234ze(E) may be performed at a range of pressures, e.g. from about 1 to about 20 bar. Continuous or semi-continuous fractional distillation of TFMA from R-1234ze(E) may be effective in allowing separation of TFMA from R-1234ze(E). Fractional distillation involves distillation using column having a plurality of separation stages, equipped with a reboiler and partial or total condenser, having its feed material fed as liquid or vapour (or a two-phase mixture) to an intermediate point on the column and with withdrawal of liquid bottom product of the purified R-1234ze(E) and a vapour or liquid top product enriched in TFMA.

Preferably the distillation is effected at a pressure greater than atmospheric pressure, so that the top distillate may be condensed at a temperature of greater than about -30°C, preferably greater than -10°C, even more preferably so that it may be condensed against cooling water (at about 20-25°C). The operating pressure of the column should be less than 20 bara, preferably less than 15 bara, more preferably less than 10 bara.

The distillative separation may be performed in a column which is at least 10 theoretical stages tall, preferably 20 theoretical stages tall. The concept of a theoretical distillation stage is explained in the standard industry reference text "Distillation Design" by Henry Z Kister (published McGraw-Hill 1992) (which is incorporated herein by reference). The ideal (theoretical) distillation stage is a device that meets the following criteria:

1. It operates in steady state and has a liquid product and a vapour product.

2. All vapour and liquid entering the stage are intimately contacted and perfectly mixed.

3. Total vapour leaving the stage is in equilibrium with total liquid leaving the stage.

By describing a distillation column as being a certain number of theoretical stages tall, we include the meaning that the separation which can be achieved using the column is equivalent to the separation achieved using that number of individual distillation stages.

The boilup rate required will be from about 1 mol (per mol fed) to about 5 mol/mol fed, preferably about 1 to about 4 mol/mol fed.

The distillation may result in an overhead product rate of from about 1 to about 4 moles of vapour per mole of material fed to the column, with the boilup rate required being a function of operating pressure. If the feed material enters as a liquid and the top and bottom products are removed as liquid, then the heat load on the column reboiler and cooling load on the overhead reflux condenser are both approximately proportional to this overhead product rate. The optimal combination of column pressure and operating conditions for minimisation of energy costs will depend on the local costs of heating and refrigeration. In order to reduce the operating costs of the column it may be possible to operate the separation with the aid of a vapour recompression heat pump cycle to transfer heat between condenser and reboiler. This is especially attractive if a relatively low TFMA content is selected in the overhead product, in which case the temperature difference between top and bottom of the column will be reduced. For example if 10% TFMA is selected as the purity of the top product then the temperature at the top of the column will be about 1 K colder than the temperature at the bottom of the column. The heat pump cycle may use the refrigerant vapour from the top of the column directly as refrigerant, or it may use an external refrigeration circuit whose evaporator functions as the still condenser and whose condenser provides heat to the column reboiler. This latter solution is attractive if a low operating pressure is desired, since it allows use of a high pressure refrigerant fluid such as R-32 or R-410A and hence may result in a lower size and cost of gas compressor. The column may be optimally operated under conditions of total or near total reflux, with intermittent withdrawal of vapour product from the top of the column to purge TFMA from the system. Accordingly, a preferred aspect of the invention includes such operation under total reflux - where the reflux rate is at least enough to give the desired boilup rate for effective separation. This may be coupled with intermittent vapour withdrawal from the top of the column, where rate and duration is controlled by monitoring pressure at the top of the column, either via on-off control or a proportional control scheme with reset. Alternative methods that may be used to separate TFMA from a mixture containing TFMA and R-1234 include using membrane separation methods, for example micro- filtration using membranes with sub-micron pore sizes which are capable of differentiating between TFMA and R-1234. The membrane used is preferably a hollow pore type membrane or rubbery polymer membrane. If the former then a membrane whose effective pore diameter is such that the rate of diffusion of TFMA along the membrane is usefully higher than that of R-1234ze(E) is preferred. In the latter case it is preferable that the rubber polymer membrane be selected so that the TFMA is more soluble in the membrane material than is the R-1234

Separation of TFMA from a mixture comprising R-1234, for example R-1234ze, and TFMA may also be accomplished by adsorption techniques, such as pressure swing or temperature swing cyclic separation cycles. In such processes, the TFMA is the more strongly adsorbed species and can therefore be removed effectively from the R-1234 by passing the material to be purified over an adsorbent, and generating a stream with low or essentially no TFMA content. The regeneration of the adsorbent by purging with inert gas, temperature or pressure cycling then contains an enriched TFMA/R-1234ze mixture. This may be recycled to a reaction step for reconversion to R-1234ze or otherwise disposed of.

Such adsorptive separation processes to separate TFMA from a mixture containing TFMA and R-1234 include those using an aluminium-containing absorbent, activated carbon, or a mixture thereof.

A preferred aluminium-containing adsorbent for use in processes according to the invention is an alumina or alumina-containing substrate. Advantageously, the substrate is porous. Further information on the various crystalline forms of alumina can be found in Acta. Cryst., 1991, B47, 617, the contents of which are hereby incorporated by reference. Preferred aluminium-containing adsorbents (e.g. alumina) for use according to the invention will have functionality that facilitates their combination with the compounds the adsorbent is removing. Examples of such functionality include acidity or basicity, which can be Lewis-type or Bronsted-type in nature, which will facilitate its combination with the compounds the adsorbent is removing. The acidity or basicity can be modified in a manner well known to those skilled in the art by using modifiers such as sodium sulphate. Examples of aluminium-containing adsorbents with acidic or basic functionality include Eta-alumina, which is acidic, and Alumina AL0 04, which is basic.

Aluminosilicate molecular sieves (zeolites) are a further preferred group of aluminium- containing adsorbent that may be used in the subject invention. Typically, the zeolites have pores having openings which are sufficiently large to allow the desired and undesired compounds to enter into the interior of the zeolite whereby the undesired compounds are retained. Accordingly, zeolites having pores which have openings which have a size across their largest dimension in the range of 3A to 12A are preferred.

Preferred zeolites have a pore opening sufficiently large to allow the undesired compounds to enter into the interior of the zeolite whereby the undesired compounds are retained, whilst excluding the desired compound from entering the interior of the zeolite. Such zeolites typically have openings which have a size across their largest dimension in the range of 3A to 12A, preferably from 3A to 10A or 4A to 12A. Particularly preferred are those molecular sieves having pores which have openings having a size across their largest dimension in the range of 4A to 10A, such as 4A to 8A (e.g. 4A to 5A) and may include zeolite Y, ultra-stable Y (dealuminated-Y), zeolite beta, zeolite X, zeolite A and zeolite ZS -5, AW-500.

By opening in this context we are referring to the mouth of the pore by which the undesired compound enters the body of the pore, where it may be retained. The openings to the pores may be elliptically shaped, essentially circular or even irregularly shaped, but will generally be elliptically shaped or essentially circular. When the pore openings are essentially circular, they should have a diameter in the range of about 3A across their smaller dimension. They can still be effective at adsorbing compounds provided that the size of the openings across their largest dimension is in the range of from about 3A to about 12A. Where the adsorbent has pores having elliptically shaped openings, which are below 3A across their smaller dimension, they can still be effective at adsorbing compounds provided that the size of the openings across their largest dimension is in the range of from about 3A to about 12A.

By "activated carbon", we include any carbon with a relatively high surface area such as from about 50 to about 3000 m ² or from about 100 to about 2000 m ² (e.g. from about 200 to about 1500 m ² or about 300 to about 1000 m ² ). The activated carbon may be derived from any carbonaceous material, such as coal (e.g. charcoal), nutshells (e.g. coconut) and wood. Any form of activated carbon may be used, such as powdered, granulated, extruded and pelleted activated carbon.

Activated carbon is preferred which has been modified (e.g. impregnated) by additives which modify the functionality of the activated carbon and facilitate its combination with the compounds it is desired to removed. Examples of suitable additives include metals or metal compounds, and bases.

Typical metals include transition, alkali or alkaline earth metals, or salts thereof. Examples of suitable metals include Na, K, Cr, Mn, Au, Fe, Cu, Zn, Sn, Ta, Ti, Sb, Al, Co, Ni, Mo, Ru, Rh, Pd and/or Pt and/or a compound (e.g. a halide, hydroxide, carbonate) of one or more of these metals. Alkali metal (e.g. Na or K) salts are currently a preferred group of additive for the activated carbon, such as halide, hydroxide or carbonate salts of alkali metals salts. Hydroxide or carbonate salts of alkali metals salts are bases. Any other suitable bases can be used, including amides (e.g. sodium amide).

The impregnated activated carbon can be prepared by any means known in the art, for example soaking the carbon in a solution of the desired salt or salts and evaporating the solvent. Examples of suitable commercially available activated carbons include those available from Chemviron Carbon, such as Carbon 207C, Carbon ST1X, Carbon 209M and Carbon 207EA. Carbon ST1X is currently preferred. However, any activated carbon may be used with the invention, provided they are treated and used as described herein. Preferably, the R-1234 is predominantly R-1234ze, i.e. R-1234yf is substantially absent, e.g. the ratio of the amount of R-1234ze to the amount of R-1234yf is at least about 10:1 , preferably least about 100:1 or more preferably least about 1000:1. Preferably, the R-1234ze is predominantly in the form of the E-isomer, i.e. the Z-isomer is substantially absent, i.e. the ratio of the amount of the E-isomer to the amount of the Z- isomer is at least about 10:1, preferably least about 100:1 or more preferably least about 1000:1.

The process of the invention may produce R-1234, for example R-1234ze, with a TFMA content of less than about 500 ppm w/w, for example less than about 200 ppm w/w, preferably less than about 100 ppm w/w, and more preferably less than about 40 ppm w/w, for example 30 ppm w/w or less.

TFMA as obtained by the processes of the invention may further be used to generate further R-1234ze, e.g. by reaction with HF under suitable conditions. This allows for economic recycling of the TFMA by-product of the synthesis of R-1234ze. TFMA obtained by the processes of the invention may similarly be used to generate R-1234yf.

The processes of the invention described above may also be combined with one or more additional purification steps in order to purify the R-1234, for example the R-1234ze, (e.g. to increase or further increase the amount of the E-isomer of R-1234ze relative to the corresponding Z-isomer).

Such additional purification may be performed before, during or after the process for the reduction of the TFMA content is performed. The present invention provides a process for producing R-1234, for example R-1234ze with a TFMA content of less than about 500 ppm w/w, e.g. less than about 200 ppm w/w, preferably less than about 100 ppm w/w, and more preferably less than about 40 ppm w/w, for example 30 ppm w/w or less. This process can produce R-1234, particularly R-1234ze, more particularly R-1234ze(E), which is considered non-flammable at an ambient test temperature of about 23°C using the ASHRAE-34 methodology.

The above processes for reducing the amount of TFMA present in R-1234, particularly R-1234ze, more particularly R-1234ze(E), may be performed simultaneously with the synthesis of R-1234, or as a separate step thereafter. The amount of TFMA in R-1234 particularly R-1234ze, more particularly R-1234ze(E), as obtained from commercially available sources such as Apollo Scientific, may also be reduced by the processes described herein.

R-1234, particularly R-1234ze, may be prepared from 1,1,1,3,3-pentafluoropropane (HFC-245fa), or preferably 1,1,1,2,3-pentafluoropropane (HFC-245eb), by dehydrofluorination. Examples of processes using HFC-245eb may be found in, for example, WO 2009/137656.

R-1234, particularly R-1234ze, may also be prepared from 2-chloro-1 , 1,1,3- tetrafluoropropane (HCFC-244db), or preferably 3-chloro-1,1,1,3-tetrafluoropropane (HCFC-244fa), by dehydrochlorination.

One or more of the by-products of these dehydrohalogenation reactions may be a flammable compound. By-products may include TFMA. The products of these dehydrofluorination and dehydrochlorination reactions may be subsequently purified by the partial or total removal of said by-products, e.g. by the methods outlined above.

For the avoidance of doubt, by the term "dehydrofluorination" (or dehydrofluorinating), we refer to the removal of hydrogen fluoride (HF) from the hydrofluorocarbon starting material. The term "dehydrochlorination" (or dehydrochlorinating) refers to the removal of hydrogen chloride (HCI), similarly.

This dehydrohalogenation process may be carried out by any suitable reaction conditions effective to dehydrofluorinate 245eb or 245fa, or to dehydrochlorinate 244fa or 244db, to produce R-1234ze.

The dehydrohalogenation reaction may be carried out in the liquid or vapour phase, preferably the vapour phase. A temperature of from about -25 to about 300 °C may be used. Preferred temperatures for liquid phase dehydrochlorination are from about 0 to about 180 °C, e.g. from about 15 to about 120 °C. Preferred temperatures for vapour phase dehydrohalogenation are from about 0 to about 300 °C, such as from about 20 to about 250 °C, e.g. from about 50 to about 200 °C.

The dehydrohalogenation reaction may be carried out at atmospheric, sub- or super- atmospheric pressure, preferably super-atmospheric pressure. For example, the dehydrohalogenation may be carried out at a pressure of from about 0 to about 40 bara, such as from about 1 to about 30 bara, e.g. from about 5 to about 20 bara. It may be possible to perform the dehydrohalogenation reaction under conditions which are modified compared to those which would normally be used by the skilled person, in order to reduce the amount of impurities such as TFMA that may be produced. Such modifications may include reducing the reaction time, lowering the reaction temperature, and/or lowering the pressure in the reaction vessel.

The dehydrohalogenation may be induced thermally, may be base-mediated and/or may be catalysed by any suitable catalyst. Suitable catalysts include metal and carbon based catalysts such as those comprising activated carbon, main group (e.g. alumina-based catalysts) and transition metals, such as chromia-based catalysts (e.g. zinc/chromia) or nickel-based catalysts (e.g. nickel mesh).

Other transition metal hydrogenation catalysts include those comprising nickel (Ni), palladium (Pd), platinum (Pt), rhenium (Re), rhodium (Rh), ruthenium (Ru) and mixtures thereof. Such catalysts may be supported on, for example, alumina, titania, silica, zirconia, fluorides of the foregoing, calcium fluoride, carbon or barium sulphate, or they may be unsupported, for example Raney Ni or Pt metal produced by reduction of Pd0 ₂ . Examples of catalysts suitable for use in the present invention include Pd/alumina, Pd/barium sulphate, Pd/C and chlorotris(triphenylphosphine)rhodium(l). In one aspect of the present invention the catalyst is palladium supported on carbon (Pd/C) or chlorotris(triphenylphosphine)rhodium(l) (Wilkinson's catalyst) or platinum supported on alumina (Pt/Al ₂ 0 ₃ ) or Adams catalyst, Pt0 _2l reduced in situ to platinum metal. When Pd/C is used as the catalyst, the Pd is present in an amount of from about 0.01 to about 10% by weight of the catalyst, such as from about 0.1 to about 5 %

Preferably the catalyst is stable in the presence of HF and/or HCI. In the vapour phase the contact time for the dehydrohalogenation catalyst with the hydro(halo)fluoroalkane suitably is from about 1 to about 200 seconds, such as from about 2 to about 150 seconds.

In the liquid phase the contact time for the dehydrohalogenation catalyst with the hydro(halo)fluoroalkane suitably is from about 1 to about 180 minutes, such as from about 2 to about 60 minutes. The dehydrohalogenation can be carried out in any suitable apparatus, such as a static mixer, a stirred tank reactor or a stirred vapour-liquid disengagement vessel. Preferably, the apparatus is made from one or more materials that are resistant to corrosion, e.g. Hastelloy® or Inconel®. The process may be carried out batch-wise or continuously and in the liquid or gas phase.

One preferred method of effecting the dehydrohalogenation of 245eb, 245fa, 244fa or 244db to produce R-1234ze and/or R-1234yf is to contact the starting material with a catalyst based on chromia such as those described in EP-A-0502605, EP-A-0773061, EP-A-957074, WO 98/10862, WO 2006/106353 and WO2008/040969 (all of which are incorporated herein by reference) (e.g. a zinc/chromia catalyst).

Another preferred method of effecting the dehydrohalogenation of 245eb, 245fa, 244fa or 244db to produce R-1234ze and/or R-1234yf is to contact the starting material with a base (base-mediated dehydrohalogenation).

This base-mediated dehydrohalogenation process comprises contacting the hydro(halo)fluoroalkane with base such as a metal hydroxide or amide (preferably a basic metal hydroxide or amide, e.g. an alkali or alkaline earth metal hydroxide or amide).

Unless otherwise stated, as used herein, by the term "alkali metal hydroxide", we refer to a compound or mixture of compounds selected from lithium hydroxide, sodium hydroxide, rubidium hydroxide, caesium hydroxide and, particularly, potassium hydroxide. Similarly, by the term "alkali metal amide", we refer to a compound or mixture of compounds selected from lithium amide, sodium amide, potassium amide, rubidium amide and caesium amide.

Unless otherwise stated, as used herein, by the term "alkaline earth metal hydroxide", we refer to a compound or mixture of compounds selected from beryllium hydroxide, magnesium hydroxide, calcium hydroxide, strontium hydroxide and barium hydroxide. Similarly, by the term "alkaline earth metal amide", we refer to a compound or mixture of compounds selected from beryllium amide, magnesium amide, calcium amide, strontium amide and barium amide.

Typically, the base-mediated dehydrohalogenation process is conducted at a temperature of from -50 to 300 °C. Preferably, the process is conducted at a temperature of from 20 to 250 °C, for example from 50 to 200 °C. The base-mediated dehydrohalogenation may be conducted at a pressure of from 0 to 30 bara.

The residence/reaction time for the base-mediated dehydrohalogenation process may vary over a wide range. However, the reaction time will typically be in the region of from 0.01 to 100 hours, such as from 0.1 to 50 hours, e.g. from 1 to 20 hours.

Of course, the skilled person will appreciate that the preferred conditions (e.g. temperature, pressure and reaction time) for conducting the base-mediated dehydrohalogenation may vary depending on a number of factors such as the base being employed, and/or the presence of a catalyst etc.

The base-mediated dehydrohalogenation process may be carried out in the presence or absence of a solvent. If no solvent is used, the hydro(halo)fluoroalkane may be passed into or over molten base or hot base, for example in a tubular reactor. If a solvent is used, in some embodiments a preferred solvent is water, although many other solvents may be used. In some embodiments solvents such as alcohols (e.g. propan-1-ol), diols (e.g. ethylene glycol) and polyols such as polyethylene glycol (e.g. PEG200 or PEG300) may be preferred. These solvents can be used alone or in combination. In further embodiments, solvents from the class known as polar aprotic solvents may be preferred. Examples of such polar aprotic solvents include diglyme, sulfolane, dimethylformamide (DMF), dioxane, acetonitrile, hexamethylphosphoramide (HMPA), dimethyl sulphoxide (DMSO) and N-methyl pyrrolidone (NMP). The boiling point of the solvent is preferably such that it does not generate excessive pressure under reaction conditions.

A preferred base is an alkali metal hydroxide selected from the group consisting of lithium hydroxide, sodium hydroxide and potassium hydroxide, more preferably, sodium hydroxide and potassium hydroxide and most preferably potassium hydroxide. Another preferred base is an alkaline earth metal hydroxide selected from the group consisting of magnesium hydroxide and calcium hydroxide, more preferably calcium hydroxide.

The base is typically present in an amount of from 1 to 50 weight % based on the total weight of the components in the dehydrohalogenation reaction mixture. Preferably, the base is present in an amount of from 5 to 30 weight %. The molar ratio of base:hydro(halo)fluoroalkane is typically from 1:20 to 50:1, preferably from 1 :5 to 20: 1 , for example from 1 :2 to 10: 1.

As mentioned above, the base-mediated dehydrohalogenation may preferably employ water as the solvent. Thus, the dehydrohalogenation reaction may preferably use an aqueous solution of at least one base, such as an alkali (or alkaline earth) metal hydroxide, without the need for a co-solvent or diluent. However, a co-solvent or diluent can be used for example to modify the system viscosity, to act as a preferred phase for reaction by-products, or to increase thermal mass. Useful co-solvents or diluents include those that are not reactive with or negatively impact the equilibrium or kinetics of the process and include alcohols such as methanol and ethanol; diols such as ethylene glycol; ethers such as diethyl ether, dibutyl ether; esters such as methyl acetate, ethyl acetate and the like; linear, branched and cyclic alkanes such as cyclohexane, methylcyclohexane; fluorinated diluents such as hexafluoroisopropanol, perfluorotetrahydrofuran and perfluorodecalin.

The base-mediated dehydrohalogenation is preferably conducted in the presence of a catalyst. The catalyst is preferably a phase transfer catalyst which facilitates the transfer of ionic compounds into an organic phase from, for example, a water phase. If water is used as a solvent, an aqueous or inorganic phase is present as a consequence of the alkali metal hydroxide and an organic phase is present as a result of the fluorocarbon. The phase transfer catalyst facilitates the reaction of these dissimilar components. While various phase transfer catalysts may function in different ways, their mechanism of action is not determinative of their utility in the dehydrohalogenation reaction provided that they facilitate that reaction. The phase transfer catalyst can be ionic or neutral and is typically selected from the group consisting of crown ethers, onium salts, cryptands and polyalkylene glycols and derivatives thereof (e.g. fluorinated derivatives thereof).

An effective amount of the phase transfer catalyst should be used in order to effect the desired reaction, influence selectivity to the desired products or enhance the yield; such an amount can be determined by limited experimentation once the reactants, process conditions and phase transfer catalyst are selected. Typically, the amount of catalyst used relative to the amount of hydro(halo)fluoroalkane present is from 0.001 to 20 mol %, such as from 0.01 to 10 mol %, e.g. from 0.05 to 5 mol %.

Examples of suitable crown ethers, cryptands, onium salts and polyalkylene glycol compounds useful as phase transfer catalysts are described in WO2008/075017, which is incorporated herein by reference.

Crown ethers are cyclic molecules in which ether groups are connected by dimethylene linkages. Crown ethers form a molecular structure that is believed to be capable of receiving or holding the alkali metal ion of the hydroxide and to thereby facilitate the reaction. Particularly useful crown ethers include 18-crown-6 (especially in combination with potassium hydroxide), 15-crown-5 (especially in combination with sodium hydroxide) and 12-crown-4 (especially in combination with lithium hydroxide).

Derivatives of the above crown ethers are also useful, such as dibenzyl-18-crown-6, dicyclohexanyl-18-crown-6, dibenzyl-24-crown-8 and dibenzyl-12-crown-4. Other compounds analogous to the crown ethers and useful for the same purpose are compounds which differ by the replacement of one or more of the oxygen atoms by other kinds of donor atoms, particularly N or S. Fluorinated derivatives of all the above may also be used.

Cryptands are another class of compounds useful in the base-mediated dehydrohalogenation as phase transfer catalysts. These are three dimensional polymacrocyclic chelating agents that are formed by joining bridgehead structures with chains that contain properly spaced donor atoms. The donor atoms of the bridges may all be O, N, or S, or the compounds may be mixed donor macrocycles in which the bridge strands contain combinations of such donor atoms. Suitable cryptands include bicyclic molecules that result from joining nitrogen bridgeheads with chains of (-OCH2CH2-) groups, for example as in [2.2.2]cryptand (4,7, 13,16,21, 24-hexaoxa-1 , 10- diazabicyclo[8.8.8]hexacosane, available under the brand names Kryptand 222 and Kryptofix 222).

Onium salts that may be used as catalysts in the base-mediated dehydrohalogenation process include quaternary phosphonium salts and quaternary ammonium salts, which may be represented by the formulae R ¹ R ² R ³ R P ⁺ Z ^" and R ¹ R ² R ³ R ⁴ N ⁺ Z ^" , respectively. In these formulae, each of R ¹ , R ² , R ³ and R ⁴ typically represent, independently, a C-MO alkyl group, an aryl group (e.g. phenyl, naphthyl or pyridinyl) or an arylalkyl group (e.g. benzyl or C _1-10 alkyl-substituted phenyl), and Z ^" is a halide or other suitable counterion (e.g. hydrogen sulphate). Specific examples of such phosphonium salts and quaternary ammonium salts include tetramethylammonium chloride, tetramethylammonium bromide, benzyltriethylammonium chloride, methyltrioctylammonium chloride (available commercially under the brands Aliquat 336 and Adogen 464), tetra-n-butylammonium chloride, tetra-n-butylammonium bromide, tetra-n-butylammonium hydrogen sulphate, tetra-n-butylphosphonium chloride, tetraphenylphosphonium bromide, tetraphenylphosphonium chloride, triphenylmethylphosphonium bromide and triphenylmethylphosphonium chloride. Benzyltriethylammonium chloride is preferred for use under strongly basic conditions.

Other useful onium salts include those exhibiting high temperature stabilities (e.g. up to about 200 °C), for example 4-dialkylaminopyridinium salts, tetraphenylarsonium chloride, bis[tris(dimethylamino)phosphine]iminium chloride and tetrakis[tris(dimethylamino)phosphinimino]phosphonium chloride. The latter two compounds are also reported to be stable in the presence of hot, concentrated sodium hydroxide and, therefore, can be particularly useful.

Polyalkylene glycol compounds useful as phase transfer catalysts may be represented by the formula R ^e O(R ⁵ 0) _m R ⁷ wherein R ⁵ is a C _1-10 alkylene group, each of R ⁶ and R ⁷ are, independently H, a C-MO alkyl group, an aryl group (e.g. phenyl, naphthyl or pyridinyl) or an arylalkyl group (e.g. benzyl or C _1-10 alkyl-substituted phenyl), and m is an integer of at least 2. Preferable both R ⁶ and R ⁷ are the same, for example they may both by H.

Such polyalkylene glycols include diethylene glycol, triethylene glycol, tetraethylene glycol, pentaethylene glycol, hexaethylene glycol, diisopropylene glycol, dipropylene glycol, tripropylene glycol, tetrapropylene glycol and tetramethylene glycol, monoalkyl glycol ethers such as monomethyl, monoethyl, monopropyl and monobutyl ethers of such glycols, dialkyl ethers such as tetraethylene glycol dimethyl ether and pentaethylene glycol dimethyl ether, phenyl ethers, benzyl ethers of such glycols, and polyalkylene glycols such as polyethylene glycol (average molecular weight about 300) and polyethylene glycol (average molecular weight about 400) and the dialkyl (e.g. dimethyl, dipropyl, dibutyl) ethers of such polyalkylene glycols.

Combinations of phase transfer catalysts from within one of the groups described above may also be useful as well as combinations or mixtures from more than one group. Crown ethers and quaternary ammonium salts are the currently preferred groups of catalysts, for example 18-crown-6 and its fluorinated derivatives and benzyltriethylammonium chloride. The products from the dehydrohalogenation reaction may then be subjected to one or more purification steps. The purification may be achieved, for example, by separation of the desired product(s) or reagents by one or more distillation, condensation or phase separation steps and/or by scrubbing with water or aqueous base. Removal of one or more undesired product may also be achieved by such processes.

The product of the dehydrohalogenation of 245eb, 245fa, 244fa or 244db may contain a mixture of isomeric forms of tetrafluoropropene, e.g. R-1234yf, and/or either or both of the E and Z isomers of R-1234ze.

Preferably, the starting material for the dehydrohalogenation is either 1,1 ,1,2,3- pentafluoropropane (HFC-245eb) or 3-chloro-1,1 ,1,3-tetrafluoropropane (HCFC-244fa), in which case, the reaction mixture may contain either or both of the E and Z isomers of R-1234ze in the substantial absence of R-1234yf (e.g. less than 1 wt% or, preferably, 0.1 wt% R-1234yf).

Where necessary, these isomers may be separated from each other using one or more conventional methods known to those skilled in the art, e.g. distillation, condensation or phase separation.

These isomers may be separated from each other either before or after the amount of TFMA impurity in the R-1234 product is reduced. Products of this invention that comprise R-1234ze, particularly R-1234ze(E), may be classified as non-flammable at a test temperature of 23°C using the ASHRAE-34 methodology.

Products of the invention may comprise R-1234, particularly R-1234ze, more particularly R-1234ze(E), and TFMA, wherein the amount of TFMA is less than about 500 ppm w/w, e.g. less than about 200 ppm w/w, preferably less than about 100 ppm w/w, and more preferably less than about 40 ppm w/w, for example 30 ppm w/w or less.

R-1234ze, e.g. as made by any of the processes of the invention, may be used to form a blend in combination with other components such as R-744 (C0 ₂ ). Such compositions may consist essentially of (or consist of) R-1234ze(E), R-744 and optionally one or more additional components. By the term "consist essentially of, we mean that the compositions of the invention contain substantially no other components, particularly no further (hydro)(fluoro)compounds (e.g. (hydro)(fluoro)alkanes or (hydro)(fluoro)alkenes) known to be used in heat transfer compositions. We include the term "consist of within the meaning of "consist essentially of.

For the avoidance of doubt, any of the compositions of the invention described herein, including those with specifically defined compounds and amounts of compounds or components, may consist essentially of (or consist of) the compounds or components defined in those compositions.

In one aspect, the one or more additional components may be any one of difluoromethane (R-32), 1 ,1-difluoroethane (R-152a), fluoroethane (R-161), 1 ,1,1,2- tetrafluoroethane (R-134a), propylene or propane. Thus, the compositions of the invention may be ternary blends of R-1234ze(E), R-744 and one of R-32, R-152a, R-161, propylene, propane or, particularly, R-134a.

Preferably, the composition consists essentially of:

i) R-1234ze(E) as obtained by the processes described herein, e.g. present at from about 89% to about 97% by weight of the total composition;

ii) R-134a, e.g. present at from about 2% to about 10% by weight of the total composition; and

iii) R-744, e.g. present at from about 1% to about 5% by weight of the total composition.

Another preferred composition consists essentially of:

i) R-1234ze(E) as obtained by the processes described herein, e.g. present at from about 88% to about 98% by weight of the total composition; and

ii) R-134a, e.g. present at from about 2% to about 12% by weight of the total composition.

However, the one or more additional components may be mixtures of one or more of these compounds. For example, the third component may include R-134a together with one of R-32, R-152a, R-161, propylene or propane. The R-134a typically is included to reduce the flammability of the equivalent composition that does not contain R-134a. R-744 and R-134a are substances that are known to be non-flammable under standard conditions.

R-1234ze (e.g. R-1234ze(E)), or a composition containing R-1234ze (e.g. R-1234ze(E)) according to the invention may be a refrigerant, heat transfer composition, foaming agent, blowing agent, cleaning agent, carrier fluid, fire extinguisher/retardant, aerosol propellant or solvent or included in a composition for one of these uses.

Flammability Testing

Test standard

Flammability was assessed using a test performed at ambient temperature and at atmospheric pressure in accordance with ASTM E 681-04, "Standard Test Method for Concentration Limits of Flammability of Chemicals" modified, per Annex A1 , "Test Method for Materials with Large Quenching Distances, Which May Be Difficult to Ignite." (which is incorporated herein by reference).

The modification was to conduct the test in a 12-liter glass combustion chamber (American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (ASHRAE) requirements, size for combustion flask chamber, corresponding to ISO 817 recommended method of testing for refrigerants). Electrical spark was used as an ignition source. Further details may be found in the ANSI/ASHRAE Standard 34-2010. Setup

Concentration Limits of Flammability tests were conducted at ambient (23 °C) and 60 °C temperatures and atmospheric pressure.

The test apparatus consisted of a one glass 12-liter (actual capacity: 12,600-ml) chamber (flask), heating oven, vacuum pump, stirrer/propeller with drive motor, high voltage power supply (for ignition source), and supply of controlled relative humidity air. The electrical oven was equipped with mechanical air circulation to heat the flask uniformly. A precision digital vacuum gauge, Model 68920, 800 mmHg range (PSI-Tronix) was used to measure vacuum and partial pressure. A vacuum pump model 47577 (Dayton) was used to evacuate the flask to approximately 5mmHg.

Procedure

The heating temperature was set at the desired test temperature. For each trial, a measured amount (by partial pressure) of the sample mixture was inserted into the prevacuumed glass flask. The flask was then brought back to atmospheric pressure with 50 ± 5% relative humidity air.

A stirrer/propeller combination was turned on to obtain complete mixing and attainment of thermal equilibrium. After at least three-five minutes of mixing, the stirrer was turned off. The high voltage luminous tube transformer (15,000 V, 30 mA) equipped with a timer (0.4 sec) was turned on to create a spark between two 1 mm diameter tungsten electrodes. The electrode gap was 6.3 mm (¼ in.). Observations were made to determine whether or not the gas mixture ignited inside the flask.

The flask was then purged with compressed air to clean out residue. The amount of the sample was varied between trials until the concentrations, which could just sustain propagation and non-propagation of flame were determined.

Flame Propagation Criterion

Propagation of flame - as used in this test method, the upward and outward movement of the flame front from the ignition source to the vessel walls beyond the size of a narrow cone. This cone is produced by drawing a 90° angle from the electrode tips to the base of the neck of the flask and rotating about the axis of the neck. The upward and downward propagation of the flame away from the ignition was rated by visual observation.

A flame propagation is any combustion that, having moved upward and outward from the point of ignition to the walls of the flask, is continuous along an arc that is greater than that subtended by an angle equal to 90 degrees, as measured from the point of ignition to the walls of the flask.

The invention will now be illustrated by the following non-limiting examples. Examples

Example 1 Purification of R-1234ze(E) by distillative separation

The vapour pressure of TFMA and of R-1234ze(E) were determined over the temperature range -36 to +60°C by placing a sample of >99.5% pure material in an evacuated sample cell of known volume containing an agitator, inside a temperature controlled aluminium block, and allowing the cell to come to a steady equilibrium pressure at regular intervals over the temperature range. Next, the vapour pressures of binary mixtures of the fluids were determined by charging known masses of each component into the cell, allowing the pressure and temperature to stabilise, and recording the steady values. This was carried out at 10% weight fraction increments over the composition range 0-100% TFMA and over the temperature range -36 to +60°C.

The vapour pressure, composition and temperature data were then reduced to fit a thermodynamic model of vapour liquid equilibrium (VLE) of the species. The data reduction technique used was the Barker method, as outlined in Chapter 8 of the standard reference text "The Properties of Gases and Liquids" 5th edition (editors BE Poling, JM Prausnitz, JP O'Connell pub. McGraw-Hill 2000) (which is incorporated herein by reference). The VLE model was of the "Gamma-Phi" form as described in the reference text "Models for Thermodynamic and Phase Equilibria Calculations" (by SI Sandler pub. Marcel Dekker Inc 1994) and used the Redlich Kwong equation of state to represent the vapour phase non-ideality and the Wilson equation to represent liquid phase non-ideality. The pure component critical point was measured using standard techniques for R-1234ze(E) and estimated using the methods of Joback as described in Poling et al for TFMA. Upon successful reduction of the data to this model it was found that the mixture of TFMA and R-1234ze(E) exhibited non-ideal behaviour, in other words that it did not obey Raoult's Law, at all pressures greater than atmospheric pressure, with the degree of deviation from ideal mixture behaviour increasing as pressure increased. From the measurements of vapour pressure data the atmospheric boiling point of TFMA was determined as approximately -47°C, in contrast to the boiling point of R-1234ze(E) which is -19.4°C. This difference in boiling point may suggest that simple batch distillation could be considered for reduction of the TFMA content. This was investigated however and found to be ineffective. Simple batch distillation although it can generate a stream of the required purity (<40ppm) results in poor yield of said purified stream. The distillative separation of TFMA from R-1234ze(E) was then studied at a range of pressures from 1 to 20 bar. It was found that either continuous or semi-continuous fractional distillation of TFMA from R-1234ze(E) would be effective in allowing separation of TFMA from R-1234ze(E). By fractional distillation is meant here a distillation column having a plurality of separation stages, equipped with a reboiler and partial or total condenser, having its feed material fed as liquid or vapour (or a two-phase mixture) to an intermediate point on the column and with withdrawal of liquid bottom product of the purified R-1234ze(E) and a vapour or liquid top product enriched in TFMA.

It is found furthermore that the size of column required to effect a good separation is significantly greater than would be expected on the basis of application of Raoult's Law. This is because the relative volatility of TFMA to R-1234ze(E) Is surprisingly low compared to that predicted by Raoult's Law.

In the analysis it was assumed that the feed stream of material to be distilled comprised 1500ppm w/w of TFMA in R-1234ze(E) and the target purity of R-1234ze(E) was fixed as 40ppm w/w.

Example 2 Flammability testing of R-1234ze(E) containing 40 ppm w/w TFMA

R-1234ze(E) containing approximately 40 ppm w/w TFMA was tested according to the Flammability Testing method described above, and using the following test criteria: Test Chamber: 12-litre Glass Chamber

Test (Chamber) Temperature: 60 ± 1 °C

Environment: Air with 50± 5 % RH

Ignition Source: Electrical Spark

Material Tested: R-1234ze(E)

The data showed that flame propagation did not occur for this sample, in 25 of the 27 tests performed, at partial pressures ranging from 46 to 106 mm HG. The lack of flame propagation was determined on the basis of the approximate flame angle being less than 90°.

In 2 of the 27 tests performed, the flame angle was between 90° and 92°. The test methodology for the ANSI/ASHRAE Standard indicates a typical experimental error of ± 5.0° for the measurement of the flame angle (see B1.9 in Appendix B of ANSI/ASHRAE Standard 34-2010).

The difference between these data and the non-flammable criterion is therefore within experimental error. The test material was concluded to be non-flammable under these test conditions.

Example 3 Flammability testing of R- 1234ze(E) containing 1180 ppm w w TFMA

R-1234ze(E) containing approximately 1180 ppm w/w TFMA was tested according to the Flammability Testing method described above, and using the following test criteria: Test Chamber: 12-litre Glass Chamber

Test (Chamber) Temperature: 60 ± 1 °C

Test pressure: approx. 760 mm Hg

Ignition Source: Electrical Spark

Material Tested: R-1234ze(E)

The data showed that flame propagation did occur for this sample in the tests performed at concentrations of 7 to 15% by volume in air.

The test material was therefore found to be flammable under these test conditions.

Example 4

Flammability testing of ternary refrigerant blend (using R-1234ze(E) containing 40 ppm w/w TFMA)

A blend of R744, R134a and R-1234ze(E) in the proportions 5%/9%/86% by weight was tested at 23°C and 60°C. Test Chamber: 12-litre Glass Chamber

Test (Chamber) Temperatures: 23 ± 1 °C, 60 ± 1 °C

Test pressure: 760-761 mm Hg

Ignition Source: Electrical Spark

Material Tested: blend of R744, R134a and R-1234ze(E) in the proportions 5/9/86 wt%

In this example, R-1234ze(E) containing approximately 40 ppm w/w TFMA was used to prepare the blend.

The blend was found to be non-flammable at both temperatures. The maximum flame angle observed at the higher test temperature was less than 80 degrees.

Example 5

Flammability testing of ternary refrigerant blend (using R-1234ze(E) containing 1180 ppm w/w TFMA)

The flammability of the ternary blend, as described in Example 4 was assessed using the method of Example 1 , and the test criteria below.

Test Chamber: 12-litre Glass Chamber

Test (Chamber) Temperature: 23 ± 1 °C

Test pressure: 760-761 mm Hg

Ignition Source: Electrical Spark

Material Tested: blend of R744, R134a and R-1234ze(E) in the proportions 5/9/86 wt%

In this example, R-1234ze(E) containing approximately 1180 ppm w/w TFMA was used to prepare the blend.

Flame propagation was observed in 11 of the 21 tests performed, at partial pressures ranging from 53 to 99 mm Hg, on the basis of flame angles observed at up to 125°. The blend was found to be flammable under these test conditions. Example 6

Modelled distillation of a mixed feed of TFMA from R1234ze(E) A conventionally configured fractional distillation operation was considered having a feed stream of the liquid mixture of TFMA and R1234ze(E) and producing two liquid streams: a small distillate stream enriched in TFMA and a larger bottom product stream containing R1234ze(E) whose TFMA content has been reduced as a result of the distillation. Figure 1 shows this column in schematic form.

In this example, the following conditions were assumed: · The distillation column is assumed to operate at a pressure of 5 bar absolute with negligible pressure drop over the column.

• The feed rate is taken as 1 kmol/hr as liquid at 20°C.

• The feed composition is 1500ppm weight basis of TFMA in R1234ze(E) (which is 0.18% on a molar basis).

· The distillate composition is 10% (molar) basis of TFMA.

• The column bottoms product contains 40ppm weight basis of TFMA (which is 0.005% on a molar basis).

• The column has 21 theoretical distillation stages and is assumed perfectly lagged (no heat gain or loss to ambient).

· The overhead condenser acts as a total condenser.

• The liquid reflux to the column is taken as 111% of the theoretical minimum reflux for the separation.

The measured vapour liquid equilibrium data for the mixture of R1234ze(E) and TFMA were incorporated into the thermodynamic property model to enable simulation of this distillation. The vapour and liquid enthalpy of mixtures of the compositions arising around the distillation were then estimated using the Peng Robinson equation of state, which is a thermodynamic fluid model commonly used for modelling of distillation separations. The use of the Peng Robinson equation is discussed in Chapter 3 and 4 of the standard reference text "The Properties of Gases and Liquids" 4 ^th edition (RC Reid; JM Prausnitz; BE Poling, pub McGraw-Hill 1987) ("Reid"; which is incorporated herein by reference). Use of this equation requires knowledge of the critical temperature, critical pressure and acentric factor for each component in the mixture and a knowledge of the ideal gas heat capacity for each component as a function of temperature. For TFMA, the Joback estimation technique as discussed in Reid Chapter 2 was used to estimate the ideal gas enthalpy and critical constants (temperature and pressure) for TFMA. The acentric factor for TFMA was then derived from its definition by using experimentally measured vapour pressure data in combination with the estimated critical temperature for TFMA.

For R1234ze(E) experimentally measured critical temperature, critical pressure and vapour pressure data were used in the simulation. The ideal gas heat capacity for R1234ze(E) was estimated as a function of temperature using molecular modelling software (Hyperchem 7.5).

Results of the simulation:

Using the above specifications the following results were found:

• The liquid reflux rate was 1.9 kmol/hr.

• The distillate product rate was 0.0177 kmol/hr.

• The distillate temperature was 20.7°C.

• The bottom product temperature was 25.1 °C.

· The bottoms product rate was therefore 0.982 kmol/hr.

• The condenser cooling duty was 37230 kJ/hr.

• The reboiler heating duty was 38020 kJ/hr.

Example 7

Modelled distillation of a mixed feed of TFMA from R1234ze(E) using overhead vapour recompression.

In this scheme, vapour overhead product is drawn from the column and compressed to a higher pressure. The reboiler and condenser of the conventional scheme are replaced with a single heat exchanger in which the compressed vapour is condensed and supplies the heat to vaporise fluid in the reboiler. This is shown in schematic form in Figure 2.

In Example 7, the same conditions were assumed as for Example 6, and in addition the following assumptions were made: • The overhead vapour is compressed to 7 bar absolute to ensure that there is a temperature difference of >5K between the condensing vapour and boiling liquid in the heat exchanger.

• The compressor is assumed to operate with an isentropic efficiency of 70%.

It was found that in this case:

• The compressed refrigerant condensed over the temperature range of 32.5°C to 34.5°C.

• The compressor consumed 2000 kJ/hr of mechanical power.

• Distillate liquid left the system at 32.5°C. Cooling of this stream and the liquid reflux to a temperature of 20.7X as in the conventional still above would require 3000 kJ/hr of cooling.

• Compositions and flowrates of overhead, reflux and top/bottom products were as in the previous example.

It is evident that the use of overhead recompression can in this case significantly reduce the requirement for providing heating and cooling utility to the separation system.