crosslinking reaction of DABA-containing copolyimidemembranes for gas separations

Sandra Hessa, Claudia Staudtb*

aInstitute of Physical Chemistry, University of Heidelberg, Im Neuenheimer Feld 253, 69120 Heidelberg, GermanybInstitute of Organic and Macromolecular Chemistry, Heinrich-Heine-University,

Universitätsstr. 1, 40225 Düsseldorf, GermanyTel. +49 (0)211-81-15362; Fax +49 (0)211-81-10696; email: staudt@uni-duesseldorf.de

Received 9 August 2006; Accepted 18 January 2007

Abstract

In membrane based gas separation, crosslinking of membrane materials was shown to be an effective method tostabilize the polymer structure against undesirable plasticization effects. In this study a crosslinkable copolyimidewas synthesized by using the dianhydride 6FDA (4,4N-(hexafluoroisopropylidene)diphthalic anhydride) and thediamines ODA (4,4N-oxydianiline), 4MPD (2,3,5,6-tetramethyl-1,4-phenylene diamine) and DABA (3,5-diaminobenzoic acid). The last mentioned diamine contains a free carboxylic acid group where crosslinking of the polymerchains is possible by esterfication reaction with a diol. In the current work EG (ethylene glycol) was chosen ascrosslinker. The esterfication conditions were varied to optimize the solid-state crosslinking reaction. To investigatethe influence of crosslinking conditions on the separation properties of the membranes pure gas permeationexperiments were performed with non-crosslinked and the different EG-crosslinked films at 35°C for the gases CO2using feed pressures up to 40 bar and CH4 at pressures up to 10 bar.

Keywords: Copolyimides; Plasticization; Covalent crosslinking; Variation of crosslinking conditions; CO2/CH4pure gas permeation

1. Introduction

6FDA based copolyimides are very promisingmaterials for membrane based separation pro-

*Corresponding author.

cesses due to their high resistance against heatand aggressive chemicals. Furthermore theypossess excellent film formation and separationproperties [1,2]. But even copolyimides show agreat loss in separation performance if they are

Variation of esterfication conditions to optimize solid-state

Desalination 217 (2007) 8–16

0011-9164/07/$– See front matter © 2007 Published by Elsevier B.V. doi:10.1016/j.desal.2007.01.011

S. Hess, C. Staudt / Desalination 217 (2007) 8–16 9

exposed to highly soluble gases like CO2 orhigher hydrocarbons (≥C3) at higher pressures aswell as to liquids like benzene inducing plasti-cization or swelling of the polymer structure.Plasticization effectuates an increase in freevolume of the polymer as well as an increase insegmental motion of the polymer chains. As aconsequence permeability of all components inthe feed mixture increases while selectivitydecreases resulting in a loss of separation perfor-mance. A method to receive plasticization-resistant membranes is crosslinking the polymerstructure because crosslinks reduce the dilatationof the polymer caused by the swelling gases.Therefore sorption of the gas is limited.Additionally the polymer matrix is stiffened bythe crosslinker and thermal motions of thepolymer chains are restricted [3]. Staudt-Bickeland Koros [4] presented a new method to cross-link polyimides: crosslinkable copolyimides weresynthesized by using 3,5-diamino benzoic acid asone of the monomers. As a result a copolymer isobtained having free carboxylic acid groupswhere crosslinking with e.g. diols is possible.This method allows to control the maximumdegree of crosslinking by the stoichiometry of thedifferent diamines used in the polycondensationreaction leading to reproducible crosslinkedmembranes. To crosslink the polymer withethylene glycol a solid-state esterfication iscarried out by adding the diol to the castingsolution. After evaporating the solvent at lowtemperatures, the obtained membrane is heated at150°C initiating the crosslinking reaction. Thismethod was successfully applied to suppressplasticization in membranes for CO2/CH4- [3,4]and aromatic/aliphatic separations [5]. In CO2/CH4 pure gas experiments crosslinking with EGled to an increase in permeability compared to thenon-crosslinked film [3,4].

This observation can be explained by theincrease of the free volume in the polymer causedby the crosslinker which acts as spacer betweenthe polymer chains. But in pervaporation experi-

ments with cyclohexane/benzene mixtures adecrease in permeability was found for the EG-crosslinked film due to the bulkier feed com-ponents [5]. In all cases an increase in selectivitywas determined and swelling effects wererestricted to a certain degree. Pithan et al. [6]optimized this solid-state esterfication by addingsulfuric acid as catalyst to the diol/solvent/polymer casting solution. It was observed that theamount of diol as well as the amount of acidinfluences the stability of the crosslinkedmembrane. The most suitable ratio seems to be anequimolar amount of sulfuric acid and a six timesexcess of diol referring to the amount of carboxy-lic acid groups in the polymer. Pervaporationexperiments were carried out using toluene/cyclohexane as feed mixture with 6FDA-4MPD/DABA 4:1 membranes [7]. One of the mem-branes was crosslinked with 1,4-butanediol usingsulfuric acid as catalyst, another one wascrosslinked without a catalyst. As referencematerial a non-crosslinked film was used. Thehighest flux was found for the crosslinkedmembrane followed by the non-crosslinked film.The membrane crosslinked using the catalystshows the lowest fluxes. The highest selectivitywas observed for the crosslinked membrane,where sulfuric acid was applied as catalyst, whilefor the crosslinked film without a catalyst nearlythe selectivity of the native film was found. Thisobservation indicates that crosslinking withoutusing an acid catalyst leads to an incompletelycrosslinked film. As a consequence a higher fluxcombined with a poor selectivity is observedcaused by swelling of the membrane material.

Unfortunately the use of sulfuric acid ascatalyst leads in a few cases to a turbidity of thefilm. As a consequence instead of the equimolaramount of sulfuric acid 0.6 equivalents of p-toluene sulfuric acids were used as catalystresulting in clear membranes [7].

In current work different methods to crosslinka copolyimide with ethylene glycol were carriedout in order to investigate their influence on gas

S. Hess, C. Staudt / Desalination 217 (2007) 8–1610

permeability and selectivity. Hence, pure gaspermeation experiments were performed at 35°Cwith CO2 feed pressures of up to 40 bar and withCH4 at feed pressures of up to 10 bar.

2. Theory

The gas separation performance of a mem-brane is characterized by the permeability andselectivity. The permeability coefficient Pi of acomponent i in a mixture is defined as flux Jinormalized by the membrane thickness R and bythe partial pressure difference Δpi between theupstream- and downstream-side of the membrane.

(1)i ii

P Jp

=Δ

Permeability values are usually expressed inbarrers [1 barrer = 10!10 (cm3 (STP) cm)/(cm2scmHg)]. For gas permeation processes thesolution/diffusion model can be applied. Withthis model the permeability coefficient can bedescribed as product of solubility coefficient Siand diffusion coefficient Di.

(2)i i iP S D= ⋅

The solubility coefficient depends on the amountof free volume in the glassy polymer, the con-densibility of the penetrating molecules and theinteraction between polymer and penetrants. Thediffusion coefficient is a measure of mobility ofthe penetrant and is influenced by the packingand motion of the polymer chains and by theshape and size of the permeating molecules.

A measure of the selectivity of a membrane isthe ideal separation factor , which can beij

idαexpressed as ratio of the pure gas permeabilitiesof the components i and j, respectively. Therebyi denotes the faster permeating component in themixture.

(3)id iij

j

PP

α =

Substitution of Eq. (2) in Eq. (3) leads to

(4)id i iij

j j

S DS D⋅

α =⋅

where Si/Sj is called solubility selectivity andDi/Dj is the diffusivity selectivity.

3. Experimental

3.1. Materials and synthesis

In the current work a 6FDA (4,4N-hexa-fluoroisopropylidene diphthalic anhydride) basedcopolyimide was synthesized using followingdiamines: C 4MPD (2,3,5,6-tetramethyl-1,4-phenylene dia-

mine), whose bulky methyl groups ensurehigher permeabilities;

C ODA (4,4N-oxydianiline) to increase the lowselectivity caused by the bulky 4MPD;

C DABA as crosslinkable group

The monomers 6FDA, ODA and DABA werepurified by sublimation and 4MPD by recrystalli-zation in methylene chloride, respectively.

The polycondensation was performed by thereaction of 6FDA with the mentioned diamines indimethyl acetamide. The resulting polyamic acidwas imidized using a 1:1-mixture of aceticanhydride and triethylamine as described else-where [4].

In Fig. 1 the structure of the synthesized6FDA-ODA/4MPD/DABA 1:1:1 polymer is pre-sented. The expression 6FDA-ODA/4MPD/DABA 1:1:1 indicates that in the polymer theratio of 6FDA-ODA- to 6FDA-4MPD- to 6FDA-DABA-units is 1:1:1. The different units aredistributed statistically.

S. Hess, C. Staudt / Desalination 217 (2007) 8–16 11

Fig. 1. Crosslinkable copolyimide 6FDA-ODA/4MPD/DABA 1:1:1 synthesized.

The polymer constitution was verified by 1H-NMR-spectroscopy. A glass transition tem-perature of 360.7°C was found.

3.2. Film formation and crosslinking

3.2.1. Non-crosslinked membranesNon-crosslinked films were obtained by dis-

solving the dried polymer in tetrahydrofurane(2–3 wt%), casting the filtered solution onto aplane glass plate and evaporating the solvent atambient temperature. The membranes were driedunder vacuum at 150°C for 24 h.

3.2.2. EG-crosslinked membranesAs crosslinker the diol ethylene glycol (EG)

was chosen which has been proved to be efficientfor membranes for CO2/CH4 separation and forpervaporation of aromatic/aliphatic mixtures. InFig. 2 the crosslinking reaction is shown, whichrepresents an esterfication between the carboxylicacid group of the DABA units in the polymer andthe hydroxylic groups of EG. Different variationsof the esterfication reaction were tested to findout the most effective crosslinking method.

1. Crosslinking without use of a catalyst. Thedried polymer was dissolved in dimethylacetamide (2–5 wt%), 6 equivalents of EG wereadded and after stirring the solution was cast ontoa plane glass plate. Then the solvent was eva-porated at 110°C. After drying the membrane atambient temperature for 3 h the film was heatedat 150°C for 24 h initiating the crosslinkingreaction. After that the membrane was storedunder vacuum at 150°C for 24 h to removeexcessive crosslinking reagent as well as possible

remains of solvent and the water set free duringthe crosslinking reaction.

2. Crosslinking using a catalyst. The free car-boxylic acid group of DABA owns only a lowreactivity of its carbonyl group, as a consequencethe reaction with alcohols is slow. However, theesterfication can be highly accelerated by addinga strong acid as catalyst.C Crosslinking using sulfuric acid as catalyst.

The polymer was treated as described in (1),adding one equivalent of sulfuric acid addi-tional to the 6 times excess of EG. This pro-cedure results in membranes which are notclear; therefore no further investigations weredone.

C Crosslinking using p-toluenesulfonic acid(p-TSA) as catalyst. Instead of sulfuric acid0.6 equivalents of p-TSA were given to thesolution of polymer and EG in dimethylacetamide. With this method clear membraneswere obtained.

C Crosslinking using methane sulfonic acid(MSA) as catalyst. 0.6 equivalents of the cata-lyst were added and clear crosslinked polymermembranes were received.

3. Crosslinking by formation of a monoesterfollowed by a transesterfication reaction. Cross-linking of the polymer chains is achieved by atwo-step method developed by Wind et al. [8].The reaction scheme is presented in Fig. 3. In afirst step the monoester of the polymer is synthe-sized by a liquid-state reaction: The dried poly-mer was dissolved in water-free NMP (30 wt%polymer). After adding 1–2 mg p-TSA and a 70times excess of EG (referred to the amount ofpolymer) the solution was stirred at 130°C under

S. Hess, C. Staudt / Desalination 217 (2007) 8–1612

Fig. 2. Crosslinking reaction with ethylene glycol using the monoester route.

Fig. 3. Crosslinking reaction with ethylene glycol using esterification reaction.

reflux and under nitrogen for 12 h. After coolingto ambient temperature the solution was slowlypoured into water and the obtained monoesterthreads were hackled and washed with water.Then the monoester was dried, first at room

temperature and after 3 h at 65°C under vacuumovernight. The low temperature was chosen toprevent crosslinking reactions. In a second step asolid-state transesterfication is performed result-ing in a crosslinked polymer structure: the

S. Hess, C. Staudt / Desalination 217 (2007) 8–16 13

monoester was dissolved in tetrahydrofuran, 1 mgp-TSA was added and after stirring the filteredsolution was cast onto a glass plate. Afterevaporating the THF at room temperature themembrane was heated at 150°C under vacuum for24 h initiating the transesterfication reaction.Vacuum is necessary to push the equilibrium ofthe reaction to the formation of crosslinks byremoving the diol set free from the membraneduring transesterfication.

A big advantage of this method is the fact, thatthe monoester is still soluble (in contrast to thecrosslinked films). Therefore it is possible tocalculate at least the conversion of the polymer tothe monoester interpreting the 1H-NMR-spectrumas described in [8]. In this study the conversion tothe monoester was 72.1%.

The characterization of the crosslinked mem-branes themselves is more problematic becausethe films are insoluble in solvents. Therefore aspectroscopic characterization to get informationsabout the degree of crosslinking is not possible.But the determination of the permeability andselectivity is a suitable method to investigate suchpolymer films, because these two factors are verysensitive indicators not only concerning thepermeation of different kind of gases but alsoconcerning the changes in the structure of thepolymeric film. Therefore experimental data weredetermined for the permeabilities of CO2 and CH4of the membranes crosslinked by the differentroutes described before.

3.3. Apparatus and procedureSteady-state pure gas permeation experiments

were performed at 35°C with CO2 up to a feedpressure of 40 bar and with CH4 up to 10 bar,respectively. A standard gas permeation appa-ratus was used described in detail in a previouspublication [9]. To avoid microdefects in themembrane caused by the quenched viton O-ringin the permeation cell, the border of the filmswere masked with self-adhesive aluminium foil.

The effective membrane area was between 3 and4.5 cm2 and the thickness of the films werebetween 15 and 22 µm.

The reproducibility of the determined permea-bility values were proved by investigating the P-values of different membrane samples of onepolymer batch as well as of membrane samples ofdifferent batches of the same polymer.

4. Results

In Fig. 4 the CO2 permeabilities of the variousEG-crosslinked 1:1:1 membranes at different feedpressures are shown. Thereby as reference a non-crosslinked 1:1:1 film was used. An increase inpermeability appears at feed pressures above15 bar indicating the presence of plasticizationinduced changes in the polymer structure. At10 bar, plasticization effects did not occur. A per-meability of 32.6 barrers was found. For the EGcrosslinked membrane where no catalyst wasused during crosslinking reaction conspicuouslyhigher permeabilities were observed (about 46.5barrer at 10 bar feed pressure) because the dis-tance between the polymer chains increases dueto the presence of the crosslinker. Additionallythe crosslinker affects a depletion of plastici-zation, in this case permeability increases abovea feed pressure of 20 bar. But this means not anexplicit improvement of plasticization resistancecompared to the non-crosslinked film indicatingthat after the crosslinking reaction not enoughcrosslinks exist to suppress plasticization effec-tively. Hence, this method of crosslinking is notsufficient. By crosslinking of the polymer usingp-TSA as crosslinker more stable membraneswere obtained. Here a plasticization pressure of30 bar was found, so a higher degree of cross-linking exists. But in this case a disadvantage ofthe use of catalyst is obvious: At a feed pressureof 10 bar a permeability of only 13.8 wasobserved which means a loss of 70% of perme-ability compared to the EG-membrane where no

S. Hess, C. Staudt / Desalination 217 (2007) 8–1614

(a) (b)

Fig. 4. CO2 permeabilities of different 6FDA-ODA/4MPD/DABA1:1:1-membranes. (a) non-crosslinked reference film(!) and EG-crosslinked membranes not catalyzed (Δ), p-TSA- (") and MSA-catalyzed (") as well as a non-crosslinkedmonoester (ME) membrane (L) and a crosslinked film using the ME-route (G) both with p-TSA as catalyst and again anon-crosslinked membrane (!); (b) as a function of feed pressure at T = 35°C.

catalyst was used. Probably the bulky catalyst isstill present in the polymer structure decreasingthe free volume of the polymer. An attempt wasmade to remove the water soluble p-TSA byimmersing the crosslinked film in water for 24 h.But in gas permeation experiments a furtherdecrease of permeability was observed. Anenormous increase in permeability can be reachedby substituting the bulky p-TSA as catalyst by themuch smaller MSA. In this case plasticizationstarts again at a CO2-pressure of 30 bar, but thistime a permeability of 35.6 was found, which iseven higher than the corresponding value of thenon-crosslinked film.

Especially for the production of hollow fibersthe crosslinking method of Wind is very inter-esting. Here crosslinking of the polymer chainswas obtained by a transesterfication reaction ofthe corresponding monoester. In Fig. 4(b) theCO2-permeabilities of such a crosslinked film andas reference values the permeabilities of a non-

crosslinked membrane and a non-crosslinkedmonoester film, respectively, were shown. Theplasticization pressure of the non-crosslinkedmonoester film amounts to 15 bar, which is evenlower than the corresponding pressure of the non-crosslinked film which has not been esterified. Inthe latter film free carboxylic acid groups arepresent. As a consequence the formation ofhydrogen bondings is possible resulting in a“physical” crosslinking of the polymer, whichsuppresses plasticization to a certain degree.

Surprisingly the plasticization pressure of thecovalently crosslinked film starts at about 25 barindicating an incomplete conversion of the mono-ester to the crosslinks, even though p-TSA wasused as catalyst, too. Again the presence of thiscatalyst is the reason for the low permeabilities ofthe EG-crosslinked film and of the monoestermembrane. Nevertheless this method is veryinteresting for the formation of crosslinkedhollow fibers because the other methods men-

S. Hess, C. Staudt / Desalination 217 (2007) 8–16 15

Table 1Ideal separation factors for CO2/CH4 at 35°C and 10 barfeed pressure

Polymer PCO2 PCH4 αideal

Non-crosslinked 32.6 1 31.8EG-crossl., without catalyst 46.5 1.3 36.1EG-crossl., p-TSA as catalyst 13.8 0.4 33.2EG-crossl., MSA as catalyst 35.6 1 37EG-crossl., monoester route 15.2 0.4 39

tioned are problematic for this application.During the crosslinking reaction of the monoestera smaller amount of EG (1 equivalent) has to beremoved from the polymer. The excess of EGwhich was used to synthesize the monoester wasalready removed by pouring the monoester intowater. For the EG membranes which were notcrosslinked by the detouring monoester-route,primarily six equivalents of EG were used and theexcess is set free during the crosslinking reaction.The smaller the amount of EG which has to beremoved out of the membrane material during thecrosslinking reaction, the smaller the amount ofdefects in the hollow fiber. Due to the thin hollowfiber walls higher permeability is achieved(compared to the free films) which can be furtherincreased by using the smaller MSA as catalyst.

In Table 1 the ideal separation factors of thenon-crosslinked membrane and all investigatedEG-crosslinked films are presented. It can beobserved that all EG-crosslinked films show ahigher selectivity compared to the non-cross-linked membrane.

5. Conclusions

Different methods to crosslink the copoly-imide 6FDA-ODA/4MPD/DABA 1:1:1 by usingthe diol EG as crosslinker were investigated. Toreceive an effective suppression of undesirableplasticization effects the application of an acid

catalyst is necessary. But in this study difficultiesoccur removing the catalyst out of the membraneafter the crosslinking reaction. As a consequencepermeabilities of the films were influenced due tothe loss of free volume in the membrane material.In this work methane sulfonic acid (MSA) is thepreferred catalyst due to its small size. To cross-link free copolyimide films the use of six equi-valents of EG with 0.6 equivalents of MSA seemsto be the most effective method. But to crosslinkhollow fibers transesterfication of the monoestermight be a preferrable method because here alower amount of excessive EG has to be removedout of the film during the crosslinking reactionpreventing defects during the preparation offibers.

Acknowledgement

The Max-Planck Institute of PolymerResearch in Mainz, Germany, is gratefullyacknowledged for supporting the polymercharacterization. This research was supported bythe German Research Foundation (DFG: Sta474/4-1) at Bonn, Germany.

References[1] M.R. Coleman and W.J. Koros, Isomeric polyimides

based on fluorinated dianhydrides and diamines forgas separation application, J. Membr. Sci., 50 (1999)285.

[2] K. Tanaka, M. Okano, H. Kita, K.-I. Okamoto andS. Nishi, Effects of qtrifluormethyl side groups ongas permeability and permselectivity in polyimides,Polym. J., 26 (1994) 118.

[3] J.D. Wind, C. Staudt-Bickel, D.R. Paul and W.J.Koros, The effects of crosslinking chemistry on CO2

plasticization of polyimide gas separation mem-branes, Ind. Eng. Chem. Res., 41(24) (2002) 6139–6148.

[4] C. Staudt-Bickel and W.J. Koros, Improvement ofCO2/CH4 separation characteristics of polyimides bychemical crosslinking, J. Membr. Sci., 155 (1999)145.

S. Hess, C. Staudt / Desalination 217 (2007) 8–1616

[5] J. Ren, C. Staudt-Bickel and R.N. Lichtenthaler,Separation of aromatics/aliphatics with crosslinked6FDA-based copolyimides, Sep. Purif. Techn.,22–23 (2001) 31.

[6] F. Pithan, C. Staudt-Bickel, S. Hess and R.N.Lichtenthaler, Polymeric membranes for aromatic/aliphatic separation processes, Chem. Phys. Chem.,3 (2002) 856.

[7] F. Pithan, Neuartige Copolyimidmembranen für diePervaporation zum Einsatz in der Aromaten/Aliphaten-Trennung und der Prozesswasserauf-

bereitung, Ph.D. Thesis, University of Heidelberg,Heidelberg, Germany, 2003.

[8] J.D. Wind, C. Staudt-Bickel, D.R. Paul and W.J.Koros, Solid-state covalent crosslinking of polyimidemembranes for plasticization reduction in CO2/CH4

separations, Macromolecules, 36(6) (2003) 1882.[9] K.C. O’Brien, W.J. Koros, T.A. Barbari and

E.S. Sanders, A new technique for the measurementof multicomponent gas transport through polymericfilms, J. Membr. Sci., 29 (1986) 229.