BACKGROUND

The reverse osmosis (RO) membranes used today are based on work by John Cadotte at Filmtec Corporation and previously at the Midwest Research Institute (US Patent 4,277,344 in 1981) that produced the first really useful reverse osmosis membrane. This work produced a highly crosslinked aramid polymer (similar technology to Kevlar) by interfacial polymerization between an aqueous solution of m-phenylenediamine and trimesoyl chloride. A water solution of the diamine is adsorbed onto the surface of a microporous, hydrophilic substrate and then this extremely thin aqueous film is contacted with a solution of the trimesoyl chloride in a water insoluble solvent like hexane. The two reactants react almost instantaneously at the water/hexane interface yielding a thin continuous polymer film. Chemical structures of the reactants are such that the resulting film is porous on the scale that can pass water molecules but not larger molecules. Subsequent work at Dow Chemical Corporation (e.g. US Patents 4,769,148 and 4,859,384) improved the wetting of the polymer support membrane by the use of surfactants and bonding to the support membrane by incorporating monomeric materials in the surfactant solutions. It has been thought for quite some time that films with other chemistries besides the Filmtec/Dow aramid film might yield superior water throughput and impurity (e.g. salt) rejection, but methods of producing sufficient thin films with the right properties have not been found.

Lockheed Martin Corporation and others have devoted considerable resources into trying to develop graphene films for RO. Graphene (structure shown in Fig. 1, taken US Patent 8979978) is a chemically very inert material that can be produced in films one or a few molecules thick. It is believed that graphene membranes would have superior RO properties to current RO membranes. The problem is efficiently and economically producing graphene membranes on porous support substrates. One approach tried by Lockheed (US Patent 9,610,546) was to use CVD to produce a nano-film of graphene on a copper substrate, then dissolve the substrate away yielding a film that was then floated onto the proper support material. This approach is fraught with likely insurmountable problems for inexpensive volume manufacturing.

A second process explored by Lockheed was to build up a graphene film on a surface by solution polymerization. In order to produce a graphene-like film it is necessary to use arylation reactions in which one benzene ring is attached to another benzene ring. The arylation reactions commonly used in chemical synthesis today are those catalyzed by palladium and other platinum family metal-based such as Suzuki coupling of aryl boronic acids, Stille coupling of aryl tin compounds, and Kumada coupling of aryl Grignard reagents. Instead of these arylation reactions, in US Patent 9,592,475 Lockheed workers used Ullman coupling, an old, low yielding aryl coupling reaction to couple benzene rings into relatively small "platelets" that were then sucked down from a solvent suspension onto a porous substrate. This sort of technique is likely to yield films with considerably variable thicknesses and porosities that are not likely to be very acceptable.

An analysis of the potential benefits of the development of graphene or graphene-derived membranes over currently commercially available reverse osmosis membranes (S. Homaeigohar and M. Elbahri, "Graphene membranes for water desalination", NPG Asia Materials (2017) 9, e427) concludes that because of the improved performance of currently available conventional reverse osmosis membranes there likely will not be great gain in the energy efficiency of reverse osmosis deriving from reducing the energy cost of forcing water through graphene membranes as opposed to conventional membranes (Elimelech, M. & Phillip, W. A.; "The future of seawater desalination: energy, technology, and the environment"; Science 333, 712 -717 (2011)). The potential for considerable energy saving in terms of utilizing improved reverse osmosis membranes is expected largely to be due to the improved anti- fouling characteristics of these improved membranes. Fouling is the buildup of a slimy coating on the surface of reverse osmosis membrane in a desalination cartridge due to particulate contamination in the input seawater stream including the presence of bacteria and algae. The presence of the bacteria and other microscopic organisms is particularly important since they feed on the other particulate material accelerating the growth of the slimy layer. Because of this problem, the incoming stream of seawater must be pretreated to remove particulate and bacterial contamination. The energy and capital costs of this pretreatment could be eliminated if this fouling problem could be overcome.

A membrane material that has shown potentially useful bactericidal and thus anti-fouling properties is graphene oxide (structure shown in Fig. 2, taken from US Patent 8979978) (Mahmoud, K. A., Mansoor, B., Mansour, A. & Khraisheh, M.; "Functional graphene nanosheets: the next generation membranes for water desalination"; Desalination 356, 208-225 (2015); and Goh, P. & Ismail, A.; "Graphene-based nanomaterial: the state-of-the-art material for cutting edge desalination technology"; Desalination 356, 115 -128 (2015)). This material is produced by "Hummers' method" that involves oxidation of graphite with a mixture of potassium permanganate, sodium nitrate and sulfuric acid (Hummers, W. S.; Offeman, R. E.; "Preparation of Graphitic Oxide". Journal of the American Chemical Society. 80 (6): 1339 (1958). US Patent Application 20160297693 describes an osmosis membrane composed of graphene oxide. The structure of a section of a single layer of graphene oxide is shown Fig. 2 (taken from Zhao, J., Wang, Z., White, J. C. & Xing, B.; "Graphene in the aquatic environment: adsorption, dispersion, toxicity and transformation"; Environ. Sci. Technol. 48, 9995-10009 (2014)). In graphene oxide, the original graphene structure has been functionalized with hydroxy, quinone, epoxy and carboxylic acid functions. It is these oxygen containing functional groups that produce the bactericidal properties of the membrane.

An important metric for reverse osmosis membranes is their salt rejection. Ideally one would like to attain 100% salt rejection in a reverse osmosis membrane used to convert seawater into freshwater. However, in many reverse osmosis membranes salt rejection trades off with water flow rate through the medium. An attractive aspect of graphene and graphene oxide membranes is that they can combine high flow rates and high levels of salt rejection. Salt rejection is heavily dependent on the presence of functionalization with oxygen containing substituents. This was shown by the examination of graphene membranes that had had pores produced in them by a process of first bombarding the graphene with gallium ions to produce defects and then oxidizing the graphene to widen the pores (O'Hern, S. C., Boutilier, M. S., Idrobo, J. C., Song, Y., Kong, J., Laoui, T., Atieh, M. and Karnik, R.; "Selective ionic transport through tunable sub-nanometer pores in single-layer graphene membranes"; Nano Lett. 14, 1234 -1241 (2014)). The inner walls of the pores were found to be highly derivatized with oxidized functional groups. The salt rejection and also some selectivity in what salt ions were found to be rejected depended heavily on which functional groups, e.g. hydroxy versus carboxylic acid groups, were present. For instance, hydroxy groups are very effective in removing Cl- ions from water passing through the pores enabling high salt rejection (Mahmoud, op cit.). What one would like to have is a process by which graphene-like films can be inexpensively and reproducibly fabricated by interfacial polymerization in a manner like the Filmtec membranes are produced. In addition, it would be desirable that such a process also allows oxygen containing functionalities like hydroxy groups to be introduced into the films selectively and in a range of desired concentrations.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a portion of a monolayer sheet of graphene.

FIG. 2 illustrates a portion of a monolayer sheet of graphene oxide.

DETAILED DESCRIPTION

An embodiment of the current invention is a composite membrane that comprises a porous polymer substrate overlaid with an extremely thin (between 0.3 and 20 nm.) film of polymer produced by the reaction of aqueous aryl or heteroaryl diazonium salt solutions with organic quinone solutions at the interface between the two phases The interfacial chemical reaction is based on Meerwein-type arylation type small molecule and oligomer synthetic reactions on quinones as described in US Patent 4,288,147 (Koch, 1980). In this reaction an aqueous solution of the diazo salt of an aromatic amine is reacted with an organic solution of benzoquinone to yield an arylated benzoquinone.

Unlike other aryl to aryl coupling reactions (e.g. the Gomberg-Bachmann and Pschorr reactions) that involve diazonium salts, this reaction was extremely clean and high yielding. The product diphenyl benzoquinone precipitated from reaction solvent mixture as large crystals in essentially 100% yield. A relevant aspect of this reaction for the present invention is that the 2,5-dichlorobenzoquinone reactant reacts at both the 3 and 6 positions on the benzoquinone ring. Thus, it might be expected to participate in a polymerization reaction with a tetraazotized aromatic amine such as tetraazotized 4- phenylenediamine. Support for this concept is derived from the following reaction that was also carried out in US Patent 4,288,147.

Other quinones may participate in the reaction. For instance, the following reaction was carried out in US Patent 4,288,147.

The above prior art strongly suggested the following polymerization reaction was possible:

Since in patent 4,288,147 the object was to make long, linear molecules to be used as dichroic dyes, the quinone reactant was substituted with chloro groups at the 2 and 5 positions so as to yield the proper products. The following reaction was carried out in the prior art.

Different substitution patterns on reactants will yield highly crosslinked products resembling graphene. For instance, tetraazotized 6-methoxyphenylene-l,3-diamine could be reacted with benzoquinone to yield a two-dimensional polymer sheet.

Once membranes are produced the quinone functions may easily be reduced to phenols for instance by treating the membrane material with sodium hydrosulfite in acetic acid.

The reduced polymer from reaction above may be methylated yielding an all methoxy substituted membrane material or it may be acetylated with acetic anhydride. The type and number of oxygen containing substituents on the polymer may be controlled by varying the substituents on the diazotized arylamine and by using mixtures of various diazotized arylamines. Various quinones including condensed ring system quinones may be used. The quinones used may be chosen from, but are not limited to, the following:

yrtnisMii' ⁰⁶²⁵⁴⁸

The use of quinones with less than four reactive positions will loosen the matrix resulting larger pore sizes.

A number of diazonium salts produced from aromatic amines may be used as monomers in the membrane fabrication. The aromatic amine compounds should comprise at least two amino groups. The diazonium salts used may be chosen from, but are not limited to, the following:

In this application we will follow the nomenclature conventions suggested by H. Zollinger, Diazo Chemistry I, VCH (1994). Thus, the last compound listed in the list directly above (with X = Cl) may either be called 3,3"-dimethoxy-4,4"-diazoniobiphenyl dichloride or 3,3"-dimethoxybiphenyl-4,4-diazonium dichloride. In the above compounds the counterions X- may be chloride, bromide, nitrate, bisulfate, tetrafluoroborate, hexafluorophosphate, tosylate, mesylate, or any other negative counterion as is common in the art.

In the aromatic diazonium compounds listed above benzyloxy groups may be substituted for methoxy groups. The product polymer membranes produced using these benzyloxy substituted diazonium salts may then be treated with boron trichloride containing solutions (for instance solutions containing the boron trichloride dimethylsulfide complex) with the result that the benzyloxy groups are debenzylated to hydroxy groups.

In addition to the above monomer reactants, the diazonium ions and quinones maybe oligomeric in nature. An example of an oligomeric reactant is prepared by the reaction of a large excess of naphthalenetetrone with diphenyl ether-3, 5, 3', 5' -tetradiazonium chloride.

The resulting polyquinone compound may then be used as the quinone reactant with another diazonium salt in the interfacial polymerization to form the reverse osmosis membrane. Similarly an oligomeric reactant may be formed by reacting a monodiazotized polyaminoarene with a quinone.

The amino groups of the resultant tetraaminoquinone may then be converted into a diazonium salt and the resultant salt can then be used to react with another quinone compound in the interfacial polymerization to form the reverse osmosis membrane. In this way the molecular structure may be systematically varied using multiple monomer components.

An issue in producing the inventive reverse osmosis membranes is that while our past work showed that benzoquinone can be diarylated in very high yield, steric issues may considerably reduce the yields of the third and fourth arylation reactions required to produce the tetraaminoquinone shown in the previous paragraph. If the yields of the arylation reactions required to produce tetraarylated quinones required as intermediates in the RO membranes are too low, membrane performance may be affected. Thus, it may be advantageous to synthesize and isolate the tetraaminoquinone above in a preliminary synthesis before using this compound as a reactant in the interfacial polymerization used to produce the reverse osmosis membrane.

Other arylation reactions than the Meerwein-type arylation described above may be used in the preliminary syntheses of the tetraaminoquinone above or other oligomeric reactants. As an example, 2,5-(4-nitrophenyl)benzoquinone may be synthesized by the Meerwein-type arylation. This quinone may then be reduced to the corresponding hydroquinone that then may be alkylated and dibrominated to produce 4,4"-dinitro-2',5'-dibromo-3',6'-dimethoxy-p-terphenyl. This compound may then be reacted with potassium 4-nitrophenyltrifluoroborate in a Suzuki arylation reaction catalyzed by palladium acetate in dioxane. The potassium aryltrifluoroborate salt is useful in this case because these salts are known to produce excellent yields in the Suzuki arylation even when the reactants are sterically highly hindered (S. Darses, et al., Tetrahedron Letters 38, 4393 (1997)).

The resulting tetranitro compound may then be reduced to the corresponding tetraamine.

The Meerwein-type arylation of quinones with aryl diazonium salts to produce the reverse osmosis membranes may be catalyzed with transition metal halide salts such as cuprous or cupric halides. Copper(2) acetate has been found to be the most effective catalyst for these reactions (Honraedt, A., etal.; JOC 78; (9); 7604-4609 (2013)

As was described in US Patent 4,769,148 the porous polymer substrates used to prepare the composite reverse osmosis membranes may be advantageously coated with aqueous solutions of anionic or cationic surfactants before they are wetted with the aqueous monomer solutions used in the membrane formation process. This ensures that the aqueous diazonium salt solutions uniformly wet the polymer substrates. In particular, polymeric cationic surfactant solutions may be favored for this application. Patent 4,769,148 suggests cationic polymeric surfactants derivatized with "onium" ions such as sulfonium, quaternary ammonium, pyridinium, phosphonium, thiazolium, imidazolium, etc. be used in these surfactant solutions. Particular favored materials were water-soluble copolymers of sulfonium and quaternary ammonium substituted monomers with methacrylic acid derivatives. For instance, copolymers produced from hydroxyethylmethacrylate, vinylbenzyldimethylsulfonium chloride or hydroxide, p-nonylphenoxynonaethoxyethyl methacrylate and aminoethyl methacrylate were used. In addition to these materials the aqueous surfactant solutions may advantageously comprise quinones or quinone derivatized water-soluble polymers.

Another carbon-carbon bond forming reaction that may be used to form RO membranes of this type is the Suzuki-Miyaura arylation. Since, in order for the interfacial polymerization to be successful some reaction participants must be in the aqueous phase and others in the organic phase, Suzuki-Miyaura reaction participants/conditions must be identified that make this possible. A number of potential combinations of this type are possible given that that the majority of Suzuki-Miyaura reactions involve four types of participants: 1-A compound containing an aryl ring system substituted with a boronic acid group or a boronic acid derivative, 2-A compound containing an aryl ring system with a substituent that may be displaced by the aryl ring system in compound 1 as a result of the reaction (These compounds are often referred to as the coupling partners of compound 1 reactants), 3-A transition metal catalyst (usually palladium or nickel-based and 4-A Bronstead or Lewis base. Fortunately, all four of the reaction participants can be made to be either organic solvent or aqueous phase soluble.

Type 1: Reaction participants 1, 2, and 3 in an organic solvent and 4 in water -

This type of interfacial polymerization has been used to make very thin graphene-like polymer films. D.

Zhou, et al reacted benzene-l,3,5-triboronic acid tripinacol ester with 1,3,5-triodobenzene in the presence of tetrakis(triphenylphosphine)palladium(0) catalyst. All three of these reaction participants were in a toluene phase while the base, potassium carbonate was in an aqueous phase. The result was that a very thin membrane of a graphene-like polymer was formed at the toluene/water interface [D.Zhou. et al., Angew. Chem. Int. Ed. 58, 1376-1381 (2018)]. Individuals from the same group have published additional papers describing similar membrane formation reactions. For instance, the following membrane forming interfacial polymerization reaction at water/toluene interface between benzene-l,3,5-triboronic acid tripinacol ester and 2,7-dibromopyrene catalyzed by (PPhsh Pd was described [D. Wang and M. Li, Chemistry, a European Journal 26, (29) 6490-6491 (2019)].

A last example of membrane formation through interfacial polymerization published by this same group (C. Li, etal., Angew. Chem. Int. Ed. 59, (24) 9403-9407 (2020) utilized a toluene phase containing a heterocyclic compound, 2,7,12-tribromo-5,10,15-triethyltriindole and 1,4-benzenediboronic acid dipinacol ester and the catalyst bis(tri-tert-butylphosphine)palladium(0). The water phase base was sodium hydroxide.

All three of the above examples produced membranes formed between a bulk organic phase and a bulk aqueous phase. No porous support substrate was involved. In an application where the membrane needs to formed on the surface of a porous support, there is a potential issue with the Type 1 polymerization reaction. Since there is only base in the aqueous phase, there is no material or mechanism that would adhere the membrane to the surface of the support. In the prior art RO membranes a mixture of a monomer reactant and a mixture of wetting agents are found in the aqueous phase before interfacial polymerization such that after polymerization there is some bonding of the RO membrane to the surface of the porous support.

In an embodiment of the invention using the Type 1 fabrication method, an aryl or heteroarylboronic acid derivative such as the free boronic acid, a boronic acid ester, or aryl trifluoroborate salt; and an aryl or heteroaryl compound coupling partner with a leaving group such as a halide, pseudohalide, such as a tosylate, mesylate, or triflate, or an alkylthio group; and a transition metal catalyst, for instance, a palladium or a nickel catalyst are dissolved in the toluene phase. For instance, the toluene solution may contain benzenel,3-diboronic acid dipinacol ester, 2,3,5,6-tetra(4-bromophenyl)-hydroquinone, and palladium-tetrakis(triphenylphosphine) catalyst. The palladium-tetrakis(triphenylphosphine) catalyst may not be sufficiently active to insure rapid and complete formation of a membrane given that one of the reactants is sterically hindered or that one of the reactants is sluggishly reactive such as a chloro or tosylato groups. Considerable effort has been exerted to develop phosphine-based ligands for use in palladium and nickel catalysts that have high steric demand and thus enable the Suzuki-Miyaura reaction with sterically hindered or sluggish reactants. Catalysts of this type that may be advantageously used in the Type 1 membrane fabrication include bis(tri-t-butylphosphine)palladium(0) (Littke, A.F., etal.; JACS 122; 4020-4028 (2000)). In many instances the high steric demand catalyst is generated in-situ by adding a source of palladium(O) or palladium(2) to the toluene phase along with the desired ligand. Palladium sources that may be used in this way include tris(dibenzylideneacetone)dipalladium(0) (Pdjdbas) and palladium(2) acetate. Sterically demanding ligands that may be used in this way include tri-t-butylphosphine, 2-dicyclohexylphosphino-2',6'- dimethoxybiphenyl (SPhos), or 2-dicyclohexylphosphino-2',4',6'-triisopropylbiphenyl (XPhos) (Martin, R. and Buchwald, S.L., Accts. Of Chem. Res. 41, (11), 1461-1473 (2008).

In this embodiment of the invention that uses the Type 1 polymerization a small amount of reaction participant 2 is present in the aqueous phase along with the base (reaction participant 4) and wetting agents. In order to ensure that the reaction participant 2 present in the aqueous phase becomes involved in the interfacial polymerization, it is advantageous that this reaction participant is more reactive than the larger concentration of reaction participant 2 in the organic phase. An ideal material to be used as the aqueous reaction participant 2 is an aryl diazonium salt because of the water solubility and high reactivity of these materials. The reactivity of these materials has been shown to be higher than aryl bromides or iodides in the Suzuki arylation (F_X. Felpin and S. Sengupta, Chem. Soc. Rev. 48, 1150-1193 (2019). Aryl diazonium tetrafluoroborate salts are preferred because of their stability. Aryl tosylates and mesylates may also be used. An example of a diazonium salt useful for this application is:

This material may also be derivatized with a crosslinking group like the methacrylate group shown below and then copolymerized with other wetting compound monomers such as vinylbenzyl dimethylsulfonium chloride and methacrylic acid.

In an example of a procedure for producing an RO membrane on a polymer support, the aqueous solution of the wetting agents including a diazonium fluoroborate compound, a base such as potassium carbonate and possibly a small portion of a co-solvent such as methanol or acetonitrile is coated onto the surface of a porous substrate and any excess removed by a roller. Surface is then flooded with a toluene solution of benzenel,3-diboronic acid dipinacol ester, 2,3,5,6-tetra(4-bromophenyl)- hydroquinone, and palladium-tetrakis(triphenylphosphine) catalyst. A membrane is formed by polymerization at the interface between the water and toluene phases. It is also possible to use reaction participant 1 type compound in the aqueous phase to bind the membrane to the substrate in the Type 1 interfacial polymerization. A reaction participant 1 compound that may be usedin this way in a low concentration in the aqueous phase is a arylltrifluoroborate salt. An example is:

As was the case with the diazonium salts, this material may also be derivatized with a crosslinking group like the methacrylate group shown below and then copolymerized with other wetting compound monomers such as vinylbenzyl dimethylsulfonium chloride and methacrylic acid.

Type 2: Reaction participants 1 and 2 in an organic solvent and reaction participants 3 and 4 in water - Since, as in the Type 1 method, it may be necessary for small amounts of water soluble versions of participants 1 or 2 in the aqueous phase to participate in the polymerization to bind the membrane to the substrate, it may be advantageous for the catalyst to be in aqueous phase as well. While other water soluble ligands may be used in conjunction with palladium salts such as palladium(2) acetate or palladium(2) chloride high stearic demand ligands such as 2-(di-t-butylphosphino)ethyl- trimethylammonium chloride, 4-( di-t-butylphosphino)-l,l-dimethylpiperidinium chloride (Shaughnessy, K.H. and Booth, R.S., Org. Lett. 3 (17), 2757-2759 (2001)) and sodium 2'-dicyclohexylphosphino-2,6- dimethoxy-l,l'-biphenyl-3-sulfonate (Anderson, K.W. and Buchwald, S.L., Angew. Chem. Int. Ed. 44, 6173-6177 (2005). Otherwise the materials requirement for the Type 2 method are the same as the Type 1 method.

Type 3: Reaction participants 1 and 3 in an organic solvent and reaction participants 2 and 4 in water - In this type of membrane formation method a water soluble compound with labile substitutents reacts with a boronic acid or a boronic acid derivative at the interface between an organic solvent phase and an aqueous phase. The compounds of choice for the aqueous phase reactants are aryl or heteroaryl compounds with labile diazonio substituents. This is because of the water solubility and reactivity of the diazonio compounds. The counterions of choice for the diazonio substituted compounds are tetrafluoroborate ions. Diazonium tetrafluoroborates are preferred because they are sufficiently stable to be isolated and purified in crystalline form and because of this it is easy to adjust their concentration in the aqueous solution used. Various sulfonate counterions such as tosylates and mesylates can be comprised by relatively stable aryl and heteroaryl diazonium salts that may also be used in this application. A diazonium tetrafluoroborate compound that may be used in this type of membrane forming process is:

The solubility of the aryl and heteroaryl diazonium salts in the aqueous phase can be enhanced by the addition of a co-solvent such as methanol or acetonitrile. In addition, fluoride salts, e.g. potassium fluoride, may advantageously be used as the base in the aqueous phase in this type of membrane forming reaction since it may increase the stability of the diazonium salt in aqueous solution.

The participant type 1 compounds used in this membrane formation approach may be boronic acids or boronic acid derivatives such as boronic acid esters. An example is:

It is preferred that the catalysts used incorporate high steric demand ligands as were discussed above.

Type 4: Reaction participant 2 in an organic solvent and reaction participants 1, 3 and 4 in water -

In this type of membrane formation method a aqueous-phase soluble, aryl or heteroaryl compound with two or more substituent trifluoroborate groups is dissolved in the aqueous phase. This material reacts with an organic phase-soluble aryl or heteroaryl compound substituted at two or more positions at the interface between an organic solvent phase and an aqueous phase. A water-soluble transition metal (preferably palladium or nickel) catalyst is present in the aqueous phase. This catalyst is preferred to be a catalyst comprising ligands with high steric demand. A base such as potassium carbonate is also present in the aqueous phase. An example of an aqueous-phase soluble trifluoroborate compound is:

The solubility of the aryl and heteroaryl trifluoroborate compound in the aqueous phase can be enhanced by the addition of a co-solvent such as methanol or acetonitrile. A organic-phase soluble compound that may be reacted with the trifluoroborate compound is:

The resulting polymer membrane may be treated with aqueous hydrochloric acid to cleave the dioxole rings yielding a membrane with the structure:

Preferred microporous polymer substrates comprise polysulfone and polyethersulfone polymers. Other microporous polymers as are known in the art may be used.