摘要:

CHROMATOGRAPHY OF COLOURLESS SUBSTANCES AND THE RELATION BETWEEN CONSTITUTION AND ADSORPTION AFFINITY BY H. BROCKMANN Received 18th July 1949 h method for the chromatographic separation of colourless substances based on 'the use of fluorescent adsorbents is described. Relations between adsorption affinity and chemical constitution of azobenzene and stilbene derivatives are demonstrated. The dependence of adsorption affinity on the nature of the adsorbent and solvent was investigated. Water soluble salts like copper sulphate or zinc sulphate were found t o be useful adsorbents for the chromatographic separation of azobenzene compounds. Chromatography of Colourless Substances. l-The adsorption zones s f colourless substances can be followed by the use of fluorescent adsorbents.Ultra-violet light of a wavelength absorbed by the adsorbed substance is employed. The part of the column containing adsorbed substance then appears as a dark band. Adsorbents can be made to fluoresce by the addition of small quantities of irreversibly adsorbed fluorescing compounds. Morin is very satisfactory for alumina magnesium oxide and calcium carbonate and berberine for silica. Salicylic acid and z-hydroxynaphthalene-3- carboxylic acid can also be used. Experimental.-A very convenient method for the preparation of a fluorescent column consists in the mixing of the adsorbent with z yo to 5 yo of finely powdered luminous paint. This method is generally applicable. Calcium sulphate dihydrate prepared by the precipitation of calcium chloride with the calculated quantity of sulphuric acid a t 8oo-goo C can be converted into a valuable adsorbent for colourless compounds and also incidentally an excellent one for coloured ones.Heating a t temperatures between 150' and zooo C yields three different products the hemihydrate has an activity equivalent to that of alumina IV; the soluble anhydrous salt can be obtained in two grades one equivalent to alumina I11 and the other to alumina 11. The product obtained by heating above 250' is only slightly active and is of no practical value. The large adsorption affinity of the soluble anhydrite is very likely to be due t o numerous faults in the crystal lattice. The fluorescent chromatograms were illuminated with light of 366 mp from a mercury vapour lamp and an appropriate filter.For compounds which absorb a t shorter wavelengths a Chlorfilter was used with windows of UGS glass (Schott Jena) which transmits only the Hg line a t 253-7 mp. In our most recent experi- ments on the separation of colourless compounds the following apparatus was used. Light from an electric arc (iron electrodes) was passed through a slit 35 mm. long and 3-4 mm. wide a quartz lens 5 cm. thick and a quartz or rock salt prism 50 mm. high. The resulting line spectrum (length of lines 50 mm.) was projected onto a fluorescent screen ; after focusing the fluorescent screen was replaced by a cardboard sheet on which the position of the spectral lines in the ultra-violet spectrum had been marked. The adsorption column con- tained in a quartz tube was placed in front of the cardboard and moved until the optimum contrast was obtained making the bands of the chromatogram clearly recognizable.It is advisable to employ a combination of adsorbent and fluorescent compound which does not continue to emit light after illumination (luminous paint N4 green I Francke Frankfurt was used). H. BROCKMANN 59 Appearance of Zones.-There are two ways in which the zones can be made apparent (i) in th.e case of adsorbents rendered fluorescent by adsorbed morin or berberine the zones become apparent through extinction of the fluorescence by the adsorbed compound; (ii) adsorbents which have been mixed with luminous paint behave differently. In this case the zones appear as follows.The particles of the adsorbent allow the light to penetrate the column to a certain extent and fluorescence excited there is emitted. Actual adsorption of the light is stronger at places where the adsorbed compound is present so that the fluorescing material in this case the luminous paint particles is not as strongly excited as in those places where the column is free luminous paint particles thus act as small fluorescent screens. Experimental.-That the outer layer of adsorbent allows light to penetrate is shown as follows. A test-tube the diameter of which is such that i t fits in the chromatographic tube with 3-5 mm. clearance all round is filled with luminous paint and placed in the tube. The space between is filled with adsorbent and a colourless substance is adsorbed onto it.On illumination with ultra-violet light the zone appears surprisingly clearly. This filter action of the outer layer of the column can also play a part in the appearance of the zones when adsorbent materials containing adsorbed morin or berberine are employed. In this case each particle of adsorbent acts as a small fluorescent screen. Whether an adsorbed colourless substance is detected through extinction of the the behaviour of an adsorbent + luminous paint with that of the adsorbent + fluorescence or through the filter effect can be decided in some cases by comparing morin or berberine. Table I gives the results of such a comparison. TABLE I -~ ~~~~~~~~ I hl,O,-Luminous paint 1 Al,O,-Morin Visible Non-visible # * ,? I Benzaldehyde P-Tolylaldehyde Acetophenone Anisaldehyde .. . . : ~ . . I A mercury lamp with a filter allowing the transmission of light of wavelength the zones do not appear when adsorbent + luminous paint is employed. The 360-370 mp was used. Since the above compounds hardly absorb a t 365 m p appearance of the zones when adsorbent + morin is used must therefore be due t o extinction of fluorescence. The effectiveness of the extinction is greater in the case of morin than that of berberine. Thus the phenacyl esters of aliphatic carboxylic acids which hardly absorb a t 365 mp are visible with light of this wavelength on Al,O,-morin but not on silica-berberine. Relations Between Constitution and Adsorption Affinity.-The adsorption affinity that is the firmness with which organic compounds are adsorbed from non-aqueous solvents inter alia depends on the nature of the skeleton and of the functional groups of the adsorbed compound.In order to find the effect of the functional groups on adsorption affinity several compounds containing the same basic skeleton but differing only in one functional group were investigated. The first experiments were carried out with $-substituted stilbenes and azobenzenes. These compounds were adsorbed on A1,0 from benzene or from carbon tetrachloride. Since stilbene and azobenzene under these conditions are only very weakly adsorbed the adsorption affinity of the derivatives can be taken t o be almost entirely dependent on the nature of the substituents.The functional groups in Table I1 are arranged in order of decreasing adsorption affinity. 60 CHROMATOGRAPHY OF COLOURLESS SUBSTANCES Compounds separated by a horizontal dash could be separated the others were not fully separable. The two series are in almost complete agreement with each other. Differences are only found in the three unnumbered acyl derivatives on the right-hand side those in the azobenzene series lie higher than those in the stilbene series; the order however remains the R-COOH - -_____ ~ R-COOH R-CONH R-OH R-OH R-NH-AC R-0-AC R-NH R-NH R-0-BZ R-NH-AC R-0-AC -____- R-COOCH R-0-BZ -___ R-COOCH R-N(CH3) 2 R-NO R-N(CH3) 2 R-NO R-OCH ____.__ R-H R-ocH,- R-H same. This difference is probably due to the fact that the azobenzene derivatives were adsorbed from benzene and the stilbene derivatives from carbon tetrachloride.The solubility of the azobenzene derivatives in carbon tetrachloride was unfortunately too small for adsorption experiments to be carried out using this solvent. Dependence of Adsorption Affinity on the Adsorbent.-In the adsorption of organic compounds from non-aqueous solvents only certain points on the adsorbent particle (the so-called active points) play a part. These are mainly at positions where the crystal lattice is faulty. The adsorption activity of an adsorbent therefore does not only depend on its chemical nature but also on the nature of its crystal lattice. a-Al,O does not adsorb whereas y-Al,O and Bohmite are very effective as adsorbents.CaSO, like the " soluble " anhydrous salt possesses the same crystal lattice as the hemihydrate ; it shows good adsorption properties and is in the light of our experience very efficient in the separation of hydroxy-anthraquinones. As the lattice of the soluble anhydrite is changed into that of the insoluble anhydrite by heating to 2joo the capacity for adsorption falls very considerably. We have previously shown that y-Al,O can be obtained in five different grades (by treatment with water vapour or by regulated heating) which can differentiate by adsorption of azo dyes under standardized conditions. The employment of A1,0 with different and reproducible activities has proved so useful that we have attempted to prepare graded specimens of other adsorbents.By varying its water content bentonite was obtained in five grades of activity silica gel and precipitated silica in three CaSO and MgO also in three and CaCO in two different grades. The standardization of these adsorbents was carried out using the same dyes as in the case of alumina. In the course of these experiments we have tried to find out whether with the same solvents our test dyes were adsorbed in the same order as on A1,0,. The following experiments show this not to be the case. 61 H. BROCKMANN Adsorption Sequence of Azo Dyestuffs in Columns of Different Adsorbents. DYESTUFFS $-hydroxyazobenzene 9-amino-azobenzene Sudan Red (0-CH,).C,H,.N:N.C,H,(o-CH,).N:N.C,,H,.OH Sudan Yellow C,H,.N:N C,,H,.OH $-met hoxyazobenzene. SOLVENT Benzene-Petrol ether r/4. The most strongly adsorbed dyestuff is listed at the top of each column the least adsorbed at the bottom. A1203 ~ _ _ - SiO Sudan Red Oxyazobenzene Amino-azobenzene Sudan Yellow Met hoxyazobenzene Hydroxyazobenzene Amino-azobenzene Sudan Red Sudan Yellow Met hoxyazobenzene ~~ CaSO Hydroxyazobenzene Sudan Yellow Amino-azobenzene Sudan Red Methoxyazobenzene observations show. SOLVENT Benzene-Petrol ether r/4. .- ________ Al,O alkaline H ydroxyazobenzene Amino-azobenzene Sudan Red Sudan Yellow Methoxyazobenzene Si02 treated with NaOH - ~ _ _ ~ _ _ _ - - Sudan Red Amino-azobenzene Hydroxy azobenzene Sudan Yellow Met hoxyazobenzene nlg2 H ydroxyazobenzene Sudan Yellow Sudan Red Amino-azobenzene Met hoxyazobenzene __-_ c u s o Amino-azobenzene Sudan Red Hydroxy azobenzene Sudan Yellow Met hoxyazobenzene Effect of the Acid or Alkaline Character of the Adsorbent on the Adsorption Sequence.-The adsorption sequence of acidic or basic dyes is influenced by the acidity or alkalinity of the adsorbent as the following A1,0 treated with HCl ______ Hydroxyazobenzene Sudan Red Amino-azobenzene Sudan Yellow Methoxyazobenzene SiO treated with HC1 Sudan Red H ydrox y azobenzene Amino-azobenzene Sudan Yellow Met hoxyazobenzene Effect of Different Solvents on the Adsorption Sequence.-The firmness with which a compound is adsorbed from a non-aqueous solvent depends on the nature of the solvent in two ways (i) the solvent tends to occupy the active points of the adsorbent ; (ii) the solvent reacts with the dissolved substance (solvation).Both these effects are negligible with non-polar solvents such as petrol ether cyclohexane carbon tetrachloride and benzene but considerable with polar solvents. The effect of various solvents on the adsorption sequence of dyestuffs is shown in the following results. 62 CHROMATOGRAPHY OF COLOURLESS SUBSTANCES H ydroxyazobenzene Amino-azobenzene Sudan Red Sudan Yellow Methoxyazobenzene Chloroform H ydroxyazobenzene Amino-azobenzene Sudan Yellow Sudan Red Met hoxyazobenzene ADSORBENT Al,O Benzene-Petrol ether 114 Carbon tetrachloride - ADSORBENT CaSO Benzene-Petrol ether r/4 Sudan Yellow hydroxyazobenzene Amino-azobenzene Sudan Red Met hoxyazobenzene Chloroform Sudan Yellow hydroxyazobenzene Amino-azobenzene Sudan Red Met hoxyazobenzene There does not seem to be a relationship between the adsorption sequence and the solubility of the test dyestuffs.It is remarkable that Sudan Yellow and hydroxyazobenzene are adsorbed equally firmly from ether. A similar effect with ether as solvent was. observed in the case of the three following Sudan dyes. These compounds are adsorbed on Al,O from benzene-liglit petroleum 1/4 in the order (i) (ii) (iii) but from ether in the order (ii) (iii).,. (I). Sudan Yellow Hydroxyazobenzene Sudan Red Amino-azobenzene Met hoxyazobenzene ____ Tetrahydrofuran Dioxan Cyclohexanone H ydroxyazobenzene Sudan Red Sudan Yellow Amino-azobenzene Met hoxyazobenzene Sudan Yellow hydroxyazobenzene Amino-azobenzene Sudan Red Met hoxyazobenzene Ether Ether H.BROCKMANN R = C6H,.N:N.C6H4 (i) R-NH.CO.CH (ii) R-NH. co .c ,H (iii) R-0-CO.CH or 0,N-R-N(CH,) (iv) R-COOCH (v) R-NO (vi) R-OCH 63 Even though no satisfactory explanation can at the moment be given of the effect of the solvent on the adsorption sequence our observations lead to the following procedure. In the case of a complex mixture which cannot be separated by the use of one adsorbent and one eluent it has been suggested that separation may be effected by adsorption from a non-polar solvent onto a strong adsorbent followed by elution with a series of solvents of increasing adsorption affinity.We believe that this procedure when applied to compounds such as our dyestuffs which in the adsorption sequence are s o dependent on the nature of the solvent is not as satisfactory as a procedure which in our experience has proved to be of value and which involves repeated chromatography of the mixture from one non-polar solvent on several adsorbents of varying activity. The Adsorption Sequence which is independent of both Adsorbent and Solvent.-The marked dependence of the adsorption sequence of our dyestuffs on the nature of the solvent is perhaps due to variations in the solvation of the hydroxyl and amino groups present.Furthermore it is possible that depending on the nature of the solvent the dyestuffs are adsorbed partly in the azobenzene form partly in the tautomeric quinone- imine form. These complications are avoided if the following compounds are used. The adsorption affinity decreases as one goes down the series. Instead of dyestuff (iii) which will be partly hydrolyzed on alkaline adsorbents p-bromo-P’-dimethylamino-azobenzene was employed. This compound as a result of polarized adsorption forms a bluish-violet zone. These compounds were always adsorbed in the same order from the following solvents on the following adsorbents. A1,03 benzene-petrol ether 1/4 ether chloroform. From ether separation of (i) and (ii) is incomplete. SiO benzene-petrol ether 1/4 ether chloroform.CaSO benzene-petrol ether 1/4 benzene ether. hlg0 benzene-petrol ether 1/4. CuSO benzene-petrol ether. From the above results we believe that we may draw the following conclusions concerning the relations between constitution and adsorption. The adsorption affinity of a compound is practically an additive function of the adsorption affinities of the basic skeleton and of the functional groups only when there is no possibility of tautomerisrn and when solvation by the solvent employed does not occur. The above-mentioned series can be taken to represent the relative adsorption affinities of the functional groups. Deviations from the above sequence are to be observed only when the adsorption behaviour of the functional groups is very similar.Strain 2 has drawn similar conclusions using carotinoids. The knowledge that adsorption affinity depends largely on the number and nature of the functional groups only when the skeleton of the molecules is not very different enables one to predict in many cases whether chromatographic separation is possible. CHROMATOGRAPHY OF COLOURLESS SUBSTANCES Colour of the zones R = -C6H*.N:N.CeH4- Flesh-coloured Bluish red Purple Blue Reddish brown Pale red (i) H-R-NH (ii) (CH,),N-R-N(CH,) (iii) H-R-N-(CH,) (iv) Br-R-N(CH,) (v) H-R-NH.CO.CH (vi) H-R-NH. co. C ,H eluted yellow zone. 64 Water-soluble Salts as Adsorbents.-Anhydrous copper sulphate has shown itself t o be particularly well suited to the separation of azobenzene derivatives.The following azobenzene derivatives were adsorbed from benzene-light petroleum r/4 and were eluted with the same solvent mixture or with benzene alone the compounds were found to be adsorbed in the following order of decreasing adsorption affinity. The following azobenzene derivatives which cannot be separated on alumina are easily separated on anhydrous copper sulphate. (I) R-COOCH, R-N(CH,),. R-N (CH,) is well adsorbed from benzene onto copper sulphate R-COOCH on the other hand forms an easily ( 2 ) R-NH, R-NH.CO.C,H,. R-NH is firmly adsorbed onto CuSO from benzene as a flesh-coloured zone from which R-NH.CO.C,H is easily separated by elution. (3) R-NH, R-0-CO.CH,. R-NH is adsorbed from benzene much more firmly than R-0-CO.CH,.(4) R-NO, R-N(CH,),. The purple zone of R-N(CH,) is much more firmly adsorbed than the orange zone of the nitro compound. Anhydrous zinc manganese aluminium and magnesium sulphates can also bc used for the separation of azobenzene derivatives. We believe that it is probable that the addition compounds of the type (I) analogous with salts of P-amino-azobenzene (11) are formed on the surface of the adsorbent. The above-mentioned salts particularly aluminium sulphate can be used for the separation of other compounds such as hydroxy-anthraquinones. Compounds that are very firmly adsorbed onto these salts can be isolated simply by dissolving the inorganic material in water and extracting the organic compounds with a suitable solvent. Organisch-Chemisches Institut der Universitat Hospitalstrasse 8-1 I Gottingen Germany.CHROMATOGRAPHY OF COLOURLESS SUBSTANCES AND THE RELATION BETWEEN CONSTITUTION AND ADSORPTION AFFINITY BY H. BROCKMANN Received 18th July 1949 h method for the chromatographic separation of colourless substances based on 'the use of fluorescent adsorbents is described. Relations between adsorption affinity and chemical constitution of azobenzene and stilbene derivatives are demonstrated. The dependence of adsorption affinity on the nature of the adsorbent and solvent was investigated. Water soluble salts like copper sulphate or zinc sulphate were found t o be useful adsorbents for the chromatographic separation of azobenzene compounds. Chromatography of Colourless Substances.l-The adsorption zones s f colourless substances can be followed by the use of fluorescent adsorbents. Ultra-violet light of a wavelength absorbed by the adsorbed substance is employed. The part of the column containing adsorbed substance then appears as a dark band. Adsorbents can be made to fluoresce by the addition of small quantities of irreversibly adsorbed fluorescing compounds. Morin is very satisfactory for alumina magnesium oxide and calcium carbonate, and berberine for silica. Salicylic acid and z-hydroxynaphthalene-3-carboxylic acid can also be used. Experimental.-A very convenient method for the preparation of a fluorescent column consists in the mixing of the adsorbent with z yo to 5 yo of finely powdered luminous paint. Calcium sulphate dihydrate prepared by the precipitation of calcium chloride with the calculated quantity of sulphuric acid a t 8oo-goo C can be converted into a valuable adsorbent for colourless compounds and also incidentally an excellent one for coloured ones.Heating a t temperatures between 150' and zooo C yields three different products the hemihydrate has an activity equivalent to that of alumina IV; the soluble anhydrous salt can be obtained in two grades, one equivalent to alumina I11 and the other to alumina 11. The product obtained by heating above 250' is only slightly active and is of no practical value. The large adsorption affinity of the soluble anhydrite is very likely to be due t o numerous faults in the crystal lattice. The fluorescent chromatograms were illuminated with light of 366 mp from a mercury vapour lamp and an appropriate filter.For compounds which absorb a t shorter wavelengths a Chlorfilter was used with windows of UGS glass (Schott, Jena) which transmits only the Hg line a t 253-7 mp. In our most recent experi-ments on the separation of colourless compounds the following apparatus was used. Light from an electric arc (iron electrodes) was passed through a slit 35 mm. long and 3-4 mm. wide a quartz lens 5 cm. thick and a quartz or rock salt prism 50 mm. high. The resulting line spectrum (length of lines 50 mm.) was projected onto a fluorescent screen ; after focusing the fluorescent screen was replaced by a cardboard sheet on which the position of the spectral lines in the ultra-violet spectrum had been marked.The adsorption column con-tained in a quartz tube was placed in front of the cardboard and moved until the optimum contrast was obtained making the bands of the chromatogram clearly recognizable. It is advisable to employ a combination of adsorbent and fluorescent compound which does not continue to emit light after illumination (luminous paint N4 green I Francke Frankfurt was used). This method is generally applicable H. BROCKMANN 59 Appearance of Zones.-There are two ways in which the zones can be made apparent (i) in th.e case of adsorbents rendered fluorescent by adsorbed morin or berberine the zones become apparent through extinction of the fluorescence by the adsorbed compound; (ii) adsorbents which have been mixed with luminous paint behave differently.In this case the zones appear as follows. The particles of the adsorbent allow the light to penetrate the column to a certain extent and fluorescence excited there is emitted. Actual adsorption of the light is stronger at places where the adsorbed compound is present so that the fluorescing material in this case the luminous paint particles is not as strongly excited as in those places where the column is free luminous paint particles thus act as small fluorescent screens. Experimental.-That the outer layer of adsorbent allows light to penetrate is shown as follows. A test-tube the diameter of which is such that i t fits in the chromatographic tube with 3-5 mm. clearance all round is filled with luminous paint and placed in the tube. The space between is filled with adsorbent and a colourless substance is adsorbed onto it.On illumination with ultra-violet light the zone appears surprisingly clearly. This filter action of the outer layer of the column can also play a part in the appearance of the zones when adsorbent materials containing adsorbed morin or berberine are employed. In this case each particle of adsorbent acts as a small fluorescent screen. Whether an adsorbed colourless substance is detected through extinction of the fluorescence or through the filter effect can be decided in some cases by comparing the behaviour of an adsorbent + luminous paint with that of the adsorbent + morin or berberine. Table I gives the results of such a comparison. TABLE I -~ ~~~~~~~~ I hl,O,-Luminous paint 1 Al,O,-Morin Benzaldehyde .. Acetophenone . . P-Tolylaldehyde . . : ~ Anisaldehyde I Non-visible Visible # * ,? I , A mercury lamp with a filter allowing the transmission of light of wavelength 360-370 mp was used. Since the above compounds hardly absorb a t 365 m p the zones do not appear when adsorbent + luminous paint is employed. The appearance of the zones when adsorbent + morin is used must therefore be due t o extinction of fluorescence. The effectiveness of the extinction is greater in the case of morin than that of berberine. Thus the phenacyl esters of aliphatic carboxylic acids which hardly absorb a t 365 mp are visible with light of this wavelength on Al,O,-morin but not on silica-berberine. Relations Between Constitution and Adsorption Affinity.-The adsorption affinity that is the firmness with which organic compounds are adsorbed from non-aqueous solvents inter alia depends on the nature of the skeleton and of the functional groups of the adsorbed compound.In order to find the effect of the functional groups on adsorption affinity, several compounds containing the same basic skeleton but differing only in one functional group were investigated. The first experiments were carried out with $-substituted stilbenes and azobenzenes. These compounds were adsorbed on A1,0 from benzene or from carbon tetrachloride. Since stilbene and azobenzene under these conditions are only very weakly adsorbed the adsorption affinity of the derivatives can be taken t o be almost entirely dependent on the nature of the substituents.The functional groups in Table I1 are arranged in order of decreasing adsorption affinity 60 CHROMATOGRAPHY OF COLOURLESS SUBSTANCES Compounds separated by a horizontal dash could be separated the others were not fully separable. The two series are in almost complete agreement with each other. Differences are only found in the three unnumbered acyl derivatives on the right-hand side those in the azobenzene series lie higher than those in the stilbene series; the order however remains the R-COOH R-CONH ~ R-OH - -_____ R-NH R-NH-AC R-0-AC -____-R-COOCH, R-N(CH3) 2 R-0-BZ R-NO -___ R-ocH,- ____.__ R-H R-COOH R-OH R-NH-AC R-0-AC R-NH, R-0-BZ R-COOCH, R-N(CH3) 2 R-NO R-OCH, R-H same. This difference is probably due to the fact that the azobenzene derivatives were adsorbed from benzene and the stilbene derivatives from carbon tetrachloride.The solubility of the azobenzene derivatives in carbon tetrachloride was unfortunately too small for adsorption experiments to be carried out using this solvent. Dependence of Adsorption Affinity on the Adsorbent.-In the adsorption of organic compounds from non-aqueous solvents only certain points on the adsorbent particle (the so-called active points) play a part. These are mainly at positions where the crystal lattice is faulty. The adsorption activity of an adsorbent therefore does not only depend on its chemical nature but also on the nature of its crystal lattice. a-Al,O, does not adsorb whereas y-Al,O and Bohmite are very effective as adsorbents.CaSO, like the " soluble " anhydrous salt possesses the same crystal lattice as the hemihydrate ; it shows good adsorption properties, and is in the light of our experience very efficient in the separation of hydroxy-anthraquinones. As the lattice of the soluble anhydrite is changed into that of the insoluble anhydrite by heating to 2joo the capacity for adsorption falls very considerably. We have previously shown that y-Al,O can be obtained in five different grades (by treatment with water vapour or by regulated heating) which can differentiate by adsorption of azo dyes under standardized conditions. The employment of A1,0 with different and reproducible activities has proved so useful that we have attempted to prepare graded specimens of other adsorbents.By varying its water content bentonite was obtained in five grades of activity silica gel and precipitated silica in three CaSO and MgO also in three and CaCO in two different grades. The standardization of these adsorbents was carried out using the same dyes as in the case of alumina. In the course of these experiments we have tried to find out whether with the same solvents our test dyes were adsorbed in the same order as on A1,0,. The following experiments show this not to be the case H. BROCKMANN 61 Adsorption Sequence of Azo Dyestuffs in Columns of Different Adsorbents. DYESTUFFS $-hydroxyazobenzene 9-amino-azobenzene Sudan Red (0-CH,).C,H,.N:N.C,H,(o-CH,).N:N.C,,H,.OH Sudan Yellow C,H,.N:N, C,,H,. OH $-met hoxyazobenzene. The most strongly adsorbed dyestuff is listed at the top of each column, the least adsorbed at the bottom.SOLVENT Benzene-Petrol ether r/4. A1203 ~ _ _ - SiO nlg2 Hydroxyazobenzene Sudan Red H ydroxyazobenzene Amino-azobenzene Oxyazobenzene Sudan Yellow Sudan Red Sudan Red Amino-azobenzene Amino-azobenzene Sudan Yellow Sudan Yellow Met hoxyazobenzene Met hoxyazobenzene Met hoxyazobenzene CaSO ~~ __-_ c u s o , Hydroxyazobenzene Sudan Yellow Amino-azobenzene Amino-azobenzene Sudan Red Sudan Red Hydroxy azobenzene Methoxyazobenzene Sudan Yellow Met hoxyazobenzene Effect of the Acid or Alkaline Character of the Adsorbent on the Adsorption Sequence.-The adsorption sequence of acidic or basic dyes is influenced by the acidity or alkalinity of the adsorbent as the following observations show.SOLVENT Benzene-Petrol ether r/4. A1,0 treated with HCl .- ______ Al,O alkaline ________ H ydroxyazobenzene Hydroxyazobenzene Amino-azobenzene Sudan Red Sudan Red Amino-azobenzene Sudan Yellow Sudan Yellow Methoxyazobenzene Methoxyazobenzene Si02 treated with NaOH Sudan Red Amino-azobenzene Hydroxy azobenzene Sudan Yellow Met hoxyazobenzene - ~ _ _ ~ _ _ _ - - SiO treated with HC1 Sudan Red H ydrox y azobenzene Amino-azobenzene Sudan Yellow Met hoxyazobenzene Effect of Different Solvents on the Adsorption Sequence.-The firmness with which a compound is adsorbed from a non-aqueous solvent depends on the nature of the solvent in two ways (i) the solvent tends to occupy the active points of the adsorbent ; (ii) the solvent reacts with the dissolved substance (solvation).Both these effects are negligible with non-polar solvents such as petrol ether cyclohexane carbon tetrachloride and benzene but considerable with polar solvents. The effect of various solvents on the adsorption sequence of dyestuffs is shown in the following results 62 CHROMATOGRAPHY OF COLOURLESS SUBSTANCES ADSORBENT Al,O, Benzene-Petrol ether 114 Carbon tetrachloride -Ether H ydroxyazobenzene Sudan Yellow Hydroxyazobenzene Amino-azobenzene Sudan Red Sudan Red Amino-azobenzene Sudan Yellow Met hoxyazobenzene Methoxyazobenzene Chloroform ____ Tetrahydrofuran Dioxan Cyclohexanone H ydroxyazobenzene H ydroxyazobenzene Amino-azobenzene Sudan Red Sudan Yellow Sudan Yellow Amino-azobenzene Sudan Red Met hoxyazobenzene Met hoxyazobenzene ADSORBENT CaSO, Benzene-Petrol ether r/4 Ether Sudan Yellow hydroxyazobenzene Amino-azobenzene Amino-azobenzene Sudan Red Sudan Red Met hoxyazobenzene Met hoxyazobenzene Sudan Yellow hydroxyazobenzene Chloroform Sudan Yellow hydroxyazobenzene Amino-azobenzene Sudan Red Met hoxyazobenzene There does not seem to be a relationship between the adsorption sequence and the solubility of the test dyestuffs.It is remarkable that Sudan Yellow and hydroxyazobenzene are adsorbed equally firmly from ether. A similar effect with ether as solvent was. observed in the case of the three following Sudan dyes. These compounds are adsorbed on Al,O from benzene-liglit petroleum 1/4 in the order (i) (ii) (iii) but from ether in the order (ii) (iii).,.(I) H. BROCKMANN 63 Even though no satisfactory explanation can at the moment be given of the effect of the solvent on the adsorption sequence our observations lead to the following procedure. In the case of a complex mixture which cannot be separated by the use of one adsorbent and one eluent it has been suggested that separation may be effected by adsorption from a non-polar solvent onto a strong adsorbent followed by elution with a series of solvents of increasing adsorption affinity. We believe that this procedure when applied to compounds such as our dyestuffs which in the adsorption sequence are s o dependent on the nature of the solvent is not as satisfactory as a procedure, which in our experience has proved to be of value and which involves repeated chromatography of the mixture from one non-polar solvent on several adsorbents of varying activity.The Adsorption Sequence which is independent of both Adsorbent and Solvent.-The marked dependence of the adsorption sequence of our dyestuffs on the nature of the solvent is perhaps due to variations in the solvation of the hydroxyl and amino groups present. Furthermore it is possible that depending on the nature of the solvent the dyestuffs are adsorbed partly in the azobenzene form partly in the tautomeric quinone-imine form. These complications are avoided if the following compounds are used. The adsorption affinity decreases as one goes down the series. R = C6H,.N:N.C6H4 (i) R-NH.CO.CH, (ii) R-NH. co .c ,H, (iii) R-0-CO.CH or 0,N-R-N(CH,), (iv) R-COOCH, (vi) R-OCH, (v) R-NO, Instead of dyestuff (iii) which will be partly hydrolyzed on alkaline adsorbents, p-bromo-P’-dimethylamino-azobenzene was employed.This compound as a result of polarized adsorption forms a bluish-violet zone. These compounds were always adsorbed in the same order from the following solvents on the following adsorbents. A1,03 benzene-petrol ether 1/4 ether chloroform. From ether separation of (i) and (ii) is incomplete. SiO benzene-petrol ether 1/4 ether chloroform. CaSO benzene-petrol ether 1/4 benzene, ether. hlg0 benzene-petrol ether 1/4. CuSO benzene-petrol ether. From the above results we believe that we may draw the following conclusions concerning the relations between constitution and adsorption.The adsorption affinity of a compound is practically an additive function of the adsorption affinities of the basic skeleton and of the functional groups only when there is no possibility of tautomerisrn and when solvation by the solvent employed does not occur. The above-mentioned series can be taken to represent the relative adsorption affinities of the functional groups. Deviations from the above sequence are to be observed only when the adsorption behaviour of the functional groups is very similar. Strain 2 has drawn similar conclusions using carotinoids. The knowledge that adsorption affinity depends largely on the number and nature of the functional groups only when the skeleton of the molecules is not very different enables one to predict in many cases whether chromatographic separation is possible 64 CHROMATOGRAPHY OF COLOURLESS SUBSTANCES Water-soluble Salts as Adsorbents.-Anhydrous copper sulphate has shown itself t o be particularly well suited to the separation of azobenzene derivatives.The following azobenzene derivatives were adsorbed from benzene-light petroleum r/4 and were eluted with the same solvent mixture or with benzene alone the compounds were found to be adsorbed in the following order of decreasing adsorption affinity. R = -C6H*.N:N.CeH4- Colour of the zones (i) H-R-NH, (ii) (CH,),N-R-N(CH,), (iii) H-R-N-(CH,), (iv) Br-R-N(CH,), (vi) H-R-NH. co. C ,H (v) H-R-NH.CO.CH, Flesh-coloured Bluish red Purple Blue Reddish brown Pale red The following azobenzene derivatives which cannot be separated on alumina are easily separated on anhydrous copper sulphate. (I) R-COOCH, R-N(CH,),. R-N (CH,) is well adsorbed from benzene onto copper sulphate R-COOCH on the other hand forms an easily eluted yellow zone. R-NH is firmly adsorbed onto CuSO, from benzene as a flesh-coloured zone from which R-NH.CO.C,H, is easily separated by elution. (3) R-NH, R-0-CO.CH,. R-NH is adsorbed from benzene much more firmly than R-0-CO.CH,. (4) R-NO, R-N(CH,),. The purple zone of R-N(CH,) is much more firmly adsorbed than the orange zone of the nitro compound. Anhydrous zinc manganese aluminium and magnesium sulphates can also bc used for the separation of azobenzene derivatives. We believe that it is probable that the addition compounds of the type (I) analogous with salts of P-amino-azobenzene (11) are formed on the surface of the adsorbent. The above-mentioned salts particularly aluminium sulphate can be used for the separation of other compounds such as hydroxy-anthraquinones. Compounds that are very firmly adsorbed onto these salts can be isolated simply by dissolving the inorganic material in water and extracting the organic compounds with a suitable solvent. ( 2 ) R-NH, R-NH.CO.C,H,. Organisch-Chemisches Institut der Universitat, Hospitalstrasse 8-1 I, Gottingen Germany

ISSN:0366-9033    

DOI:10.1039/DF9490700058

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

THE FUNCTION OF ADSORBENT ACTIVITY IN THE CHROMATOGRAPHIC SEPARATION OF CERTAIN ANTHRAQUINONE COMPOUNDS BY A. STEWART 8 Miiller Vet%. Ver. Schweiz. Physiol. 1942 21 29; Helv. chim. A d a 1943 26 1945; 19443 27 404. Schroeder Ann. N.Y. Acad. Sci. 1948 49 204. Received 8th August 1949 The activity oi adsorbents used in chromatography with organic solvents is easily controlled by preliminary incorporation of water in the adsorbent. The effect on the band movement of simple anthraquinone compounds in toluene on activated alumina and unactivated magnesium carbonate is described. The conditions for preparing activated alumina by heating between 250' and 500' C have considerably less effect on the behaviour than has the deactivation treatment. The band movement measure- ments were made under satisfactory comparative conditions but are affected by several errors ; in particular the greater solvent movement in mid-column than a t the periphery renders values obtained a t the periphery 11 % low.The distribution of the anthraquinone compounds between solvent and partially deactivated alumina is independent of concentration. The length of the initial chromato- graph band and the mean rate of band movement are in satisfactory accord with theory. The considerable band widening in passage through a column is independent of the adsorption little affected by the rate of solvent flow and may be caused by non- uniformity of flow through the interstitial capillary spaces ; its effect on the separation o f a mixture causes the various bands to have approximately equal width a t the same position in the column.The relation between constitution and chromatographic behaviour of simple anthra- quinone compounds largely influenced by the conditions used is briefly described. The successful application of adsorption chromatography ultimately depends on the empirical selection of operating conditions if for no other reason than that the composition of the mixtures examined is rarely fully known until the separation has been achieved. Various methods have been described for the partial deactivation of adsorbents 1-5 and it is the purpose of this paper to draw attention to the simplicity of this operation and the wide range of control exercised on the band movement enabling this variable to be systematically applied in selecting the operating conditions.To illustrate the behaviour band movement measurements have been made using the highly coloured simple anthraquinone compounds in toluene solution ; these conditions do not necessarily represent the optimum for such separations for which their fairly low solubility is often a controlling factor. The effect on activated alumina and magnesium carbonate adsor- bents is described; it has been applied to other adsorbents but with the wide range of control obtained on deactivation these adsorbents cover most of the activity range required when the separation is carried out from organic solvent solution. The effect of variation in the conditions for the activation of alumina is shown to have much less effect on the behaviour than has the subsequent deactivation.Partial deactivation of activated alumina gives a distribution between adsorbent and solvent which is substantially independent of the concen- 1 Zechnieister and Cholnoky trans. Bacharach and Robinson. Principles a??d Practice of Clzromatografihy. (Chapman & Hall London 1941 p. 48.) 2 Brockmann and Schodder Ber. 1941 74 73. Stewart and I.C.I. Ltd. Brit. Pat. 565J405. C 65 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 66 tration and under these conditions the mean band movement and the length of initial band are shown to be in agreement with the theory described by LeRosen and others.'-1° Conditions affecting the widening of bands often a limiting factor in the chromatographic separation are considered.of adsorbent/equilibrium concentration of solution in gram per gram this value being substantially constant under the conditions employed. Distribution constant prevailing in unit length of the column Theoretical The symbols used have the following significance L . . . . Weight of interstitial solvent per unit length of the column. . . Weight of adsorbent per unit length of the column. K . . . . Distribution constant = the weight of solute adsorbed per gram 171 . . c . . . . Concentration of solution. 'W . . (top) of the column. Distance the developing solvent has moved in the column. . . Initial concentration of solution added to column. x . . . . Length of the initial band formed by adding w to column.x ~ x t D . . Go . . D,"B referring to initial lengths of components a and p. . . Length of column occupied by interstitial solvent of weight w,. Distance the developing solvent requires to move to separate two bands starting from xg and x!. R Displacement of zone on column/displacement of solvent in * Do . . As LeRosen and others '-lo have shown where the distribution between adsorbent and solvent is independent of concentration and is rapidly established the conditions for the formation and separation of the bands can be evaluated. If a solution of concentration c of a pure compound is added to a column in amount 'wo equal to length Do of interstitial solvent the ratio of inter- stitial solvent to adsorbent in unit length of the column being i/m then the solute is adsorbed until it reaches equilibrium with the initial concentra- tion of the solution and a band of length x is formed such that X,C + K'x,c = Doco or x = D,/(I + K') = D,/(I + Kw/i) .(1) If pure solvent is now allowed to flow through the column then the band moves down at a fraction of the rate of linear movement of the solvent equal to the fraction of the total substance present in the interstitial solvent. The movement of the band relative to the linear movement of solvent through the column R is given by or (4 (3) 6 LeRosen J . Amev. CJLem. SOC. 1947 69 87. 7 De Vault J . Amev. Chem. SOL 1943 65 5 3 2 . K' = Kmli. . . Weight of solution. x . . . . Distance of any point in the column measured from the beginning column R = dx/dD.R for the leading edge of a band ; R for the following edge ; RM for the mean of RL and R,. RM expressed as a percentage of the solvent movement. Extent of water deactivation; ml. water addition per IOO g. adsorbent the latter including any moisture initially present. ,4 measure of the widening of a band edge in passage through the column. R = %,/Do = I/(I + K') = I/(I + Km/i) . I(' = (I - R)/R and K = [(I - R)/R] i/m . * 8 Weiss J. Chem. Soc. 1943 297. SGlueckauf J . Chern. SOL 1947 1302. 10 \Veil-Malherbe J . Chem. SOC. 1943 303 A. STEWART 67 The movement of the chromatograph bands through a column is con- veniently expressed l1 as the relative movement 2'2 which is given here as a percentage of the solvent movement.The value depends on the ratio of solvent to adsorbent ilm for the particular column so that the results obtained from different columns require to be calculated on the basis of eqn. (2) and (3) to correspond with a particular value of i/m. (4) FIG. 1.-Effect of water deactivation treatment applied to activated alumina on the band movements of anthraquinone compounds in toluene. Curve I . I -Chloroanthraquinone. 2. I -Methylaminoanthraquinone . 3. I -Aminoanthraquinone. . 4. I-Dimethylaminoanthraquinone. 5. I-Amino-4-methylaminoanthraquinone. , 6 . 2-Aminoanthraquinone. , 7. I 4-Diaminoanthraquinone. , 8. I 4 5-Triaminoanthraquinone. Under the above conditions the band should move down the column without changing its length the small region in front and rear where adsorption and desorption occur remaining constant.The position of the (5) * - following and leading edges of a band is given by Following edge x = x D/Do Leading edge x = x + x D/Do . 11 LeRosen J . Aupzev. Chem. SOC. 1942 64 1905. ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 68 If a second component is present in the initial solution and there is no mutual interference in the distribution the bands will draw apart and commence to separate when the following edge of the faster reaches the leading edge of the slower the actual position on the column being given by RFD,. x! Ds/Do = x + x D,/D or D;P/Do = x;/(x{ - q) . (6) Certain complications limit the simple application of the above relations.The leading edge of a band moves considerably (20 %-30 %) faster than the following edge and the diminishing concentration at the band edges increases the difficulty of measuring the band movement and determining the position of separation. The bands move at a different rate in mid-column than at the periphery probably 16 yo faster in the columns considered below. FIG. z .-Eff ect of water deactivation treatment applied to unactivated magnesium carbonate on the band movement of anthraquinone compounds in toluene. Curve I. I -Hydroxy-4-~-toluidinoanthraquinone. , 2. r 4 5-Triaminoanthraquinone. , 3 . I 4 5 8-Tetraminoanthraquinone. The solvent flow required to produce the initial separation should depend on the ratio of the lower R value to the difference in R values while the position at which this separation occurs and the actual rate at which the bands subsequently move apart depend on the value of R.69 A. STEWART Experimental The Control of Adsorbent Activity by Partial Deactivation .-The preferred method of deactivation consists in adding liquid water to the adsorbent shaking to distribute the soft wet portion initially formed and mixing for 2 hr. conveniently by rotating the container. The addition necessary causes little or no change in the handling properties of the adsorbents ; the method is easy to carry out and gives a reproducible behaviour. The deactivation is expressed as ml. water addition per IOO g. adsorbent abbreviated to yo H the weight of adsorbent including any moisture initially present.The effect of partial deactivation on the band movement of simple anthra- quinone compounds in toluene on activated alumina (Type 0 ; P. Spence & Sons Ltd.) and on unactivated magnesium carbonate (Ponderous. B.P. quality ; Cuxson Gerrard and Co. Ltd.) is shown in Fig. I and 2 the band movement being given as the mean of the leading and following edge movement yo €2~1 determined and corrected for variation in column packing as described below. This reproducible behaviour enables the rate of movement to be controlled over a wide range so that in conjunction with a suitable choice of organic solvent and a sufficiently high initial adsorbent activity most separations can be obtained at a convenient rate. In practice the two adsorbents mentioned which are of satisfactory grain size for chromatographic use meet most requirements.Inter- FIG. 3.-The effect of water deactivation treatment applied to activated alumina on the distribution constant of anthraquinone compounds between toluene and adsorbent. The curves are numbered as in Fig. I. ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 70 mediate values of the deactivation tend to give most effective separations as the maximum increase of movement with deactivation generally occurs at low yo H values for weakly adsorbed compounds and a t higher values for those more strongly adsorbed. Occasionally the order of the band movement changes with the deactivation generally at low values.of yo H (Fig. 3). Partial deactivation of activated alumina gives a distribution between adsorbent and solvent sub- stantially independent of concentration and bands do not form long trails at one or other edge when this applies.When the solvent used has affinity for water equilibrium appears to be estab- lished with the deactivated adsorbent and any initial water present. in the solvent affects the chromatographic behaviour to some extent. For separalions from an aqueous medium this deactivation treatment is ineffective. FIG. 4.-Band behaviour of the test mixture in toluene on alumina activated a t 330' C for 4 hr. Curve 3 . Purple band. , 4. Violet band. Curvc I . Red band. , 2. Blue band. Fig. 3 shows the effect of partial deactivation on the distribution constant calculated from the yo RM values shown in Fig.I by means of eqn. (3). There is a very considerable change in the distribution with practically a linear relation between log K and yo H when the deactivation exceeds 2-5 yo H . The Effect of Conditions of Activation of Alumina on the Deactivation Behaviour .-Activated alumina may be prepared la by heating aluminium trihydroxide at atmospheric pressure the product y alumina retaining some water to an extent varying with the activation temperature; it may also contain A. STEWART 400 330-360 330 4 4 330 a 4 8.8 24 5'3 7.9 3'9 71 alkaline impurities l3 which can if necessary be removed but these rarely upset the chromatographic behaviour. Satisfactory commercial products are available which obviate the rather troublesome activation.To find the effect of activation conditions on the behaviour after water deactivation dried hydrate of alumina (British Aluminium Co. Ltd.) was heated in an open pan at temperatures and for periods on temperature given below together with the moisture content of the products determined by ignition to about 2300' C. 500-550 4 I.7 Temperature . . " C 250 Time hr. Moisturecontent % 11.3 Each sample was deactivated with from 1.5 % to 15 % water addition (% H ) and 9 2 , 2 . FIG. 5.-Effect of conditions of activation on the relative movement of the violet band. Curve I . Activated at 250° C 4 hr. 330°c 4 9 9 t I 0 1 , > 1 , 3. 330° c 8 , , 330"-360" C 24 , 400°c 4 , , 4. , 500°-5500c 4 , band movements measured using a toluene solution of four components which gave distinctive highly coloured bands covering a convenient range of adsorption affinity.The 10 ml. solution added to each column contained 0.77 mg. I -Methylaminoanthraquinone (red least adsorbed band) 0.34 mg. I 4-Dimethylaminoanthraquinone (blue band) I -28 mg. I -Amino-q-methylaminoanthraquinone (purple band) 0.58 mg. I 4-Diaminoanthraquinone (violet most adsorbed band) Is Siewert and Jungnickel A.C.S. Abstr. 1943 37 5898. ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 72 the actual concentration of the components being unimportant provided sufficient is present to give a highly coloured band. Fig. 4 shows the band behaviour of these components on one sample of adsorbent while Fig.5 shows the effect of conditions of activation on the behaviour of the most strongly adsorbed com- ponent the results for the others being similarly distributed and displaced to an extent indicated in Fig. 4. The curves in Fig. 5 show the effect of activation temperature the effect of the heating period between 4 and 24 hr. a t 330' C being within the experimental error. The activation conditions employed have considerably less effect on the be- haviour than has the subsequent deactivation treatment. There is no direct relation between the initial moisture content and the water added on deactivation e g . samples prepared a t 250° and 530" C differ by 9.6 yo in initial moisture but give practically the same relative band movement when a further 10 yo water is added to both.For low values of yo H the band movement increases with activation temperature but the extent to which the band movement increases on further addition of water decreases with activation temperature. Qgalitative tests show that activation temperatures above 5o0°-5500 C give a progressive decrease in activity. The quality of activated alumina probably depends to some extent on the nature of the starting material and other factors but the general behaviour on deactivation is similar FIG. 6.-Effect of the correction for variation in the ratio of solvent to adsorbent in alumina columns ( x ) and curves show the uncorrected values ; (e) show the corrected values corresponding to the curves shown in Fig. 4. The i/m values were 1-5 yo H 0.580 ; 2.5 yo H 0.583 ; 5.0 yo H 0.542 ; 10 % H 0.462 ; 15 ?< H 0.425.The Measurement of Band Movement.-The tubes used were z cm. diam. and 55 cm. long having a compact cotton wool retaining plug. Well- packed columns free from air bubbles were obtained from 150 g . alumina or A. STEWART 73 70 g. magnesium carbonate which were added as a slurry in toluene and settled with assistance of intermittent tapping to a constant column length before use ; the interstitial solvent was determined by weighing the column before and after adding the adsorbent. The temperature remained close to 20' C throughout and no temperature correction was applied. The solvent used was high-grade technical toluene which had been passed through a column of the same adsorbent so as t o minimize any possible effect on the activity.The solution under test was added (10 ml. of 0.1 % conc.) when the solvent level had dropped to the adsorption surface and as soon as this entered the column i t was washed in with several small (0.5-1 ml.) lots of solvent before raising the solvent level to about 10 cm. above the adsorbent where i t was maintained throughout the measurement by means of a constant level arrangement. Care was taken a t all stages not to disturb the adsorbent or allow air to enter the column. Gravity flow was employed and was measured by weight. The position of the band edges and the solvent f l o ~ . gave satisfactory linear relations against time from which the RL and RF values were obtained by means of the previously determined length of column occupied by I g.of interstitial solvent and these are given throughout this paper as a percentage of the solvent movement. A typical column was of 3.303 cm.2 cross-sectional area and 43.9 cm. length L containing 150.3 g . (m x L) activated alumina of 5 yo H and 81-29 g. (i x L) interstitial solvent; the latter occupies 0-540 cm. per g. solvent and for the observed flow of 54-34 g./hr. the linear movement of solvent was 29.29 cm./hr. The solvent movement varied between 23 and 45 average 28.8 cm./hr. for the various columns. The value of i/m was fairly constant for a single sample of adsorbent but varied from 0.6 to 0.4 as the yo H increased from 1.5 to 15 yo the mean value being close to 0.5. Values of R corresponding to this mean value of i/m were obtained from the measurements on alumina columns from The magnesium carbonate columns gave a mean value for ilnz of 1.17 which was used in place of 0.5 for their correction.The effect of the correction for i/rn is shown in Fig. 6 where the uncorrected values corresponding to Fig. 4 are shown by ( x ) with curves drawn through these points and the corrected values-corresponding to the curves shown in Fig. 4 -are indicated by ( ). The correction is relatively small and without significant effect on the general relation between band movement and adsorbent activity. The reproducibility of the band movement measurement was tested in a series of columns using a band having R 51 % ; the standard deviation for the RL and R values was 3'3 and 2-7 and for R 2.0.RL/RM = 1-15 with standard deviation 0.06. The behaviour of these columns was found sub- stantially unchanged after use in a series of measurements The reproducibility is reasonably satisfactory for comparative purposes involving fairly large differences but apart from errors due to the difficulty of observing the exact position of the band edge and the occasional inclination of the bands there is a further error arising from the bands and presumably the solvent moving faster in mid-column than a t the periphery. Since a uniform solvent movement over the cross-sectional area of the column has been assumed the RM values are too low results given below indicating this error to be about 11 yo but no correction has been applied for this factor.Provided the com- parisons are carried out in similar columns as in the present case this source of error should remain reasonably constant. The Relation between Adsorption and Band Movement.-Preliminary adsorption measurements were made allowing a solvent solution to reach equilibrium with an alumina adsorbent and determining the equilibrium con- centration but this procedure was limited to low values of the latter by the loti solubility of the compounds used and the photometric method employed. The results fitted the Freundlich adsorption equation log (specific adsorption) = log,,K + a log, (equilibrium concentration) f* -ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 74 cc being about 0.75 for activated alumina but increasing to 1-0 on partial deactiva- tion with water and becoming slightly greater than I for weak adsorption.Within a rather large experimental error there was agreement between the observed RM value and that calculated from K by means of eqn. (2). To obtain measurements at higher concentrations a 2-cm. column packed with 50 g. partially deactivated Type 0 alumina and of known interstitial volume was used the solution being fed through the column until the issuing solution was a t the initial concentration and the amount adsorbed determined TABLE I THE RELATION BETWEEN THE BAND MOVEMENT CALCULATED FROM ADSORPTION MEASUREMENT AND THE OBSERVED VALUE Adsorbent Partially deactivated Type 0 Activated Alumina. % H = ml. water added per roo g. adsorbent. measurement for band washing out.from peripheral Gbservation. -4 dsorbate Conditions 62.5 , 33'2 , 16.3 , 4'4 9 . . I 4-Di-p-toluidino- Toluene 5 % H anthraquinone Conc. 46 mg. % I -Methylamino- anthraqu inone Toluene 5 yo H Conc. 115 mg. yo I -Amino-4-niethyl- aminoanthraquinone , I ? I -Amino-;?-methyl- Toluene 5 % H anthraquinone Conc. 46.2 mg. 7,; 5 % H 2 s I Toluene 5 yo H Conc. 23.1 mg. 04 Benzene Conc. 4.0 mg. % 10 % H 15 % H I 4-Diaminoanthra- Toluene 5 % H quinone Conc. 7.5 mg. % by difference and also by washing the band from the column allowance being made for the solution held by the cotton wool retaining plug. The relative band movement was determined in the same column by observing the peripheral movement and in some cases by observing when the band started and finished washing from the column.The results are summarized in Table I. There is reasonable agreement between the band movement calculated from adsorption measurements and the observed Rnf values the values obtained % R obs. % R calc. from adsorption 68.7 63.1 63'3 47'5 47'2 46.4 40.1 ;:;) 9'3 17'7 46.2 80.1 RF 55'8 51'4 51.6 38.0 37'0 36-6 26.8 29'2 27'4 28.2 27'9 29'0 7'2 7'5 14'1 35'3 60.7 7'0 31.6 38.8 30.6 30.1 37'6 10-3 8.6 I 8.2 43'1 83'5 7'1 by washing from the column being 5-1 yo high and those obtained from peripheral observations 11.1 yo low ; as indicated above the latter are likely to be too low due to non-uniformity of solvent flow across the cross-sectional area of the column.The results confirm that band movement on the partially deacti- vated alumina adsorbent used is independent of concentration. A considerable difference is shown between the band movement of the leading and following edges which in the case of the peripheral results is a measure of the band widening further considered below the somewhat greater difference obtained when the band is washed out represents the widening between the leading edge in mid- column and the following edge a t the periphery. The Widening of Bands.-(i) The leading and following edges of a band move down at a uniform rate and the widening of each edge is given by RL/Rlw.The widening is not uniform over the whole band but is greater towards the edges than for the main deeply coloured portion the movement of I-methyl- aminoanthraquinone in toluene on Type 0 5 yo H alumina being 75 Edges of band Main portion of band (ii) The RL/RM value is substantially the same for the bands of all components present in a column independent of the R value and is little affected by the extent of deactivation. Table I1 gives results from a set of measurements on the activation of alumina using the four component mixture as described above which are typical of the six samples examined in this connection. . . . . 43-2 R F . . 45.1 Column Deactivation RL/RH FOR THE FOUR COMPONENT MIXTURE ON FIVE 2-CM.COLUMNS ;\LUMINA ACTIVATED AT 3 3 O o c FOR 8 HR. AND PARTIALLY DEACTIVATED TABLE I1 1.04 1-06 (% H) 1'5 2.5 5 TO I5 Average 1'5 2'5 5 I 0 I5 Average in Fig. 7. (iii) Similar band widening occurs in the absence of adsorption in passage through a column of z cm. diam. packed with non-adsorbent particles of similar grain size. I 4-Di-p-toluidinoanthraquinone is not adsorbed from toluene solution by slightly moist and unactivated dried hydrate of alumina fairly similar in particle size to the activated product used in Table 11. Tests gave RM = 95.1 yo within the anticipated peripheral error of IOO yo. The RL/RM value was 1-15. The concentration of the solution leaving the column was also measured photometrically the results confirming the band widening as shown 5'3 5'6 I 8.7 A.STEWART Rfif 51-2 RL 59'2 55'7 1-14 1-12 50'4 1-12 1-06 1'1 I 1-09 1-07 1-29 1-08 1-09 1-05 1.05 1-17 1-04 I '04 -565 '559 -526 '439 '403 - 1-07 1.14 '2 RL IRM 1-15 1'10 I -09 1.07 1'20 1-07 1-07 1 - 1 0 2'4 4'3 I 6.8 30.0 1-23 1-09 1-07 1-10 '4 -6 5'0 '4 1.6 1-09 7'9 14.1 41.2 61.1 80.6 50.8 2 0.9 15'4 39'3 12'1 41.0 -ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 76 An increase in the viscosity of the solvent by addition of liquid paraffin reduced the flow rate and band movement to a quarter but did not change the R,/R value.(iv) The above results indicate that the band widening is not dependent on adsorption but is directly proportional to the distance of travel along the column since R,/R is substantially the same for all the bands in a column AD is constant and R = AxlAD. (v) The bands from a solution containing several components move through a column in such a way that they have the same width a t any given position in the column. Fig. 8 shows the observed band width plotted against the mean band position for the four component mixture referred to above using a column of 2.4 cm. diam. There is appreciable observational error indicated by vertical lines on which the mean value is marked. In the upper part of the column the widening is proportional to the distance travelled but beyond 25 cm.the rate of widening decreases perhaps due to the band edges becoming too weak to be observed. The width of the four bands practically overlaps although the leading band passed the 15 cm. position 48 hr. before the last band. A possible explanation of the effect is that the major part of the band width is due to the widening effect which is similar for all the bands and small differences arising from the length of the initial bands are concealed by the observational error. (vi) EFFECT OF SOLVENT FLOW ON BAND WIDENING.-A moderate increase in the solvent movement D from 0.6 to 3.0 cm./min. had little effect in a 2-cm. column on the RL/RM value at 1.13. Increasing the flow to 15 cm./min. caused an appreciable initial widening of the band after which the rate of widening decreased to give an average value of 1-23.FIG. 7.-Behaviour on a non-adsorbent column. Band position and concentration leaving column for I 4-di-p-toluidinoanthraquinone in toluene on a 2-cm. column of unactivated alumina. Curve I 40 ml. (34.6 9.) initial solution conc. 17-8 mg. per roo ml. , 2 20 , (17.3 g.) * I > ,> 3 5 9 ( 4'3 g.) (vii) EFFECT OF COLUMN DIAMETER.-The band widening appears to decrease as the column diameter increases columns of 2 5 and 7-5 cm. diam. giving A. STEWART 77 RL/Rni values of 1-13 1.06 and 1-04 the behaviour of the columns being other- wise identical. Equally small values have on occasion been obtained in 2-cm. diam. columns and the result may only indicate that the most effective packing is more easily obtained in columns of larger diameter.(viii) A possible cause of band widening may be that the interstitial capillary spaces are not of uniform size and flow through some outdistancing flow through others leads to dilution and consequent widening of the band. Although the band movement is faster in mid-column the rate of diffusion in a column is considered to be too small for diffusion from the mid-column to periphery to account for the widening. In addition the diffusion would be affected by the strength of adsorption the viscosity of solvent and the rate of solvent flow. Some diffusion is to be expected when a solution moves along a capillary space between two solvent layers since the movement is a maximum in mid- capillary and zero a t the capillary surface.But if this was the full explanation band widening might be expected to be more sensitive to the viscosity of the solvent and rate of solvent movement and to increase in a more closely packed column. The Length of the Initial Band.-From eqn. (2) the length of band obtained on adding a solution of a pure compound to a column x should equal RD, the product of the relative band movement and the length of column occupied by a volume of interstitial solvent equal to the volume of solution added Fig. g shows the observed x plotted against RLDo and RFDo from the results of 50 measurements covering 20 pure compounds in which 10 or 2 0 ml. of solution was used Do being approximately 5 or 10 cm.FIG. 8.-Variation in band width with mean position in column. Solvent Toluene ; Adsorbent Alumina activated 300° C 24 hr. j yo N. Column 2-4 cm. diam. El Purple band. -0- Violet band. 0 Red band. x Blue band. There is fairly satisfactory agreement between x and R,D, whereas the value of RFD is only about 70 yo of x,. For strongly adsorbed compounds the initial band is very small and difficult to measure and the x value tends to be too ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS 78 large in these cases. The average values for the 40 measurements in which the band exceeded I cm. length were x 4-59 cm. ; R,D 4.66 cm. R,D 3.40 cm. RLDo/x 1-00 std. dev. 0.17 RFDo/x 0.71 std. dev. 0.14 The agreement between x and R,D is to be expected since the initial band is itself a leading edge and its length will be correspondingly greater than the theoretical R,D and equal RLDo.The Relation between Chemical Constitution and Chromatographic Behaviour of Anthraquinone Compounds .-Strain 1 has shown that the relation between constitution and chromatographic behaviour frequently depends on the solvent and adsorbent conditions used and this is found to apply to anthraquinone compounds in some cases a change in the activity of the adsorbent by water addition being sufficient to alter the order in which bands move. Within this qualification certain general rules serve as a guide to the behaviour and appear to be governed by two factors (i) the specific action of certain groups such as hydroxy and amino in increasing the adsorption probably as Meunier and Vinet16 suggest due to hydrogen bonding between the group and the adsorbent (ii) the avidity of the various compounds for the solvent employed acting against a fairly general adsorptive power.The order of increasing adsorption for mono-substitution is generally halogeno nitro arylamino alkylamino amino acylamido hydroxy group in side chain and hydroxy group attached to nucleus. There does not appear to be any systematic relation between mono-substitution in the I and z positions. 1 4 Strain Ind. ErJg. Chem. (Anal.) 1946 18 605. FIG. 9.-Relation between x and RLD ( 0 ) or RpD (+). Adsorption generally increases with increasing number of substituent groups of the same composition to an extent which varies with the position occupied 15 Meunier and Vinet Chromatographie et Mesomerze (Masson & Cie Paris).A. STEWART 79 and the nature of the substituent b u t decreased adsorption from aqueous medium occurs in the case of the sulphonate group. The introduction of further groups of different composition may either increase or decrease the adsorption thus methyl halogeno and arylamino groups tend to decrease and amino methoxy and hydroxy to increase the adsorption. Discussion The simple method described for the deactivation of adsorbents gives remarkable control over the chromatographic behaviour and is easily applied as a further variable to the choice of solvent and adsorbent when selecting the conditions for any particular separation.It enables the separations to be obtained at a convenient rate and in this connection is of considerable assistance when a flowing chromatogram is combined with photometric measurement for the quantitative determination of the components.16 l7 Appreciable heat of wetting is generated when a solvent is added to an adsorbent even after partial deactivation and as Muller has shown the heat generated gives a measure of the activity. It would appear that the adsorption of a solute depends on competition between the solute water and solvent for the available active surface. In the case of activated alumina a small degree of deactivation renders the distribution of a solute between solvent and adsorbent independent of the concentration a factor which facilitates the chromatographic separation ; under these conditions fairly satisfactory agreement has been obtained between the mean relative move- ment of the bands the length of the initial band and the values predicted from theory.The solvent flow is greater in mid-column than at the periphery where the almost spherical adsorbent particles are in contact with a surface of much smaller curvature and the packing would be expected to be less perfect. Although the chromatograph columns of alumina show dilatency the particles are not in closest packing-the interstitial volume is too high for this-and it seems likely that they are in a flocculated state similar to that described by Kruyt and Selms l8 for suspensions of silica particles in organic solvent in which case the glass surface of the tube may be linked t o the adsorbent particles by forces similar to those responsible for the flocculation.The troublesome band widening appears to be related to the manner of flow through the interstitial spaces. Research L a bovatories Iin$erial Chemical Industries Limited Dyestufs Division Hexagon House Blackley Maizchestev. Cropper and Strafford J . SOC. Chem. I n d . 1944 63 263. l7 Cropper Aizalyst 1946 71 263. l8 Kruyt and Selms Kec. trav. chim. 1943 62 407. THE FUNCTION OF ADSORBENT ACTIVITY IN THE CHROMATOGRAPHIC SEPARATION OF CERTAIN ANTHRAQUINONE COMPOUNDS BY A. STEWART Received 8th August 1949 The activity oi adsorbents used in chromatography with organic solvents is easily controlled by preliminary incorporation of water in the adsorbent.The effect on the band movement of simple anthraquinone compounds in toluene on activated alumina and unactivated magnesium carbonate is described. The conditions for preparing activated alumina by heating between 250' and 500' C have considerably less effect on the behaviour than has the deactivation treatment. The band movement measure-ments were made under satisfactory comparative conditions but are affected by several errors ; in particular the greater solvent movement in mid-column than a t the periphery renders values obtained a t the periphery 11 % low. The distribution of the anthraquinone compounds between solvent and partially deactivated alumina is independent of concentration. The length of the initial chromato-graph band and the mean rate of band movement are in satisfactory accord with theory.The considerable band widening in passage through a column is independent of the adsorption little affected by the rate of solvent flow and may be caused by non-uniformity of flow through the interstitial capillary spaces ; its effect on the separation o f a mixture causes the various bands to have approximately equal width a t the same position in the column. The relation between constitution and chromatographic behaviour of simple anthra-quinone compounds largely influenced by the conditions used is briefly described. The successful application of adsorption chromatography ultimately depends on the empirical selection of operating conditions if for no other reason than that the composition of the mixtures examined is rarely fully known until the separation has been achieved.Various methods have been described for the partial deactivation of adsorbents 1-5 and it is the purpose of this paper to draw attention to the simplicity of this operation and the wide range of control exercised on the band movement enabling this variable to be systematically applied in selecting the operating conditions. To illustrate the behaviour band movement measurements have been made using the highly coloured simple anthraquinone compounds in toluene solution ; these conditions do not necessarily represent the optimum for such separations for which their fairly low solubility is often a controlling factor. The effect on activated alumina and magnesium carbonate adsor-bents is described; it has been applied to other adsorbents but with the wide range of control obtained on deactivation these adsorbents cover most of the activity range required when the separation is carried out from organic solvent solution.The effect of variation in the conditions for the activation of alumina is shown to have much less effect on the behaviour than has the subsequent deactivation. Partial deactivation of activated alumina gives a distribution between adsorbent and solvent which is substantially independent of the concen-1 Zechnieister and Cholnoky trans. Bacharach and Robinson. Principles a??d 2 Brockmann and Schodder Ber. 1941 74 73. 8 Miiller Vet%. Ver. Schweiz. Physiol. 1942 21 29; Helv. chim. A d a 1943 26 1945; Practice of Clzromatografihy.(Chapman & Hall London 1941 p. 48.) 19443 27 404. Schroeder Ann. N.Y. Acad. Sci. 1948 49 204. Stewart and I.C.I. Ltd. Brit. Pat. 565J405. C 6 66 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS tration and under these conditions the mean band movement and the length of initial band are shown to be in agreement with the theory described by LeRosen and others.'-1° Conditions affecting the widening of bands, often a limiting factor in the chromatographic separation are considered. L . . . . 171 . . . . K . . . . Theoretical The symbols used have the following significance : Weight of interstitial solvent per unit length of the column. Weight of adsorbent per unit length of the column. Distribution constant = the weight of solute adsorbed per gram of adsorbent/equilibrium concentration of solution in gram per gram this value being substantially constant under the conditions c .. . . 'W . . . . x . . . . D . . Go . . . . x . . . . Do . . . . D,"B R employed. Distribution constant prevailing in unit length of the column, K' = Kmli. Concentration of solution. Weight of solution. Distance of any point in the column measured from the beginning (top) of the column. Distance the developing solvent has moved in the column. Initial concentration of solution added to column. Length of the initial band formed by adding w to column. x ~ x t referring to initial lengths of components a and p. Length of column occupied by interstitial solvent of weight w,. Distance the developing solvent requires to move to separate two bands starting from xg and x!.Displacement of zone on column/displacement of solvent in column R = dx/dD. R for the leading edge of a band ; R, for the following edge ; RM for the mean of RL and R,. RM expressed as a percentage of the solvent movement. Extent of water deactivation; ml. water addition per IOO g. adsorbent the latter including any moisture initially present. ,4 measure of the widening of a band edge in passage through the column. As LeRosen and others '-lo have shown where the distribution between adsorbent and solvent is independent of concentration and is rapidly established the conditions for the formation and separation of the bands can be evaluated. If a solution of concentration c of a pure compound is added to a column in amount 'wo equal to length Do of interstitial solvent the ratio of inter-stitial solvent to adsorbent in unit length of the column being i/m then the solute is adsorbed until it reaches equilibrium with the initial concentra-tion of the solution and a band of length x is formed such that X,C + K'x,c = Doco or x = D,/(I + K') = D,/(I + Kw/i) .(1) If pure solvent is now allowed to flow through the column then the band moves down at a fraction of the rate of linear movement of the solvent equal to the fraction of the total substance present in the interstitial solvent. The movement of the band relative to the linear movement of solvent through the column R is given by R = %,/Do = I/(I + K') = I/(I + Km/i) . * (4 or I(' = (I - R)/R and K = [(I - R)/R] i/m .* (3) 6 LeRosen J . Amev. CJLem. SOC. 1947 69 87. 7 De Vault J . Amev. Chem. SOL 1943 65 5 3 2 . 8 Weiss J. Chem. Soc. 1943 297. SGlueckauf J . Chern. SOL 1947 1302. 10 \Veil-Malherbe J . Chem. SOC. 1943 303 A. STEWART 67 The movement of the chromatograph bands through a column is con-veniently expressed l1 as the relative movement 2'2 which is given here as a percentage of the solvent movement. The value depends on the ratio of solvent to adsorbent ilm for the particular column so that the results obtained from different columns require to be calculated on the basis of eqn. (2) and (3) to correspond with a particular value of i/m. FIG. 1.-Effect of water deactivation treatment applied to activated alumina on the band movements of anthraquinone compounds in toluene.Curve I . I -Chloroanthraquinone. 2. I -Methylaminoanthraquinone . 3. I -Aminoanthraquinone. . 4. I-Dimethylaminoanthraquinone. 5. I-Amino-4-methylaminoanthraquinone. , 6 . 2-Aminoanthraquinone. , 7. I 4-Diaminoanthraquinone. , 8. I 4 5-Triaminoanthraquinone. Under the above conditions the band should move down the column without changing its length the small region in front and rear where adsorption and desorption occur remaining constant. The position of the following and leading edges of a band is given by Following edge x = x D/Do * (4) Leading edge x = x + x D/Do . - (5) 11 LeRosen J . Aupzev. Chem. SOC. 1942 64 1905 68 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS If a second component is present in the initial solution and there is no mutual interference in the distribution the bands will draw apart and commence to separate when the following edge of the faster reaches the leading edge of the slower x! Ds/Do = x + x D,/D or D;P/Do = x;/(x{ - q) (6) .the actual position on the column being given by RFD,. FIG. z .-Eff ect of water deactivation treatment applied to unactivated magnesium carbonate on the band movement of anthraquinone compounds in toluene. Curve I. I -Hydroxy-4-~-toluidinoanthraquinone. , 2. r 4 5-Triaminoanthraquinone. , 3 . I 4 5 8-Tetraminoanthraquinone. The solvent flow required to produce the initial separation should depend on the ratio of the lower R value to the difference in R values while the position at which this separation occurs and the actual rate at which the bands subsequently move apart depend on the value of R.Certain complications limit the simple application of the above relations. The leading edge of a band moves considerably (20 %-30 %) faster than the following edge and the diminishing concentration at the band edges increases the difficulty of measuring the band movement and determining the position of separation. The bands move at a different rate in mid-column than at the periphery probably 16 yo faster in the columns considered below A. STEWART 69 Experimental The Control of Adsorbent Activity by Partial Deactivation .-The preferred method of deactivation consists in adding liquid water to the adsorbent, shaking to distribute the soft wet portion initially formed and mixing for 2 hr., conveniently by rotating the container.The addition necessary causes little or no change in the handling properties of the adsorbents ; the method is easy to carry out and gives a reproducible behaviour. The deactivation is expressed as ml. water addition per IOO g. adsorbent abbreviated to yo H the weight of adsorbent including any moisture initially present. FIG. 3.-The effect of water deactivation treatment applied to activated alumina on the distribution constant of anthraquinone compounds between toluene and adsorbent. The curves are numbered as in Fig. I. The effect of partial deactivation on the band movement of simple anthra-quinone compounds in toluene on activated alumina (Type 0 ; P. Spence & Sons Ltd.) and on unactivated magnesium carbonate (Ponderous.B.P. quality ; Cuxson Gerrard and Co. Ltd.) is shown in Fig. I and 2 the band movement being given as the mean of the leading and following edge movement yo €2~1, determined and corrected for variation in column packing as described below. This reproducible behaviour enables the rate of movement to be controlled over a wide range so that in conjunction with a suitable choice of organic solvent and a sufficiently high initial adsorbent activity most separations can be obtained at a convenient rate. In practice the two adsorbents mentioned which are of satisfactory grain size for chromatographic use meet most requirements. Inter 70 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS mediate values of the deactivation tend to give most effective separations as the maximum increase of movement with deactivation generally occurs at low yo H values for weakly adsorbed compounds and a t higher values for those more strongly adsorbed.Occasionally the order of the band movement changes with the deactivation generally at low values.of yo H (Fig. 3). Partial deactivation of activated alumina gives a distribution between adsorbent and solvent sub-stantially independent of concentration and bands do not form long trails at one or other edge when this applies. When the solvent used has affinity for water equilibrium appears to be estab-lished with the deactivated adsorbent and any initial water present. in the solvent affects the chromatographic behaviour to some extent. For separalions from an aqueous medium this deactivation treatment is ineffective.FIG. 4.-Band behaviour of the test mixture in toluene on alumina activated a t 330' C for 4 hr. Curvc I . Red band. Curve 3 . Purple band. , 2. Blue band. , 4. Violet band. Fig. 3 shows the effect of partial deactivation on the distribution constant, calculated from the yo RM values shown in Fig. I by means of eqn. (3). There is a very considerable change in the distribution with practically a linear relation between log K and yo H when the deactivation exceeds 2-5 yo H . The Effect of Conditions of Activation of Alumina on the Deactivation Behaviour .-Activated alumina may be prepared la by heating aluminium trihydroxide at atmospheric pressure the product y alumina retaining some water to an extent varying with the activation temperature; it may also contai A.STEWART 71 alkaline impurities l3 which can if necessary be removed but these rarely upset the chromatographic behaviour. Satisfactory commercial products are available which obviate the rather troublesome activation. To find the effect of activation conditions on the behaviour after water deactivation dried hydrate of alumina (British Aluminium Co. Ltd.) was heated in an open pan at temperatures and for periods on temperature given below together with the moisture content of the products determined by ignition to about 2300' C. Temperature . . " C 250 330 330 330-360 400 500-550 Time hr. 4 4 a 24 4 4 Moisturecontent % 11.3 8.8 7.9 5'3 3'9 I.7 Each sample was deactivated with from 1.5 % to 15 % water addition (% H ) and FIG.5.-Effect of conditions of activation on the relative movement of the violet band. Curve I . Activated at 250° C 4 hr. , 2 . 2 9 330°c 4 9 9 t 0 I 330° c 8 ,, , > 1 , 330"-360" C 24 ,, 1 3. , 400°c 4 ,, , 4. , 500°-5500c 4 ,, band movements measured using a toluene solution of four components which gave distinctive highly coloured bands covering a convenient range of adsorption affinity. The 10 ml. solution added to each column contained 0.77 mg. I -Methylaminoanthraquinone (red least adsorbed band), 0.34 mg. I 4-Dimethylaminoanthraquinone (blue band), I -28 mg. I -Amino-q-methylaminoanthraquinone (purple band), 0.58 mg. I 4-Diaminoanthraquinone (violet most adsorbed band), Is Siewert and Jungnickel A.C.S. Abstr. 1943 37 5898 72 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS the actual concentration of the components being unimportant provided sufficient is present to give a highly coloured band.Fig. 4 shows the band behaviour of these components on one sample of adsorbent while Fig. 5 shows the effect of conditions of activation on the behaviour of the most strongly adsorbed com-ponent the results for the others being similarly distributed and displaced to an extent indicated in Fig. 4. The curves in Fig. 5 show the effect of activation temperature the effect of the heating period between 4 and 24 hr. a t 330' C being within the experimental error. The activation conditions employed have considerably less effect on the be-haviour than has the subsequent deactivation treatment. There is no direct relation between the initial moisture content and the water added on deactivation, e g .samples prepared a t 250° and 530" C differ by 9.6 yo in initial moisture but FIG. 6.-Effect of the correction for variation in the ratio of solvent to adsorbent in alumina columns ( x ) and curves show the uncorrected values ; (e) show the corrected values corresponding to the curves shown in Fig. 4. The i/m values were 1-5 yo H 0.580 ; 2.5 yo H 0.583 ; 5.0 yo H 0.542 ; 10 % H 0.462 ; 15 ?< H 0.425. give practically the same relative band movement when a further 10 yo water is added to both. For low values of yo H the band movement increases with activation temperature but the extent to which the band movement increases on further addition of water decreases with activation temperature.Qgalitative tests show that activation temperatures above 5o0°-5500 C give a progressive decrease in activity. The quality of activated alumina probably depends to some extent on the nature of the starting material and other factors but the general behaviour on deactivation is similar The Measurement of Band Movement.-The tubes used were z cm. diam. and 55 cm. long having a compact cotton wool retaining plug. Well-packed columns free from air bubbles were obtained from 150 g . alumina o A. STEWART 73 70 g. magnesium carbonate which were added as a slurry in toluene and settled, with assistance of intermittent tapping to a constant column length before use ; the interstitial solvent was determined by weighing the column before and after adding the adsorbent.The temperature remained close to 20' C throughout and no temperature correction was applied. The solvent used was high-grade technical toluene which had been passed through a column of the same adsorbent so as t o minimize any possible effect on the activity. The solution under test was added (10 ml. of 0.1 % conc.) when the solvent level had dropped to the adsorption surface and as soon as this entered the column i t was washed in with several small (0.5-1 ml.) lots of solvent before raising the solvent level to about 10 cm. above the adsorbent where i t was maintained throughout the measurement by means of a constant level arrangement. Care was taken a t all stages not to disturb the adsorbent or allow air to enter the column. Gravity flow was employed and was measured by weight.The position of the band edges and the solvent f l o ~ . gave satisfactory linear relations against time from which the RL and RF values were obtained by means of the previously determined length of column occupied by I g. of interstitial solvent and these are given throughout this paper as a percentage of the solvent movement. A typical column was of 3.303 cm.2 cross-sectional area and 43.9 cm. length L , containing 150.3 g . (m x L) activated alumina of 5 yo H and 81-29 g. (i x L) interstitial solvent; the latter occupies 0-540 cm. per g. solvent and for the observed flow of 54-34 g./hr. the linear movement of solvent was 29.29 cm./hr. The solvent movement varied between 23 and 45 average 28.8 cm./hr. for the various columns.The value of i/m was fairly constant for a single sample of adsorbent but varied from 0.6 to 0.4 as the yo H increased from 1.5 to 15 yo the mean value being close to 0.5. Values of R corresponding to this mean value of i/m were obtained from the measurements on alumina columns from The magnesium carbonate columns gave a mean value for ilnz of 1.17 which was used in place of 0.5 for their correction. The effect of the correction for i/rn is shown in Fig. 6 where the uncorrected values corresponding to Fig. 4 are shown by ( x ) with curves drawn through these points and the corrected values-corresponding to the curves shown in Fig. 4 -are indicated by ( ). The correction is relatively small and without significant effect on the general relation between band movement and adsorbent activity.The reproducibility of the band movement measurement was tested in a series of columns using a band having R 51 % ; the standard deviation for the RL and R values was 3'3 and 2-7 and for R 2.0. RL/RM = 1-15 with standard deviation 0.06. The behaviour of these columns was found sub-stantially unchanged after use in a series of measurements, The reproducibility is reasonably satisfactory for comparative purposes involving fairly large differences but apart from errors due to the difficulty of observing the exact position of the band edge and the occasional inclination of the bands there is a further error arising from the bands and presumably the solvent moving faster in mid-column than a t the periphery. Since a uniform solvent movement over the cross-sectional area of the column has been assumed the RM values are too low results given below indicating this error to be about 11 yo but no correction has been applied for this factor.Provided the com-parisons are carried out in similar columns as in the present case this source of error should remain reasonably constant. The Relation between Adsorption and Band Movement.-Preliminary adsorption measurements were made allowing a solvent solution to reach equilibrium with an alumina adsorbent and determining the equilibrium con-centration but this procedure was limited to low values of the latter by the loti, solubility of the compounds used and the photometric method employed. The results fitted the Freundlich adsorption equation, log (specific adsorption) = log,,K + a log, (equilibrium concentration), f 74 -ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS cc being about 0.75 for activated alumina but increasing to 1-0 on partial deactiva-tion with water and becoming slightly greater than I for weak adsorption.Within a rather large experimental error there was agreement between the observed RM value and that calculated from K by means of eqn. (2). To obtain measurements at higher concentrations a 2-cm. column packed with 50 g. partially deactivated Type 0 alumina and of known interstitial volume was used the solution being fed through the column until the issuing solution was a t the initial concentration and the amount adsorbed determined TABLE I THE RELATION BETWEEN THE BAND MOVEMENT CALCULATED FROM ADSORPTION MEASUREMENT AND THE OBSERVED VALUE Adsorbent Partially deactivated Type 0 Activated Alumina.% H = ml. water added per roo g. adsorbent. . . measurement for band washing out. from peripheral Gbservation. -4 dsorbate I 4-Di-p-toluidino-anthraquinone I -Methylamino-anthraqu inone I -Amino-;?-methyl-anthraquinone I -Amino-4-niethyl-aminoanthraquinone I 4-Diaminoanthra-quinone Conditions Toluene 5 % H Conc. 46 mg. % Toluene 5 yo H Conc. 115 mg. yo , 62.5 ,, I ? 33'2 ,, Toluene 5 % H Conc. 46.2 mg. 7,; 2 s 16.3 ,, I 4'4 9, Toluene 5 yo H Conc. 23.1 mg. 04 Benzene Conc. 4.0 mg. % 5 % H 10 % H 15 % H Toluene 5 % H Conc. 7.5 mg. % % R calc. from adsorption ;:;) 9'3 17'7 46.2 80.1 % R obs.RF 55'8 51'4 51.6 38.0 37'0 36-6 26.8 29'2 27'4 28.2 27'9 29'0 7'2 7'5 14'1 35'3 60.7 7'0 68.7 63.1 63'3 47'5 47'2 46.4 40.1 31.6 38.8 30.6 30.1 37'6 10-3 8.6 I 8.2 43'1 83'5 7'1 by difference and also by washing the band from the column allowance being made for the solution held by the cotton wool retaining plug. The relative band movement was determined in the same column by observing the peripheral movement and in some cases by observing when the band started and finished washing from the column. There is reasonable agreement between the band movement calculated from adsorption measurements and the observed Rnf values the values obtained The results are summarized in Table I A.STEWART 75 by washing from the column being 5-1 yo high and those obtained from peripheral observations 11.1 yo low ; as indicated above the latter are likely to be too low due to non-uniformity of solvent flow across the cross-sectional area of the column. The results confirm that band movement on the partially deacti-vated alumina adsorbent used is independent of concentration. A considerable difference is shown between the band movement of the leading and following edges which in the case of the peripheral results is a measure of the band widening further considered below the somewhat greater difference obtained when the band is washed out represents the widening between the leading edge in mid-column and the following edge a t the periphery. The Widening of Bands.-(i) The leading and following edges of a band move down at a uniform rate and the widening of each edge is given by RL/Rlw.The widening is not uniform over the whole band but is greater towards the edges than for the main deeply coloured portion the movement of I-methyl-aminoanthraquinone in toluene on Type 0 5 yo H alumina being : Edges of band . . . . 43-2 59'2 51-2 R F RL Rfif RL IRM 1-15 Main portion of band . . 45.1 55'7 50'4 1'10 (ii) The RL/RM value is substantially the same for the bands of all components present in a column independent of the R value and is little affected by the extent of deactivation. Table I1 gives results from a set of measurements on the activation of alumina using the four component mixture as described above, which are typical of the six samples examined in this connection.TABLE I1 RL/RH FOR THE FOUR COMPONENT MIXTURE ON FIVE 2-CM. COLUMNS ;\LUMINA ACTIVATED AT 3 3 O o c FOR 8 HR. AND PARTIALLY DEACTIVATED Column Deactivation (% H) 1'5 2.5 5 I5 Average TO 1'5 2'5 5 I5 Average I 0 -565 '559 -526 '439 '403 -1-14 1-29 1-08 1-09 1.14 1-12 1.04 1-06 1-23 1-09 1-07 1-10 1-05 1.05 1-17 1-04 I '04 1-07 1-12 1-06 1-09 1-07 1-09 1'1 I I -09 1.07 1-07 1-07 1'20 1 - 1 0 '2 '4 1.6 I 8.7 5'3 5'6 '4 -6 5'0 15'4 39'3 12'1 7'9 14.1 41.2 61.1 80.6 41.0 2'4 4'3 I 6.8 30.0 50.8 2 0.9 (iii) Similar band widening occurs in the absence of adsorption in passage through a column of z cm.diam. packed with non-adsorbent particles of similar grain size. I 4-Di-p-toluidinoanthraquinone is not adsorbed from toluene solution by slightly moist and unactivated dried hydrate of alumina fairly similar in particle size to the activated product used in Table 11. Tests gave RM = 95.1 yo within the anticipated peripheral error of IOO yo. The RL/RM value was 1-15. The concentration of the solution leaving the column was also measured photometrically the results confirming the band widening as shown in Fig. 7 76 -ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS An increase in the viscosity of the solvent by addition of liquid paraffin reduced the flow rate and band movement to a quarter but did not change the R,/R, value. (iv) The above results indicate that the band widening is not dependent on adsorption but is directly proportional to the distance of travel along the column since R,/R is substantially the same for all the bands in a column, AD is constant and R = AxlAD.(v) The bands from a solution containing several components move through a column in such a way that they have the same width a t any given position in the column. Fig. 8 shows the observed band width plotted against the mean band position for the four component mixture referred to above using a column of 2.4 cm. diam. There is appreciable observational error indicated FIG. 7.-Behaviour on a non-adsorbent column. Band position and concentration leaving column for I 4-di-p-toluidinoanthraquinone in toluene on a 2-cm. column of unactivated alumina.Curve I 40 ml. (34.6 9.) initial solution conc. 17-8 mg. per roo ml. 2 20 , (17.3 g.) ,, * I 3 5 9 ( 4'3 g.) > ,> by vertical lines on which the mean value is marked. In the upper part of the column the widening is proportional to the distance travelled but beyond 25 cm. the rate of widening decreases perhaps due to the band edges becoming too weak to be observed. The width of the four bands practically overlaps although the leading band passed the 15 cm. position 48 hr. before the last band. A possible explanation of the effect is that the major part of the band width is due to the widening effect which is similar for all the bands and small differences arising from the length of the initial bands are concealed by the observational error.(vi) EFFECT OF SOLVENT FLOW ON BAND WIDENING.-A moderate increase in the solvent movement D from 0.6 to 3.0 cm./min. had little effect in a 2-cm. column on the RL/RM value at 1.13. Increasing the flow to 15 cm./min. caused an appreciable initial widening of the band after which the rate of widening decreased to give an average value of 1-23. (vii) EFFECT OF COLUMN DIAMETER.-The band widening appears to decrease as the column diameter increases columns of 2 5 and 7-5 cm. diam. givin A. STEWART 77 RL/Rni values of 1-13 1.06 and 1-04 the behaviour of the columns being other-wise identical. Equally small values have on occasion been obtained in 2-cm. diam. columns and the result may only indicate that the most effective packing is more easily obtained in columns of larger diameter.(viii) A possible cause of band widening may be that the interstitial capillary spaces are not of uniform size and flow through some outdistancing flow through others leads to dilution and consequent widening of the band. Although the band movement is faster in mid-column the rate of diffusion in a column is considered to be too small for diffusion from the mid-column to periphery to account for the widening. In addition the diffusion would be affected by the strength of adsorption the viscosity of solvent and the rate of solvent flow. FIG. 8.-Variation in band width with mean position in column. Solvent Toluene ; Adsorbent Alumina activated 300° C 24 hr. j yo N. Column 2-4 cm. diam. 0 Red band. El Purple band. x Blue band.-0- Violet band. Some diffusion is to be expected when a solution moves along a capillary space between two solvent layers since the movement is a maximum in mid-capillary and zero a t the capillary surface. But if this was the full explanation band widening might be expected to be more sensitive to the viscosity of the solvent and rate of solvent movement and to increase in a more closely packed column. The Length of the Initial Band.-From eqn. (2) the length of band obtained on adding a solution of a pure compound to a column x should equal RD,, the product of the relative band movement and the length of column occupied by a volume of interstitial solvent equal to the volume of solution added Fig. g shows the observed x plotted against RLDo and RFDo from the results of 50 measurements covering 20 pure compounds in which 10 or 2 0 ml.of solution was used Do being approximately 5 or 10 cm. There is fairly satisfactory agreement between x and R,D, whereas the value of RFD is only about 70 yo of x,. For strongly adsorbed compounds the initial band is very small and difficult to measure and the x value tends to be to 78 ADSORBENT ACTIVITY OF CERTAIN COMPOUNDS large in these cases. the band exceeded I cm. length were : The average values for the 40 measurements in which x 4-59 cm. ; R,D 4.66 cm. R,D 3.40 cm. RLDo/x 1-00 std. dev. 0.17 RFDo/x 0.71 std. dev. 0.14 The agreement between x and R,D is to be expected since the initial band is itself a leading edge and its length will be correspondingly greater than the theoretical R,D and equal RLDo.The Relation between Chemical Constitution and Chromatographic Behaviour of Anthraquinone Compounds .-Strain 1 has shown that the relation between constitution and chromatographic behaviour frequently depends on the solvent and adsorbent conditions used and this is found to apply to FIG. 9.-Relation between x and RLD ( 0 ) or RpD (+). anthraquinone compounds in some cases a change in the activity of the adsorbent by water addition being sufficient to alter the order in which bands move. Within this qualification certain general rules serve as a guide to the behaviour and appear to be governed by two factors (i) the specific action of certain groups such as hydroxy and amino in increasing the adsorption probably as Meunier and Vinet16 suggest due to hydrogen bonding between the group and the adsorbent (ii) the avidity of the various compounds for the solvent employed acting against a fairly general adsorptive power.The order of increasing adsorption for mono-substitution is generally halogeno, nitro arylamino alkylamino amino acylamido hydroxy group in side chain, and hydroxy group attached to nucleus. There does not appear to be any systematic relation between mono-substitution in the I and z positions. Adsorption generally increases with increasing number of substituent groups of the same composition to an extent which varies with the position occupied 1 4 Strain Ind. ErJg. Chem. (Anal.) 1946 18 605. 15 Meunier and Vinet Chromatographie et Mesomerze (Masson & Cie Paris) A. STEWART 79 and the nature of the substituent b u t decreased adsorption from aqueous medium occurs in the case of the sulphonate group.The introduction of further groups of different composition may either increase or decrease the adsorption thus methyl halogeno and arylamino groups tend to decrease and amino methoxy and hydroxy to increase the adsorption. Discussion The simple method described for the deactivation of adsorbents gives remarkable control over the chromatographic behaviour and is easily applied as a further variable to the choice of solvent and adsorbent when selecting the conditions for any particular separation. It enables the separations to be obtained at a convenient rate and in this connection is of considerable assistance when a flowing chromatogram is combined with photometric measurement for the quantitative determination of the components.16 l7 Appreciable heat of wetting is generated when a solvent is added to an adsorbent even after partial deactivation and as Muller has shown the heat generated gives a measure of the activity. It would appear that the adsorption of a solute depends on competition between the solute water and solvent for the available active surface. In the case of activated alumina a small degree of deactivation renders the distribution of a solute between solvent and adsorbent independent of the concentration a factor which facilitates the chromatographic separation ; under these conditions fairly satisfactory agreement has been obtained between the mean relative move-ment of the bands the length of the initial band and the values predicted from theory. The solvent flow is greater in mid-column than at the periphery where the almost spherical adsorbent particles are in contact with a surface of much smaller curvature and the packing would be expected to be less perfect. Although the chromatograph columns of alumina show dilatency the particles are not in closest packing-the interstitial volume is too high for this-and it seems likely that they are in a flocculated state similar to that described by Kruyt and Selms l8 for suspensions of silica particles in organic solvent, in which case the glass surface of the tube may be linked t o the adsorbent particles by forces similar to those responsible for the flocculation. The troublesome band widening appears to be related to the manner of flow through the interstitial spaces. Research L a bovatories, Iin$erial Chemical Industries Limited, Dyestufs Division, Hexagon House Blackley Maizchestev. Cropper and Strafford J . SOC. Chem. I n d . 1944 63 263. l7 Cropper Aizalyst 1946 71 263. l8 Kruyt and Selms Kec. trav. chim. 1943 62 407

ISSN:0366-9033    

DOI:10.1039/DF9490700065

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

79 EQUILIBRIUM AND RATE STUDIES OF CATION-EXCHANGE WITH MONOFUNCTIONAL RESINS BY D. K. HALE AND D. REICHENBERG Received 13th July 1949 The preparation of sulphonated cross-linked polystyrene and of cross-linked poly- methacrylic acid is described. The sodium-hydrogen exchange equilibria for both inaterials have been examined. The former resin is shown to behave as a monofunctional strong acid and the latter as a monofunctional weak acid. The application of the law of mass action to the exchange equilibria is discussed. CATION-EXCHANGE RESINS 80 A study of the rate of sodium-hydrogen exchange with sulphonated cross-linked polystyrene suggests that a t low concentrations of sodium ions in solution the rate- determining mechanism is diffusion of ions through a thin film of liquid surrounding the resin particle.The influence of hydroxyl ion concentration on the rate of exchange of sodium for hydrogen with both resins is described and the conditions under which the diffusion of ions within the resin particle may become the rate-controlling process are discussed. In the application of ion-exchange resins to chromatographic techniques more information is required on the fundamental molecular and ionic pro- cesses involved. Correlation of such information with the basic chemical and macromolecular structure may be expected to lead to the development of improved materials. In equilibrium and rate studies monofunctional resins obtained by addition polymerization e.g. sulphonated cross-linked polystyrene and cross-linked polymethacrylic acid offer many advantages.In contrast to earlier materials obtained by polycondensation e.g. sul- phonated phenol-formaldehyde resins these materials possess a fairly well- defined structure which may be systematically varied. They may also be examined over a wide range of pH without complicating factors arising due to the presence of different types of ionizable group. Moreover the technique of suspension polymerization enables the resins to be prepared in the form of spherical beads which are especially suitable for rate studies. In this paper the preliminary results of an investigation into the equili- brium and rate processes with sulphonated cross-linked polystyrene and cross-linked polymethacrylic acid are presented.Experimental Preparation of Cation- exchange Resins .-SULPHONATED CROSS-LI NKED POLYSTYRENE. This material was prepared by the sulphonation of a cross- linked polystyrene bead polymer as described by D'A1elio.l Styrene was co- polymerized a t 80" C for 18 hr. with ca. 10 yo divinylbenzene I y; benzoyl peroxide being employed as catalyst. The co-polymer was sulphonated with concentrated sulphuric acid a t 100" C for 8 hr. using I yo silver sulphate as catalyst. The maximum capacity of the product (5.25 milli-equivalents of basc per g. dry hydrogen form) was independent of particle size and agreed with the value calculated for a monosulphonic acid. The material was hygroscopic and the dry hydrogen form absorbed approximately 80 yo water a t zoo C. On conversion from the wet hydrogen form to the wet sodium form a decrease in volume of G y0-7 yo was observed.CROSS-LINKED POLYMETHACKYLIC ACID. The carboxylic type exchange resin was prepared from methacrylic acid and divinylbenzene. Commercial methacrylic acid redistilled 2.12 uacuo was polymerized with ca. 10 yo divinylbenzene in the presence of I yo benzoyl peroxide. Polymerization was carried out in a sealed tube a t GO" C for 24 hr. The product was treated with z N NaOI-I to remove soluble material washed and dried. The maximum capacity (9.2 milli- equivalents of base per g. dry hydrogen form) was consistent with the value calculated from the composition of the monomer mixture. The dry material absorbed approximately 130 o//o of water a t 20" C and an increase in volume of approximately 75 Yo was observed on conversion from the wet hydrogen to the wet sodium form.After preliminary cycling in a column between 2 M NaCl or 2 N NaOH and z N HC1 the cation-exchange resins were converted to the hydrogen form and washed with distilled water. Washing was continued until the pH of the effluent attained a value of 4.0 or higher. If necessary fines were removed by elutriation and the ion-exchange resins were air-dried to a uniform ~noisturc content. D'Alelio US. Pat. 2,366,007. D. K. HALE AND D. REICHENBERG 81 of the cation-exchange resins in thc. hydrogen €orm were weighed out into a series of bottles and further samples \\-ere taken for moisture content determinations. The latter were dried to constant weight over P,O in a vacuum desiccator Different amounts of NaOH solution with or without NaCl solution were added to the samples of resin and the solutions made up to 50 ml.For the sulphonated cross-linked poly- styrene the liquid-solid ratio was IOO/I for the cross-linked polymethacrylic Equilibrium Studies .-Samples acid ZOO/I. The solutions were allowed to stand with occasional shaking until equilibrium was attained (1-7 days). Aliquot samples of the solutions were then withdrawn and titrated with standard HC1 or NaOH to determine the extent of exchange. The pH values of the solutions were determined using a Cambridge pH meter and glass electrode. For the measurement of pH values greater than 9.0 a Cambridge Alki electrode was employed. obtain samples of resin of approximately uniform particle diameter the resins were sieved in the air-dry state using calibrated B.S.sieves. For the determination of rates of exchange two methods were employed. Rate Studies .-To FIG. I .-Sulphonated cross-linked polystyrene. Relationship between Na+ ion taken up by the resin and pH. A. In presence of 5 11 NaC1. B. In presence of 0-1 M NaC1. c. In absence of NaCl. (a) INDICATOR METHOD. This method which is only applicable to sulphonated cross-linked polystyrene depends on the fact that whether the solution is acid or alkaline exchange will proceed virtually to completion if the ratio [Na+]/[H'] in solution is sufficiently high (see below). The hydrogen form of the resin is stirred with a solution of NaCl and NaOH the latter being less than sufficient to neutralize the hydrogen ions liberated in the exchange process.The solution initially alkaline becomes acid when the amount of exchange just exceeds the amount of alkali originally added. An indicator preferably of the anionic type is used to show this change. A weighed amount of resin of known moisture content was added to a known volume of water containing a few drops of bromo-cresol green indicator solution (0.04 yo solution in water) in a small beaker. The mixture was stirred vigorously with a magnetic stirrer and a suitable mixture of NaC1 and NaOH solutions added. The time elapsing between the addition of the alkali solution and the colour transition (blue+yellow) was measured with a stop-watch.All experiments werecarried out a t room temperature (18'-22' C) and with a constant volume of solution. (b) SHALLOW-BED METHOD. A simple modification of the method used by Boyd ,\damson and Myers was employed. The resin sample (cn. 0.1 g.) was supported 1 Boyd Adamson and Myers J . Amer. Chcm. SOC. 1947 69 2836. 82 CATION-EXCHANGE RESINS on stainless steel gauze or a sintered-glass disc. The resin was first converted to the hydrogen form and washed free of acid. The solution containing sodium ions was then passed through the bed for an appropriate time a t a known flow- rate. The bed was then immediately washed with a stream of distilled water. In the case of the sulphonated cross-linked polystyrene the amount of residual hydrogen ion was determined by displacement using an excess of NaCl solution and titrating the solution with standard alkali.For the cross-linked poly- methacrylic acid the amount of exchange was determined by removing the sodium ion with a measured volume of standard acid and back titration of the acid solution. Results Exchange Equilibria.-The amount of sodium ion taken up a t equilibrium by the resins and its dependence on pH and on the ratio [Na+]/[H+] in solution is shown in Fig. I z and 3. 1 ; ~ . 2 .-Cross-linked polymethacrylic acid. Relationship between Na+ ion taken up by the resin and pH. A . I n presence of 2 hI NaCl. B. In presence of I M NaCl. c. I n presence oi 0-1 M NaCl. D. In absence of NaCl. I t will be seen from Fig. 3 that the Na+ ion taken up by both resins was dependent only on [Na']/[H+] in solution and not on "a+] or [H+] separately.In the case of the sulphonic acid type exchanger provided the [Na+]/[H+] ratio in solution is greater than IOO/I virtually complete replacement of hydrogen by sodium ion is effected. For the exchange resin containing carboxylic groups a [Xa+]/[Hi] ratio of a t least I O ~ / I is necessary to effect complete conversion to the sodium form. Exchange Kinetics.-Using the indicator method it was found that above a minimum rate of stirring the results obtained were independent of the stirring rate. This was found to apply over the whole range of sodium ion concentrations studied. Under the conditions employed however the degree of mixing of resin and solution may be expected to be lower than that attained with the shallow-bed method.D. K. HALE AND D. REICHENBERG FIG. 3.-Relationship between Na+ ion taken up by the resin and log, [Na,+]/[Hs+]. A . Cross-linked polymethacrylic acid. B. Sulphonated cross-linked polystyrene. FIG. 4 .-Sulphonated cross-linked polystyrene. Exchange kinetics a t low Na+ ion concentrations. (Indicator method .) A. Air-dry particle diameter 50-100 I I 1 i a “a,+] 0.026-0.029 M I1 I 1 “a,+] 0.048-0.050 M 300-400 p [Na +] 0.045-0.050 hT 1 8 1 11 B. c. D. “a,+] 0.023-0.028 M CATION-EXCHANGE RESINS 84 lo. I 10 77me - secondJ 20 I I‘ G . j.-Sulphonated J V “a,+] 30 cross-linked polystyrene. Exchange kinetics a t high Ka+ ion concentrations. (Indicator method.) 2-18 M A “a,+] 1-09 M A.Air-dry particle H. diameter Air-dry particle 300-400 p { O[Na,+] O[Na,+] 2.18 1.09 M M diameter 50-100 ---- - I t I (Jec -‘) I ? . O ? . O - yo- - I 1 1 I i I Meun sodurn sodurn Ion ion cancentrahon cancentrahon y ronllike ronllike I I 1 /. 1.5 5 FIG. 6.-Sulphonated cross-linked polystyrene. Relationship between half-life and Na + ion concentration. (Indicator method.) Air-dry particle diameter 300-400 EL. D. K. HALE AND D. REICHENBERG 85 The effect of sodium ion concentration in solution and the particle size of the resin were examined using the indicator method. The results obtained are shown in Fig. 4 and 5 . A t low sodium ion concentrations as shown in Fig.4 the exchange proceeds initially at an approximately constant rate but then slows down progressively as the exchange proceeds. At high sodium ion concentrations (Fig. 5) the form of curve obtained is similar but owing to the high rate of exchange under these conditions the initial rate of exchange cannot be determined accurately by the present method. The reciprocal of the time for half-conversion of the resin to the sodium form is plotted against the mean sodium ion concentration in solution in Fig. 6. It appears that a t high sodium ion concentrations the rate of exchange is independent of the sodium ion concentration whilst at low sodium ion con- centrations the rate is proportional to the sodium ion concentration. This is shown also in Fig.7 where the initial rate of exchange has been plotted against the mean sodium ion concentration in solution. FIG. $.-5julphonated cross-linked polystyrene. Relationship between rate of exchange and Xa+ ion concentration a t low Na+ ion concentrations. (Indicator method.) Air-dry particle diameter 300-400 u. A t low sodium ion concentrations there is a change in the sodium ion concen- tration as the reaction proceeds. It has however been found that the results are in agreement with the relationship d (Na) /dt = Kw[Na’-] where (Naj is the total amount of exchange w the weight of resin [Na ] the sodium ion concentration in solution and K a constant. It will be seen from Fig. 4 that at low sodium ion concentrations the rate of exchange with particles of diameter 50-100 p is about four times as great as that with particles of diameter 300-400 v.Thus a t low sodium ion concentra- tions the rate of exchange is approximately inversely proportional to the particle diameter. At high sodium ion concentrations no quantitative conclusions can be drawn from the present data but i t is apparent from Fig. 5 that exchange takes place more rapidly with the smaller particles. The indicator method cannot be used for the direct investigation of the effect of hydroxyl ion concentration on the rate of exchange and the shallow-bed method was employed for this purpose (see Fig. 8). Results obtained by the indicator method were confirmed in that with NaCl solutions the exchange rate was independent of sodium ion concentration CATION-EXCHANGE RESINS 86 7;me - seconds / 5 I > 20 .I FIG.8.-Sulphonated cross-linked polystyrene. Exchange kinetics at high Na+ ion concentration. Air-dry particle diameter 300-400 p. .4. Shallow-bed method. 2 N NaOH and I N NaOH solutions. Now-rates 15 cm./sec. and 30 cm./sec. R. Shallow-bed method. 2 14 NaCl and I &I NaCl solutions. Flow-rates 15 cm./sec. and 30 cm./sec. c. Indicator method. Na+ ion concentration 2 M and I M. FIG. 9.-Cross-linked polymethacrylic acid. Exchange kinetics a t high Na + con- centrations. Air-dry particle diameter 250-380 p. Flow-rate I cm./scc. A. Na+ concentration 2-2 M OH- concentration 0.2 M. 0'1 M. /O 2'1 M , B. ,* 0'025 11. ,* 2.0 M , , c . 2 D. K. HALE AND D.REICHENBERG above I M. In addition i t was shown that the rate of exchange increased with increase in hydroxyl ion concentration and was independent of hydroxyl ion concentration above I M. Variations in flow-rate from 15 cm./sec. to 30 cm./sec. did not affect the rate of exchange but the shallow-bed technique using M NaCl gave faster rates of exchange than the indicator method. The results obtained with cross-linked polymethacrylic acid using the shallow- bed method and keeping the flow-rate and Na+ concentration virtually constant are given in Fig. 9. It will be seen that increase in hydroxyl ion concentration markedly increases the rate of exchange. Discussion Exchange Equilibria.-From the study of the exchange equilibria with sulphonated cross-linked polystyrene we conclude that if the [Na+]/[H+] ratio in solution is greater than IOO the resin is fully ionized.It is not possible to say on the present evidence whether the resin is fully ionized at all values of [Na+]/[H+]. However it is clear that the resin behaves as a fairly strong acid and only one type of grouping appears to be present. Consideration of the exchange equilibria of the polymethacrylic acid resin H+R + Na+S +Na+lt + H+s leads to the mass-action relation activity coefficients being neglected. [N~+R] and [H+R] are the concentrations of Na+ ion and H+ ion in the resin in g. equiv./l. [Na+s] and [H+s] are the concentrations of Na+ ion and H+ ion in the solution in g. equiv./l. K is the relative affinity constant of sodium and hydrogen ions for the resin.By analogy with sulphonic acid resins we may expect K to have a value of from I to 2. In order to compare eqn. (I) with experimental data we must have a relationship between [H+R] and the total concentration of carboxylic hydrogen on the resin. We may assume that - I< . [H+Rl [Total carboxylic hydrogen] - where I< is a constant. This leads to the equation ”a+ ,7 . . [H+s] “a+R] - KK L. [Total carboxylic hydrogen] - I t has been found that an equation of this type fits the experimental data quite well with a value of 2-24 x 10% for KK,. Alternatively the ionization of the resin may be assumed to follow the laws holding for dilute solutions. We then write 7 (3) where [HR] and 1R-J are the concentrations of unionized and ionized resin in g.equiv./l. K may be expected to have a value of the same order (10-5 g. equiv./l.) as for the carboxylic group in simple compounds. From eqn. (I) and (4) and assuming electro-neutrality of the resin phase the following relation may be derived (5) [Na+lJ = _____ xK,h’ -+ -Ac 2 88 CA2TION-EXCHANGE RESINS where x = [Na+s]/[H+s ] c = total capacity of resin in g. equiv. per g. dry resin. a = volume of wet resin in l./g. dry resin. It is assumed that a is constant and independent of both [Na+sJ and [H+s]. While large variations in the volume of the resin are observed the effect is small considered in relation to the enormous variations in [Na+s]/[H+s]. Since K1el and [N~+R] only becomes significant for values of [Na+s]/ [H+,] greater than 103 r/K,x is less than 10-3 and may be neglected in comparison with unity.Therefore Expressing the amount of sodium ion on the resin in g. equiv. per g. dry resin we have From eqn. (5) we see that the amount of sodium taken up by the resin depends on the ratio [Na+s]/[H+s] and not on [Na+s] and [H+s] separately. -4s constants eqn. (7) contains only the capacity c and the parameter aK,K,. This latter may be estimated. K e I K e 1 0 3 . equiv./l. and since the density of the wet resin is approximately unity a e 10-3l./g. Hence aK,K e 10-8 g. equiv./g. dry resin. Alternatively a value for nK,K may be found from the experimental data. It may be shown from eqn. (7) that wheny = Qc This leads to a value of 1-05 x 10-8 g.equiv./g. for aK,K,. With c = 9-24 x 10-3 g. equiv./l. and aK,K = I O - ~ g. equiv./l. values of y were calculated for various values of log, [Na+s]/[H+s] using eqn. (7). The results are represented by the smooth curve in Fig. 3. The agreement between the calculated curve and the experimental points is good in view of the assumptions made. It is concluded that the behaviour of cross-linked polymethacrylic acid is well-accounted for by assuming that only a small fraction of the carboxylic hydrogen is ionized. The exact relationships governing the degree of ioni- zation is a matter for further investigation. Exchange Kinetics.-Boyd Adamson and Myers have studied the rates of exchange of various cations on the phenolic resin Amberlite IR-I.In their discussion they consider three mechanisms each of which might be rate-controlling under appropriate conditions (i) Diffusion of ions through a thin film of liquid surrounding the particles (or through liquid in macropores inside the particles). (ii) Diffusion of ions through the resin material itself. (iii) The chemical process of exchange. In the present work if the chemical process of exchange were the sole rate-controlling process the measured rate of exchange would be independent of the particle size. The fact that both a t high and low sodium ion con- centrations the rate varied markedly with particle size would seem t o rule out the chemical process as a major rate-controlling factor. If any type of film diffusion is a rate-controlling process this is because despite stirring rapid flow or any other attempt to make the solution D.K. HALE AND D. REICHENBERG homogeneous right up to the surface of the particle there is a thin film inside which mixing is imperfect. (The film may also be due to cracks or macropores in the particle ; in this case the rate of stirring will have practically no effect on the characteristics of the film.) If the diffusion of sodium ions across the film towards the resin is the sole rate-controlling process the concentration of sodium ions at the inside boundary of the film will be virtually zero (owing to the diffusion of ions into the particle being very rapid in comparison). Hence the concentration gradient (and the diffusion rate) of sodium ions across the film will be proportional to the concentration of sodium ions in solution.To a first approximation the film thickness may be assumed to be independent of particle size. The rate of exchange will then be inversely proportional to the particle diameter (provided macropores play a negligible part in film diffusion) since the surface area per particle is proportional to the square of the diameter and the capacity is proportional to the cube of the diameter. It is probable that the diffusion rates of sodium and hydrogen ions within the resin particle are largely coupled owing to the powerful electrostatic forces which oppose the entry of anions and also any variation of the total cation concentration in the resin. We may therefore speak of diffusion of ions within the resin being the rate-controlling process without specifying sodium or hydrogen ions separately.If diffusion of ions within the resin is the sole rate-controlling process this means that the effect of film diffusion has been eliminated and the surface of the resin is in equilibrium with the bulk of the solution. Since in all the present work with sulphonated poly- styrene the [Na+]/[H+] ratio in the solution was at least 104 the surface of the resin will be saturated with sodium ions. Hence diffusion of sodium ions into the interior of the resin particle will occur from a constant surface concentration which is independent of the sodium ion concentration in the solution. Hence the rate of exchange will be independent of the sodium ion concentration in the solution.If diffusion of hydrogen ions through the film away from the resin is the sole rate-controlling process the rate of exchange will be independent of sodium ion concentration but will be dependent on the hydroxyl ion con- centration of the solution. Hydroxyl ions will diffuse towards the surface of the particle and will neutralize hydrogen ions in the film thus increasing the rate of removal of hydrogen ions. The results obtained at low sodium ion concentrations (< 0.1 M) show exactly the characteristics expected if film diffusion of the sodium ions is the rate-controlling process. At high sodium ion concentrations the rate is independent of sodium ion concentration and hence the rate-controlling process might be either the diffusion of ions within the particle or the diffusion of hydrogen ions across the film or both.However Fig. 8 shows that using the shallow-bed technique increase in the hydroxyl ion concentration from zero to I M increased the exchange rate while a further increase to 2 M had no effect. Hence we may conclude that with M or z M NaCl solutions the film diffusion of hydrogen ions is rate controlling while with N or 2 N NaOH solutions diffusion of ions within the particle is the sole rate-controlling process. The diffusion coefficients of sodium hydrogen and hydroxyl ions are of the same order. Similar concentrations (I M) of hydroxyl or sodium ions in solution are therefore likely to be required to eliminate film diffusion of hydrogen and sodium ions as rate-controlling processes.No satisfactory explanation can be given for the observation that with the shallow-bed method increase in the flow-rate from 15 to 30 cm./sec. failed to increase the rate of exchange with NaCl solutions. The fact that at low sodium ion concentrations the rate was inversely proportional to particle diameter appears to preclude the possibility of macropores playing KINETICS OF CATION EXCHANGE 90 an appreciable part in film diffusion. We can only suppose that owing to the highly turbulent nature of the flow in our experiments the effective film thickness was not altered by the apparent flow-rate. We consider now in more detail the way in which at high Na+ concen- trations and virtually zero OH- concentrations the rate of diffusion of hydrogen ions through the film controls the exchange process.This rate of diffusion is initially zero but increases progressively as the concentration of hydrogen ions outside the particle surface builds up. However long before the rate of hydrogen ion film diffusion becomes equal to the rate of diffusion of ions within the resin the ratio [Na+]/[H+] in the solution immediately outside the particle surface will have fallen appreciably causing a decrease in the concentration of Na+ ions inside the particle surface. Hence the rate of diffusion of ions within the particle will decrease. When we consider the rate of Na-H exchange with the polymethacrylic acid resin it is apparent that the film diffusion of hydrogen ions will be much more important than with sulphonated polystyrene resins owing to the much larger [Na+]/[H+] ratio (10 *) in the solution necessary to maintain the surface of the resin particle saturated with sodium ions.The results (Fig. 9) confirm this expectation. In conclusion it appears that for the diffusion of ions within the resin to be the rate-controlling process it is necessary to have in solution high concentrations of both Na+ and OH- ions. The work described above has been carried out as part of the research programme of the Chemical Research Laboratory and this paper is published by permission of the Director of the Laboratory. The authors wish to thank their colleagues on the staff of the Chemical Research Laboratory for advice and assistance.Chemical Research Laboratory Teadington Middlesex. 79 EQUILIBRIUM AND RATE STUDIES OF CATION-EXCHANGE WITH MONOFUNCTIONAL RESINS BY D. K. HALE AND D. REICHENBERG Received 13th July 1949 The preparation of sulphonated cross-linked polystyrene and of cross-linked poly-methacrylic acid is described. The sodium-hydrogen exchange equilibria for both inaterials have been examined. The former resin is shown to behave as a monofunctional strong acid and the latter as a monofunctional weak acid. The application of the law of mass action to the exchange equilibria is discussed 80 CATION-EXCHANGE RESINS A study of the rate of sodium-hydrogen exchange with sulphonated cross-linked polystyrene suggests that a t low concentrations of sodium ions in solution the rate-determining mechanism is diffusion of ions through a thin film of liquid surrounding the resin particle.The influence of hydroxyl ion concentration on the rate of exchange of sodium for hydrogen with both resins is described and the conditions under which the diffusion of ions within the resin particle may become the rate-controlling process are discussed. In the application of ion-exchange resins to chromatographic techniques, more information is required on the fundamental molecular and ionic pro-cesses involved. Correlation of such information with the basic chemical and macromolecular structure may be expected to lead to the development of improved materials. In equilibrium and rate studies monofunctional resins obtained by addition polymerization e.g.sulphonated cross-linked polystyrene and cross-linked polymethacrylic acid offer many advantages. In contrast to earlier materials obtained by polycondensation e.g. sul-phonated phenol-formaldehyde resins these materials possess a fairly well-defined structure which may be systematically varied. They may also be examined over a wide range of pH without complicating factors arising due to the presence of different types of ionizable group. Moreover the technique of suspension polymerization enables the resins to be prepared in the form of spherical beads which are especially suitable for rate studies. In this paper the preliminary results of an investigation into the equili-brium and rate processes with sulphonated cross-linked polystyrene and cross-linked polymethacrylic acid are presented.Experimental Preparation of Cation- exchange Resins .-SULPHONATED CROSS-LI NKED POLYSTYRENE. This material was prepared by the sulphonation of a cross-linked polystyrene bead polymer as described by D'A1elio.l Styrene was co-polymerized a t 80" C for 18 hr. with ca. 10 yo divinylbenzene I y; benzoyl peroxide being employed as catalyst. The co-polymer was sulphonated with concentrated sulphuric acid a t 100" C for 8 hr. using I yo silver sulphate as catalyst. The maximum capacity of the product (5.25 milli-equivalents of basc per g. dry hydrogen form) was independent of particle size and agreed with the value calculated for a monosulphonic acid. The material was hygroscopic and the dry hydrogen form absorbed approximately 80 yo water a t zoo C.On conversion from the wet hydrogen form to the wet sodium form a decrease in volume of G y0-7 yo was observed. CROSS-LINKED POLYMETHACKYLIC ACID. The carboxylic type exchange resin was prepared from methacrylic acid and divinylbenzene. Commercial methacrylic acid redistilled 2.12 uacuo was polymerized with ca. 10 yo divinylbenzene in the presence of I yo benzoyl peroxide. Polymerization was carried out in a sealed tube a t GO" C for 24 hr. The product was treated with z N NaOI-I to remove soluble material washed and dried. The maximum capacity (9.2 milli-equivalents of base per g. dry hydrogen form) was consistent with the value calculated from the composition of the monomer mixture. The dry material absorbed approximately 130 o//o of water a t 20" C and an increase in volume of approximately 75 Yo was observed on conversion from the wet hydrogen to the wet sodium form.After preliminary cycling in a column between 2 M NaCl or 2 N NaOH and z N HC1 the cation-exchange resins were converted to the hydrogen form and washed with distilled water. Washing was continued until the pH of the effluent attained a value of 4.0 or higher. If necessary fines were removed by elutriation and the ion-exchange resins were air-dried to a uniform ~noisturc content. D'Alelio US. Pat. 2,366,007 D. K. HALE AND D. REICHENBERG 81 Equilibrium Studies .-Samples of the cation-exchange resins in thc. hydrogen €orm were weighed out into a series of bottles and further samples \\-ere taken for moisture content determinations.The latter were dried to constant weight over P,O in a vacuum desiccator Different amounts of NaOH solution with or without NaCl solution were added to the samples of resin and the solutions made up to 50 ml. For the sulphonated cross-linked poly-styrene the liquid-solid ratio was IOO/I for the cross-linked polymethacrylic acid ZOO/I. The solutions were allowed to stand with occasional shaking until equilibrium was attained (1-7 days). Aliquot samples of the solutions were then withdrawn and titrated with standard HC1 or NaOH to determine the extent of exchange. The pH values of the solutions were determined using a Cambridge pH meter and glass electrode. For the measurement of pH values greater than 9.0 a Cambridge Alki electrode was employed. Rate Studies .-To obtain samples of resin of approximately uniform particle diameter the resins were sieved in the air-dry state using calibrated B.S.sieves. For the determination of rates of exchange two methods were employed. FIG. I .-Sulphonated cross-linked polystyrene. Relationship between Na+ ion taken up by the resin and pH. A. In presence of 5 11 NaC1. B. In presence of 0-1 M NaC1. c. In absence of NaCl. (a) INDICATOR METHOD. This method which is only applicable to sulphonated cross-linked polystyrene depends on the fact that whether the solution is acid or alkaline exchange will proceed virtually to completion if the ratio [Na+]/[H'] in solution is sufficiently high (see below). The hydrogen form of the resin is stirred with a solution of NaCl and NaOH the latter being less than sufficient to neutralize the hydrogen ions liberated in the exchange process.The solution, initially alkaline becomes acid when the amount of exchange just exceeds the amount of alkali originally added. An indicator preferably of the anionic type is used to show this change. A weighed amount of resin of known moisture content was added to a known volume of water containing a few drops of bromo-cresol green indicator solution (0.04 yo solution in water) in a small beaker. The mixture was stirred vigorously with a magnetic stirrer and a suitable mixture of NaC1 and NaOH solutions added. The time elapsing between the addition of the alkali solution and the colour transition (blue+yellow) was measured with a stop-watch. All experiments werecarried out a t room temperature (18'-22' C) and with a constant volume of solution.(b) SHALLOW-BED METHOD. A simple modification of the method used by Boyd ,\damson and Myers was employed. The resin sample (cn. 0.1 g.) was supported 1 Boyd Adamson and Myers J . Amer. Chcm. SOC. 1947 69 2836 82 CATION-EXCHANGE RESINS on stainless steel gauze or a sintered-glass disc. The resin was first converted to the hydrogen form and washed free of acid. The solution containing sodium ions was then passed through the bed for an appropriate time a t a known flow-rate. The bed was then immediately washed with a stream of distilled water. In the case of the sulphonated cross-linked polystyrene the amount of residual hydrogen ion was determined by displacement using an excess of NaCl solution and titrating the solution with standard alkali.For the cross-linked poly-methacrylic acid the amount of exchange was determined by removing the sodium ion with a measured volume of standard acid and back titration of the acid solution. Results Exchange Equilibria.-The amount of sodium ion taken up a t equilibrium by the resins and its dependence on pH and on the ratio [Na+]/[H+] in solution is shown in Fig. I z and 3. 1 ; ~ . 2 .-Cross-linked polymethacrylic acid. Relationship between Na+ ion taken c. I n presence oi up by the resin and pH. A . I n presence of 2 hI NaCl. B. In presence of I M NaCl. 0-1 M NaCl. D. In absence of NaCl. I t will be seen from Fig. 3 that the Na+ ion taken up by both resins was dependent only on [Na']/[H+] in solution and not on "a+] or [H+] separately.In the case of the sulphonic acid type exchanger provided the [Na+]/[H+] ratio in solution is greater than IOO/I virtually complete replacement of hydrogen by sodium ion is effected. For the exchange resin containing carboxylic groups, a [Xa+]/[Hi] ratio of a t least I O ~ / I is necessary to effect complete conversion to the sodium form. Exchange Kinetics.-Using the indicator method it was found that above a minimum rate of stirring the results obtained were independent of the stirring rate. This was found to apply over the whole range of sodium ion concentrations studied. Under the conditions employed however the degree of mixing of resin and solution may be expected to be lower than that attained with the shallow-bed method D.K. HALE AND D. REICHENBERG FIG. 3.-Relationship between Na+ ion taken up by the resin and log, [Na,+]/[Hs+]. B. Sulphonated cross-linked polystyrene. A . Cross-linked polymethacrylic acid. FIG. 4 .-Sulphonated cross-linked polystyrene. Exchange kinetics a t low Na+ ion concentrations. (Indicator method .) A. Air-dry particle diameter 50-100 “a,+] 0.048-0.050 M B. i a I 1 I “a,+] 0.026-0.029 M c. I 1 I1 300-400 p [Na +] 0.045-0.050 hT D. 11 1 8 1 “a,+] 0.023-0.028 84 CATION-EXCHANGE RESINS lo. I 77me - secondJ 10 20 30 I I‘ G . j.-Sulphonated cross-linked polystyrene. Exchange kinetics a t high Ka+ ion concentrations. (Indicator method.) A. Air-dry particle J V “a,+] 2-18 M diameter 50-100 A “a,+] 1-09 M H. Air-dry particle O[Na,+] 2.18 M diameter 300-400 p { O[Na,+] 1.09 M ---- - I t I (Jec -‘) -I 1 1 I i I I I Meun Ion I 1 1.5 ? .O yo- sodurn ion cancentrahon y ronllike /. 5 FIG. 6.-Sulphonated cross-linked polystyrene. Relationship between half-life and Na + ion concentration. (Indicator method.) Air-dry particle diameter 300-400 EL D. K. HALE AND D. REICHENBERG 85 The effect of sodium ion concentration in solution and the particle size of the resin were examined using the indicator method. The results obtained are shown in Fig. 4 and 5 . A t low sodium ion concentrations as shown in Fig. 4 the exchange proceeds initially at an approximately constant rate but then slows down progressively as the exchange proceeds. At high sodium ion concentrations (Fig. 5) the form of curve obtained is similar but owing to the high rate of exchange under these conditions the initial rate of exchange cannot be determined accurately by the present method.The reciprocal of the time for half-conversion of the resin to the sodium form is plotted against the mean sodium ion concentration in solution in Fig. 6. It appears that a t high sodium ion concentrations the rate of exchange is independent of the sodium ion concentration whilst at low sodium ion con-centrations the rate is proportional to the sodium ion concentration. This is shown also in Fig. 7 where the initial rate of exchange has been plotted against the mean sodium ion concentration in solution. FIG. $.-5julphonated cross-linked polystyrene. Relationship between rate of exchange and Xa+ ion concentration a t low Na+ ion concentrations.(Indicator method.) Air-dry particle diameter 300-400 u. A t low sodium ion concentrations there is a change in the sodium ion concen-It has however been found that the results tration as the reaction proceeds. are in agreement with the relationship d (Na) /dt = Kw[Na’-], where (Naj is the total amount of exchange w the weight of resin [Na ] the sodium ion concentration in solution and K a constant. It will be seen from Fig. 4 that at low sodium ion concentrations the rate of exchange with particles of diameter 50-100 p is about four times as great as that with particles of diameter 300-400 v. Thus a t low sodium ion concentra-tions the rate of exchange is approximately inversely proportional to the particle diameter.At high sodium ion concentrations no quantitative conclusions can be drawn from the present data but i t is apparent from Fig. 5 that exchange takes place more rapidly with the smaller particles. The indicator method cannot be used for the direct investigation of the effect of hydroxyl ion concentration on the rate of exchange and the shallow-bed method was employed for this purpose (see Fig. 8). Results obtained by the indicator method were confirmed in that with NaCl solutions the exchange rate was independent of sodium ion concentratio 86 CATION-EXCHANGE RESINS 7;me - seconds 20 .I /O / 5 FIG. 8.-Sulphonated cross-linked polystyrene. Exchange kinetics at high Na+ ion .4. Shallow-bed method. 2 N NaOH and I N NaOH solutions. Now-rates R.Shallow-bed method. 2 14 NaCl and I &I NaCl solutions. Flow-rates c. Indicator method. concentration. 15 cm./sec. and 30 cm./sec. 15 cm./sec. and 30 cm./sec. Air-dry particle diameter 300-400 p. Na+ ion concentration 2 M and I M. FIG. 9.-Cross-linked polymethacrylic acid. Exchange kinetics a t high Na + con-centrations. Air-dry particle diameter 250-380 p. Flow-rate I cm./scc. A. Na+ concentration 2-2 M OH- concentration 0.2 M. B. ,* 2'1 M , I > 0'1 M. c . 2 , 2.0 M , ,* 0'025 11 D. K. HALE AND D. REICHENBERG above I M. In addition i t was shown that the rate of exchange increased with increase in hydroxyl ion concentration and was independent of hydroxyl ion concentration above I M. Variations in flow-rate from 15 cm./sec. to 30 cm./sec. did not affect the rate of exchange but the shallow-bed technique using M NaCl gave faster rates of exchange than the indicator method.The results obtained with cross-linked polymethacrylic acid using the shallow-bed method and keeping the flow-rate and Na+ concentration virtually constant, are given in Fig. 9. It will be seen that increase in hydroxyl ion concentration markedly increases the rate of exchange. Discussion Exchange Equilibria.-From the study of the exchange equilibria with sulphonated cross-linked polystyrene we conclude that if the [Na+]/[H+] ratio in solution is greater than IOO the resin is fully ionized. It is not possible to say on the present evidence whether the resin is fully ionized at all values of [Na+]/[H+]. However it is clear that the resin behaves as a fairly strong acid and only one type of grouping appears to be present.Consideration of the exchange equilibria of the polymethacrylic acid resin H+R + Na+S +Na+lt + H+s leads to the mass-action relation activity coefficients being neglected. [N~+R] and [H+R] are the concentrations of Na+ ion and H+ ion in the resin in g. equiv./l. [Na+s] and [H+s] are the concentrations of Na+ ion and H+ ion in the solution in g. equiv./l. K is the relative affinity constant of sodium and hydrogen ions for the resin. By analogy with sulphonic acid resins we may expect K to have a value of from I to 2. In order to compare eqn. (I) with experimental data we must have a relationship between [H+R] and the total concentration of carboxylic hydrogen on the resin.We may assume that I< . [H+Rl -[Total carboxylic hydrogen] -where I< is a constant. This leads to the equation ”a+ ,7 “a+R] - KK L. . . [Total carboxylic hydrogen] - [H+s] (3) I t has been found that an equation of this type fits the experimental data quite well with a value of 2-24 x 10% for KK,. Alternatively the ionization of the resin may be assumed to follow the laws holding for dilute solutions. We then write where [HR] and 1R-J are the concentrations of unionized and ionized resin in g. equiv./l. K may be expected to have a value of the same order (10-5 g. equiv./l.) as for the carboxylic group in simple compounds. From eqn. (I) and (4) and assuming electro-neutrality of the resin phase, the following relation may be derived : [Na+lJ = _____ xK,h’ -+ -Ac 7 (5) 88 CA2TION-EXCHANGE RESINS where x = [Na+s]/[H+s ] c = total capacity of resin in g.equiv. per g. dry resin. a = volume of wet resin in l./g. dry resin. It is assumed that a is constant and independent of both [Na+sJ and [H+s]. While large variations in the volume of the resin are observed the effect is small considered in relation to the enormous variations in [Na+s]/[H+s]. Since K1el and [N~+R] only becomes significant for values of [Na+s]/ [H+,] greater than 103 r/K,x is less than 10-3 and may be neglected in comparison with unity. Therefore Expressing the amount of sodium ion on the resin in g. equiv. per g. dry resin we have From eqn. (5) we see that the amount of sodium taken up by the resin depends on the ratio [Na+s]/[H+s] and not on [Na+s] and [H+s] separately.-4s constants eqn. (7) contains only the capacity c and the parameter aK,K,. K e I K e 1 0 3 . equiv./l. and since the density of the wet resin is approximately unity a e 10-3l./g. Hence aK,K e 10-8 g. equiv./g. dry resin. Alternatively a value for nK,K may be found from the experimental data. It may be shown from eqn. (7) that wheny = Qc, This latter may be estimated. This leads to a value of 1-05 x 10-8 g. equiv./g. for aK,K,. With c = 9-24 x 10-3 g. equiv./l. and aK,K = I O - ~ g. equiv./l., values of y were calculated for various values of log, [Na+s]/[H+s] using eqn. (7). The results are represented by the smooth curve in Fig. 3. The agreement between the calculated curve and the experimental points is good in view of the assumptions made.It is concluded that the behaviour of cross-linked polymethacrylic acid is well-accounted for by assuming that only a small fraction of the carboxylic hydrogen is ionized. The exact relationships governing the degree of ioni-zation is a matter for further investigation. Exchange Kinetics.-Boyd Adamson and Myers have studied the rates of exchange of various cations on the phenolic resin Amberlite IR-I. In their discussion they consider three mechanisms each of which might be rate-controlling under appropriate conditions : (i) Diffusion of ions through a thin film of liquid surrounding the particles (or through liquid in macropores inside the particles). (ii) Diffusion of ions through the resin material itself. (iii) The chemical process of exchange.In the present work if the chemical process of exchange were the sole rate-controlling process the measured rate of exchange would be independent of the particle size. The fact that both a t high and low sodium ion con-centrations the rate varied markedly with particle size would seem t o rule out the chemical process as a major rate-controlling factor. If any type of film diffusion is a rate-controlling process this is because, despite stirring rapid flow or any other attempt to make the solutio D. K. HALE AND D. REICHENBERG homogeneous right up to the surface of the particle there is a thin film inside which mixing is imperfect. (The film may also be due to cracks or macropores in the particle ; in this case the rate of stirring will have practically no effect on the characteristics of the film.) If the diffusion of sodium ions across the film towards the resin is the sole rate-controlling process the concentration of sodium ions at the inside boundary of the film will be virtually zero (owing to the diffusion of ions into the particle being very rapid in comparison).Hence the concentration gradient (and the diffusion rate) of sodium ions across the film will be proportional to the concentration of sodium ions in solution. To a first approximation the film thickness may be assumed to be independent of particle size. The rate of exchange will then be inversely proportional to the particle diameter (provided macropores play a negligible part in film diffusion) since the surface area per particle is proportional to the square of the diameter and the capacity is proportional to the cube of the diameter.It is probable that the diffusion rates of sodium and hydrogen ions within the resin particle are largely coupled owing to the powerful electrostatic forces which oppose the entry of anions and also any variation of the total cation concentration in the resin. We may therefore speak of diffusion of ions within the resin being the rate-controlling process without specifying sodium or hydrogen ions separately. If diffusion of ions within the resin is the sole rate-controlling process this means that the effect of film diffusion has been eliminated and the surface of the resin is in equilibrium with the bulk of the solution. Since in all the present work with sulphonated poly-styrene the [Na+]/[H+] ratio in the solution was at least 104 the surface of the resin will be saturated with sodium ions.Hence diffusion of sodium ions into the interior of the resin particle will occur from a constant surface concentration which is independent of the sodium ion concentration in the solution. Hence the rate of exchange will be independent of the sodium ion concentration in the solution. If diffusion of hydrogen ions through the film away from the resin is the sole rate-controlling process the rate of exchange will be independent of sodium ion concentration but will be dependent on the hydroxyl ion con-centration of the solution. Hydroxyl ions will diffuse towards the surface of the particle and will neutralize hydrogen ions in the film thus increasing the rate of removal of hydrogen ions.The results obtained at low sodium ion concentrations (< 0.1 M) show exactly the characteristics expected if film diffusion of the sodium ions is the rate-controlling process. At high sodium ion concentrations the rate is independent of sodium ion concentration and hence the rate-controlling process might be either the diffusion of ions within the particle or the diffusion of hydrogen ions across the film or both. However Fig. 8 shows that using the shallow-bed technique increase in the hydroxyl ion concentration from zero to I M increased the exchange rate while a further increase to 2 M had no effect. Hence we may conclude that with M or z M NaCl solutions the film diffusion of hydrogen ions is rate controlling while with N or 2 N NaOH solutions diffusion of ions within the particle is the sole rate-controlling process.The diffusion coefficients of sodium hydrogen and hydroxyl ions are of the same order. Similar concentrations (I M) of hydroxyl or sodium ions in solution are therefore likely to be required to eliminate film diffusion of hydrogen and sodium ions as rate-controlling processes. No satisfactory explanation can be given for the observation that with the shallow-bed method increase in the flow-rate from 15 to 30 cm./sec. failed to increase the rate of exchange with NaCl solutions. The fact that at low sodium ion concentrations the rate was inversely proportional to particle diameter appears to preclude the possibility of macropores playin 90 KINETICS OF CATION EXCHANGE an appreciable part in film diffusion.We can only suppose that owing to the highly turbulent nature of the flow in our experiments the effective film thickness was not altered by the apparent flow-rate. We consider now in more detail the way in which at high Na+ concen-trations and virtually zero OH- concentrations the rate of diffusion of hydrogen ions through the film controls the exchange process. This rate of diffusion is initially zero but increases progressively as the concentration of hydrogen ions outside the particle surface builds up. However long before the rate of hydrogen ion film diffusion becomes equal to the rate of diffusion of ions within the resin the ratio [Na+]/[H+] in the solution immediately outside the particle surface will have fallen appreciably causing a decrease in the concentration of Na+ ions inside the particle surface. Hence the rate of diffusion of ions within the particle will decrease. When we consider the rate of Na-H exchange with the polymethacrylic acid resin it is apparent that the film diffusion of hydrogen ions will be much more important than with sulphonated polystyrene resins owing to the much larger [Na+]/[H+] ratio (10 *) in the solution necessary to maintain the surface of the resin particle saturated with sodium ions. The results (Fig. 9) confirm this expectation. In conclusion it appears that for the diffusion of ions within the resin to be the rate-controlling process it is necessary to have in solution high concentrations of both Na+ and OH- ions. The work described above has been carried out as part of the research programme of the Chemical Research Laboratory and this paper is published by permission of the Director of the Laboratory. The authors wish to thank their colleagues on the staff of the Chemical Research Laboratory for advice and assistance. Chemical Research Laboratory, Middlesex. Teadington

ISSN:0366-9033    

DOI:10.1039/DF9490700079

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

CATION EXCHANGE WITH A SYNTHETIC PHENOLSULPHONATE RESIN Part V. Kinetics BY T. R. E. KRESSMAN AND J. A. KITCHENER Received 11th Jztly 1949 -1 study has been made of the kinetics of exchange between solutions of various simple inorganic and substituted quaternary ammonium salts and the ammonium form of the sulphonated phenol-formaldehyde resin for which equilibrium measure- ments have been reported elsewhere. Two mechanisms are observed where the rate is controlled by diffusion in the particles of the exchanger (P-mechanism) and in the bounding Kernst film (F-mechanism) respectively. These are distinguished by the form of the kinetics by interruption tests and by the influence of stirring. The factors that decide which mechanism applies in a given system are discussed.Measurements of the influence of temperature suggest that so long as the cation is small compared with the pores of the resin (e.g. Na+ and KMe,+ with the present resin) the energy of activation for diffusion is ca. 5 kcal./mole as for free diffusion in water. Larger ions which have difficulty in penetrating show a higher value e.g. 8 kcal./mole with PhNMe ,CH ,Ph +. Exchange occurs readily between a granular resin and a finely divided resin in wspension proving that soluble anions are not needed to release cations from the resin. The high mobility of the cations is consistent with a diffuse donble-layer model. T. R. E. KRESSMAN AND J. A. KITCHENER 91 The potentialities of ion exchange separation techniques can be fully exploited only if consideration is given to both equilibrium and kinetic aspects.These are largely independent and they differ with different ion exchangers even though the exchange groups may be the same. There is no simple relationship between affinities and rates of exchange for a series of ions although with the phenolsulphonate resin used in the present study the rate generally decreases as the affinity increases. The possibilities of separations based on differences of rates of exchange have not yet been fully explored. Fig. I shows an example where such an effect might be employed the concentrations of H+ NH4+ and NEt4+ in a solution (initially equimolar with respect to NH,+ and NEt,+) are shown during the course. of an exchange experiment with a H-resin (contain- ing -SO,- groups) in a stirred system.It is seen that the equilibrium affinity of NEt4+ for the resin is greater than that of -NH4+ but the rate of exchange is greater for the NH4+ ion. If the experiment were stopped after 2 or 3 min. the solution would be relatively enriched in NEt,+ but at equilibrium the reverse would apply. Both equilibria and kinetics for a series of ions depend very greatly on the structure of the solid exchanger. The equilibrium relationships for a phenoIsulphonate resin similar to Zeo-Karb 215 have already been described elsewhere in Parts I-IV of this series of papers.1 The present paper describes studies of the kinetics of exchange with the same resin. Materials.-The sample of resin was taken from the same batch of H-resin as that used in earlier studies.1 The general technique of converting the material to the salt form and of determining the equivalent weights air drying etc.is the same as described there. With the exception of the experiments illustrated in Fig. I where the H-form of the resin was used the NH,-form was used through- out the work and after preparing it from the H-resin i t was air-dried and sieved FIG. I. Experiments and Results - 10 + 22 mesh. Two samples were sieved out and all the results except those illustrated in with the second sample and although the resin granules were also - 10 + 22 Fig. 8 were obtained with one of these samples. Those in Fig. 8 were obtained mesh a larger fraction of the larger mesh size was present and the velocity of exchange is thus somewhat lower with this sample than with the first.1 Kressman and Kitchener J . Cltcm. Soc. 1949 1190. 92 KINETICS OF CATION EXCHANGE In the series of experiments illustrated in Fig. I the H-form of the resin \\as used and because of its slight instability,f it was prepared by thoroughly washing the free acid from a sample taken from the main batch air-drying rapidly and sieving - 10 +22 mesh ; i t was then used immediately. In this way the experi- ments were completed before the H-resin had begun to hydrolyze. large number of particles-of the order of 500-were counted allowed to swell in water the water removed and the resin granules mopped L)etween filter papers until the surface moisture had been removed. The swollen granules were then weighed.The density of the swollen granules \%as determined with the aid of a density bottle in the usual way. The effective sphere radius Y of the swollen granules is calculated from C m. Determination of Effective Sphere Radius of the Swollen Resin Granules .-A 413 . 7 ~ ~ p N = W where p is the density of the swollen resin N the number of granules arid IT’ the weight of the N granules. C FIG. 2. SecLbn uf x x The first sample of KH,-resin which was used for the majority of the work (see under Materials) was found in this way to have an effective sphere radius of 0.446 mm. ; and the second sample used for the experiments illustrated in Fig. 8 an effective sphere radius of 0.517 mm. The density of the swollen granules was 1-33 g./cm.3 Determination of Exchange Velocity.-The limited bath method was used throughout the work and not the column technique with infinite bath as used hy Boyd et The essential requirements in such heterogeneous studies of standardized vigorous agitation and the necessity of starting and stopping the reaction sharply were achieved by using a centrifugal type of stirrer as shown i n Fig.2 a. The resin was placed in the wire-gauze cage and as the stirrer rotated in the solution a very rapid stream of the solution passed over the resin which was forced into a roughly cylindrical “ wall ” on the insitlc of the cage (see Fig. 2 b). The constancy of speed of rotation of the stirrer x i as followed stroboscopically. H y using several stroboscope discs constant speeds of rotation from 300 rev.jmin. a t intervals to 1200 rev./min. were attained. The reaction was started by lowering the already rotating stirrer with the resin in the cage into the solution ; and i t was stopped after the appropriate interval of time by raising it. still rotat- Boyd. Adamson and RIj-ers J . A v i i e ~ . C‘kem. SOC. 1947 69 2 8 ~ 1 T. R. E. KRESSMAN AND J. A. KITCHENER 9.; ing out of the solution. An aliquot portion of the solution was then analyzed for NH,+ or H7- as appropriate. Temperatures between 14O and 45" C were thermostatically controlled to within -+ 0.2" ; outside this range the limits were FIG. 3.-Effect of stirring on the time for half-change. r . Centrifugal stirrer Na+. 1. Centrifugal stirrer NEt,+.3. Glass stirrer NEt4+. - -L 0.50. The weight of resin taken in every run was such that it contained 2-5 m.equiv. o f exchangeable cation and 2.5 m.equiv. of the other cation were also present in the aqueous phase in the form of 125 ml. 0.02 N solution. Fig. 3 shows that the mixing is a t a maximum when the speed of rotation is between 1000 and 1100 rev./min. ; higher speeds caused entrainment of air and accordingly an apparent decrease in velocity. The speed of rotation was therefore standardized a t 1000 rev./min. throughout the work. Influence of Agitation.-Fig. depends upon the speed of rotation of the stirrer for two systems NH4R + NaCl 3 shows how the time for half-change and NH,R NEt,Br respectively. The time for half-change in the first system decreases as the speed of rotation is increased approaching a minimum at 1000 rev./min.when the stationary liquid film surrounding the particles has reached a minimum. (Above about 1100 rev./min. entrainment of air occurs and erratic results are consequently obtained.) No greater velocity was obtained even when the solution was circulated over the resin (in the form of a shallow bed) extremely rapidly by means of a pump. The time for half-change in the NEt ' system is also practically at a minimum a t 1000 re\jniin. but unlike the Na+ system it remains at a minimum at speeds down to 750 rev./min. and only then when very little relative motion of solid and liquid Is occurring does it begin to increase. Fig. 3 also shows a curve obtained in the NEt4+ system with a simple bent qlass-rod stirrer (Fig.2 6 ) . This stirrer is seen to be more efficient a t slower ipeeds of rotation but at higher speeds the centrifugal stirrer results in a thinner KINETICS OF CATION EXCHA4NGE 94 liquid film than does the simple stirrer. Evidently so long as the glass stirrer keeps the particles in suspension increased agitation has an immeasurable effect upon the thickness of the liquid film. Preparation of NEt4-resin in Fine Suspension.-A series of rate experi- ments was carried out in which no soluble anions whatever were present in the system the exchange occurring between the ordinary NH,-resin and a fine aqueous suspension of NEt,-resin. This suspension was prepared by milling well-washed NEt,-resin with water.A sample of H-resin was washed free of acid and converted to the NEt,-form with a solution of NEt,Br. The rate of reaction is considerably less than with ammonium chloride and a lower rate of flow was accordingly used and a greatei- volume of solution was found necessary to remove the whole of the exchangeable hydrogen ions from the resin. The resin-about 40 g.-was thoroughly washed to remove all traces of electrolytes and then ground in a laboratory porcelaiil ball mill for about 24 hr. with about I 1. water. The milky suspension so obtained was allowed to settle for about I hr. and the liquid decanted. This consisted of a suspension of the finest particles the largest of which were of the order of OH NH2 I [A diam. It was adjusted by dilution to contain 0.02 equiv.of exchangeable SEt,+ per 1. of suspension as indicated by a Kjeldahl nitrogen determination. Care was taken when using this sus- pension to ensure that homogeneous samples were taken from it. It was kept gently shaken agitated just during prior to the and re- are 3 3 0 gNQ% vigorously OH NH2 "="Q3;3- moval of the sample. The NI-I,-resin was washed with water before each run was started to remove any traces of electrolytes which might be con- taminating it. The velocity determination with this suspension was carried out exactly as with the aqueous salt solutions. The results :"3O FIG. 4.-Anion of Chlorazol Sky Blue FFS. shown in Fig. 5 curve 10 and Fig. I I curve 6 and graphs of Qt/Q against t6 for the NEt,-resin suspension and for NEt,Br indicate that the apparent rate constants are in the ratio of 0*74/1.sodium salt was prepared from the crude dye by dissolving it in water and salting-out with sodium acetate. This was repeated three times. Adhering sodium acetate was finally removed by washing repeatedly with alcohol. The purified product was dried a t IOOO C. Preparation of the Salts of Chlorazol Sky Blue FFS.-The The NEt,-salt was prepared in solution from NEt40H and the pure dye acid. The NEt,OH was prepared in known concentration (about 0.2 N) by adding excess moist silver oxide to a solution of NEt,Br and filtering. The dye acid was prepared from the pure sodium salt by passing quantitatively a dilute solution (about 0.01 N) containing 4-96 g.of the salt through a IOO ml. column of Zeo-Karb 215 (The Permutit Co. Ltd.) containing exchangeable hydrogen ions. The solution was then evaporated on the water bath until its concentration was of the order of 0.03 N cooled and " titrated '' with the NEt,OH to pH 7 using a pH meter. The quantity of NEt,OH required was exactly that calculated from its known concentration. The solution of the NEt,-dye salt was then diluted to I 1. In the runs with the dye salts as in those with the NEt,-resin suspension the resin was washed with water before each run was started to remove traces of electrolytes. Velocity with Different Cations .-Since the equilibrium positions are very different with the various cations and since only the kinetics are at present being considered the extent of exchange at time t is expressed as Qt/O!, i.e.the fraction of the amount of exchange occurring a t time t to that occurring at equilibrium. Fig. 5 shows the graphs of Qt/Q against t for a number of different T. R. E. KRESSMAN 4ND J. A. KITCHENER 9s cations at 25" C. The same results are plotted in Fig. 6 as - log, ( I - Qt/Qc?,) against t and Qt/Qm against t4 for analysis of mechanism-see below. activation energies of the exchange reactions were determined from velocity measurements at several temperatures with the NH,-resin and respectively Na+ NMe,+ and PhNMe2CH,PhA. kcal./mole range 5.1 1'-61" Influence of Temperature .-The of exchange between NH,-resin and various cations a t 2 5 O C. FIG. .=j.-Rate I .K+ (chloride). 2. Na+ 0 chloride x interrupted A Chlorazol Sky Blue salt. 3- Li+ (chloride). 4. Mg++ and Ba++ 0 chloride A MgSO,. 5. NMe + (bromide). 6 . A1 3- + + (chloride). 7. NEt,+ (bromide). 8. NEt,' (bromide) interrupted at 5 min. for 30 min. 9. NMe ,.n-Amyl+ (bromide). 10. PhNMe,Et+ (bromide) and NEt,-resin suspension. I I . Th++++ (nitrate). 12. Th++++ (nitrate) interrupted at 2.5 min. for 20 min. 13. PhNMe,CH ,Ph+ (chloride). The results - are shown in Fig. 7 8 and 9 where they are plotted in terms of -1og,,(1 - Qt/Qa.) against t and QJQ against 14 as appropriate. The logarithms of the limiting slopes of the lines so obtained are plotted against the reciprocals of the absolute temperature in Fig. 10. The values for the activation energies given in Table I are obtained from the slopes of these straight lines.TABLE I Cation 2 O- I 4- 8 2 5 "-45 * 7 Na+ (P) Na+ (F) NMe + 5.1 8.7 YhNMe2CH2Phi 96 KINETICS OF CATION EXCHANGE (1%) Discussion The structure of the phenolsulphonate exchange resin is a three-dimensional rigid network resembling a sponge and containing five -SO groups to every seven phenol residues (see Part I).l The cations are highly mobile inside the water-filled interstices and form a Gouy diffuse double layer round each resin granule. Consequently cation exchange does not involve any " chemical " step since no covalent bond has to be broken and the rate of exchange depends simply on one or more of a series of consecutive transport steps-namely forced convect ion of the fluid diffusion through the unmixed boundary FIG.6.-Tests of mechanism. I . K+. 2. I i f . 8 . 7- NMe,+. NEt,+. 3 . Ag+. 4. Mg++ and Ha". I I . j. N b t + . Na + (2 5 ") . 9. 10. NMe,.n-Amyl+. PhNRiIe,Et+ and NEt,-resin suspension. 13. PhNMe ,CH ,Pi> + . 13. Th++++. 6. Al+++. layer and diffusion inside the particles. As Boyd et aL2 have shown three possible types of kinetics may therefore be encountered in a practical dynamic exchange process (i) Diffusion in the boundary liquid film is rate-determining (F-mechanism) . (ii) Diffusion in the solid particle is rate-determining (P-mechanism). (iii) In an intermediate region the rate is influenced by diffusional resistances in both phases (I-mechanism).The factors that may decide which mechanism applies in a given system are (I) particle size (2) degree of agitation of the solution (3) diffusion coefficients of the ions in the solution and inside the resin particles (4) temperature (5) the equilibrium distribution coefficient and (6) solution concentra- tion. Variation of any one of these factors may produce a change of mechanism. For example Boyd et aL2 have shown that the Na+-K+ exchange (under their conditions of experiment) is F at 0.001 M ionic strength I at 0.01 M and P at 0-1 M. T. R. E. KRESSMAN AND J. A. KITCHENER 97 Criteria for Distinguishing Mechanisms. (a) Influence of Stirring.-The rate increases with the degree of mixing of the liquid as long as the F- or I-mechanisms are rate-controlling whereas with P-mechanism it should be independent.With increasing speed of rotation of a given stirrer a maximum exchange rate may be reached beyond which the rate may remain constant or even decrease (Fig. 3). Only exceptionally does such a maximum indicate the point where 1-mechanism gives place to P. More usually the rate reaches a limit for hydrodynamical reasons the stirrer having reached its limiting efficiency for mixing and it may still be possible to achieve faster rates of exchange by using a different type of stirrer (Fig. 3). However in border-line cases a change of mechanism may sometimes be detected by the influence of stirring-notably when a maximum exchange rate is reached in the stirring range slower than the known hydrodynamical limit.The exchange with NEt4+ in Fig. 3 provides an instance of this. ( a ) (b) I . 2'03 c 4. 25'0°C 5. 31'8" 6. 45'7" FIG. 7.-Influence of temperature W a f . 0 Chloride i' interrupted A Chlorazol Sky Blue salt Discussiom 1947 I 37. 2. 8.4' j. 14.8" (b) Form of the Kinetics.-@) FILM DIFFUSION (F-MECHANISM) .-The well-known Nernst static diffusion film theory is clearly only a crude approxi- mation for the complex situation which exists near an irregular solid surface in a stirred liquid; in particular the "thickness" of such a layer is a mathematical fiction which has proved of little use since this quantity must usually be deduced from the kinetics (not vice versa) and its dependence on stirring temperature viscosity etc.is problematical. Only in a few idealized cases has it been possible to calculate the diffusional transport up to a body in a stirred l i q ~ i d . ~ However the Nernst layer approximation is useful for treating the kinetics of exchange under fixed conditions of stirring and temperature.2 In the present system let Qo be the number of milli-equivalents of the pure A+ form of the resin which at time t = o is brought into contact with V ml. solution containing Qo milli-equivalents of a salt of the cation B+. Suppose the resin consists of n particles of mean equivalent sphere radius Y 3Levich Acta Physicochiw 1942 17 257; 1944 19 117 133 Faraday SOG. D KINETICS OF CATION EXCHANGE 98 and that the effective Nernst diffusion layer thickness is 8.Let the amount of exchange which has occurred after time t be Qt (milli-equivalents). The diffusion process taking place in the Nernst layer is essentially a cation exchange at constant ionic strength. The rate is therefore proportional to the gradient of concentration of A+ (or B+) ; let D L be the Fick's law diffusion coefficient for the inter-diffusion of A+ and B+ under these conditions. After time t the concentration of A+ on the outer side of the Nernst layer is Qt/V. The concentration of A+ at the surface of the particle is more difficult to estimate. Boyd et aZ.2 took this as the concentration which would be found in a solution in equilibrium with the prevailing resin composition. Under their conditions this was simply proportional to the [A+] in the resin but under the present conditions it would be necessary to introduce the mass action equilibrium constant (see Part I),l and the resulting kinetics would be mathematically very complicated.However it will be shown that the cations in the exchanger are highly mobile ; there seems therefore to be no reason to distinguish between " free " and " bound'' cations in the resin phase-at least for simple inorganic cations. Further the number of cations present at the outer surface of the particles as gegenions to the -SO,- groups far exceeds those present as soluble electrolyte from the solution phase since soluble anions are largely repelled by the opposing zeta potential.* Consequently the number of A+ ions which are free to participate in diffusion away from the surface of the sphere at any instant is simply proportional to the number present in the resin let it be put equal to k'(Qo - Qt).See for example Verweg and Overbeek Theory of the StabWy of Lyophobic CoZZoids FIG. 8.-Influence of temperature NMe,+. I . Go-6' C 4. 12.6~ C 5 . I ' 2 O 2. 41.3" 3. 25.0' (Elsevier 1948) p. 31. T. R. E. KRESSMAN AND J. A. KITCHENER 99 Application of Fick's law to the Nernst layer (the diffusion gradient being assumed linear) gives Integration leads finally to the equation where Q is the amount of A+ which has passed into the solution when equilibrium has been reached and k is a constant equal to Dk'/6. This equation is of the same form as that obtained by Boyd et aL2 (for an infinite bath) which was found to fit the kinetics of their Na+-K+ exchange in dilute solutions at least up to Qt/Qw = 0.4.In the present work the equation fits the kinetics well in appropriate cases (see Fig. G a and 7 b). For example the NH,+-Na+ exchange at 25" C fits accurately from Qt/Qm = o up to at least QJQ = 0.9. Other cases which conform to these kinetics include Ag+ K+ Li+ Mg++ Ba++ Al+++. the solution effectively uniform up t o the particle surface and the rate controlled by radid diffusion inwards with a constant diffusion coefficient the kinetics of the P-mechanism should be formally similar to those for conduction of heat into a sphere from a well-stirred bath. Boyd et aL2 have successfully applied the well-known theory for conduction of heat into a sphere from an infinite bath.The corresponding theory for a limited bath is given by Carslaw and Jaeger.5 The form of their solution however is less convenient than that of an alternative solution recently given by Paterson,G viz. FIG. 9.-Influence of temperature PhNMe,CH,Ph+. 3. 50.0' C 1. 25.0' C 2. 36-5'C (ii) PARTICLE DIFFUSION (P-MECHANISM) .-Assuming 5 Carslaw and Jaeger Conduction of Heat in Solids (Oxford Univ. Press 1g47) pp. 83- 201. 6 Paterson Proc. Physic. Soc. 1947 59 50. KINETICS OF CATION EXCHANGE %t Y 2 ) roots of thc equation I . Na+ (P). 2. NMe,+. I00 where 7 I= - Y being the thermal diffusivity ; and Q and p are the Since x2 + 370'x - 3w = 0 arid ID is (heat capacity of the sphere)/(heat capacity of the bath).This solution is valid up to T = 0.1 which in practice covers most of the process (e.g. up to Qt/Qco = 0.84 when w = I). In attempting to apply Paterson's solution to the ion-exchange process with e.g. NEt,+ and NH,-resin x is identified with the cation-exchange diffusion coefficient inside the resin (Dp) and Qt/Qm with the ratio (NH,+ out in time t)/(NH4+ out at equilibrium) and ze with the ratio (NEt,+ in FIG. 10.-Activation energy data. 4. Na-" (F) (-0.5 added t o vertical scale). 5. PhNMe,.CH,Ph+. 3. Diffusion coefficient data for NaCl in water.' resin at equil.)/(NEt,+ in solution at equil.). From the value of w cc and p can be calculated and a graph of Qt/Qa as a function of z can be computed. T = (Dp/r2)t the best value of (Dp/r2) can be found t o fit the theoretical curve to the experimental points (provided of course the theory gives the correct form of the kinetics) ; hence r being determined independently a value of D p can be obtained.The attempt to fit the equation in this way to the data in Fig. 11 did not lead t o satisfactory agreement when experiment and theory were fitted at Ql/Qm = 0.5 the deviations were always in the sense that the equation predicted a more rapid reaction initially than in fact occurred. Barrer 7 Int. Crit. Tables Vol. 5 . Barrer Trans. Faraday SOC. 1949 45 358. T. R. E. KRESSMAN AND J. A. KITCHENER TABLE I1 NRle,+ 2' O ' j 0'3 0'1 I01 lias recently discussed possible causes for deviatioiis in the somewhat analogous system of gas sorption by zeolites.Lack of conformity in the present system could be due to (I) slow swelling of the initially dry resin and (2) change of diffusion coefficient with resin composition. (Curiously enough good agreement was obtained when w was taken as (NH,+ in resin at equil.)/(NH,+ in solution at equil.) ; the lines in Fig. 11 were obtained in this way. The significance if any of this is not clear.) Barrer 8 has shown that Paterson's solution like the well-known one for diffusion into a sphere from an infinite bath approximates to a &relation for small values of Qt/Qm the limiting expression being However examination of such graphs shows that whereas the line €or aii infinite bath is within z yo of linear up to about Qt/Qm = 0.5 the linear range becomes progressively shorter as the bath becomes more limited.Thus at w = I the L3 graph is linear only within 4 % up to QtlQ = 0.25. Consequently in testing for P-mechanism by this method of plotting the data a straight line is to be expected only at small values of Qt'QL. Fig. 6 0 and 7 a show examples of exchange reactions which conform approxi- inately to these expectations. Values nIEFUSION COEFFICIENT yT 230 c obtained coefficients for the nominal from these diffusion graphs arc given Cation _ _ _ ~ - ~ - - ' Dp(cm +c.-l __- x - 108) in Table 11. The ta graph is used as the first test for P-mechanism. It is fortunate PhNMe2Et+ that the F- and P-kinetics are PhNMe,CH,Ph+ o*oor sufficiently different in form to be distinguished readily by conformity to either log (I - Qt/Qm) against t or Qt/Qm against 0 graphs respectively.Fig. G (curves j and 7) shows how data which fit onc are clearly excluded from fitting the other. ~ ~ 3 ~ - A m y l 9 Carslaw and Jaeger ref. 5 p. 288. I 1 (iii) INTERMEDIATE (1-MEcHANIsM) .-The full kinetics of the 1-process transitional between F- and P- have not yet been worked out for a limited bath. The analogous problem of heat conduction into a composite sphere has been solved for the case of a constant surface t e m p e r a t ~ r e . ~ Howevcr the algebra is already very complicated and with a finite bath would be much more so. Crank and Godson l o have obtained approximate numerical bolutions for certain cases of composite infinite cylinders in a limited bath using the method of finite differences The same method could be used for the present problem but it would be necessary to have more explicit know- ledge of the constant k' in the F-kinetics theory than is at present available.The only example of what appears to be I-mechanism encountered in the present work is the Na+-NH,+ exchange within the range 15"-z5" C the kinetics at 2 5 O 35" and 46" fitting F-mechanism and those at IS- 8' and zo fitting P-mechanism. (c) Interruption tests l1 provide a simple criterion for the existence of a large concentration gradient inside the particles thus differentiating P- or I-mechanism froin F. In the present work such tests have in every case where applied confirmed the niechanisms already suggested from the l o Crank and Godson Phil.M a g 1947 38 794. l1 Kunin and Myers J . Plzj~sic. Ckewi. 1947 51 I I I I . 102 KINETICS OF CATION EXCHANGE form of the kinetics. Thus the Na+-NH,+ exchange at 25" gives F-kinetics and shows no discontinuity on interruption (Fig. 7 curve 4) whereas Na"-NH,+ at zo and NMe,+ and NEt,+-NH,+ at 2 5 O givc P-kinetics and show a large discontinuity in the expected direction. Like- wise Th+ $- + +-NH,+ gives P-kinetics and shows a similar discontinuity on interruption (see Fig. 5 curves 8 and 12). ( d ) Temperature coefficient might be expected to afford a criterion of mechanism since the temperature coefficient for diffusion through the solid might reasonably be supposed to be distinctly higher than for free diffusion in solution.This is certainly the case with large molecules such as dyes pmetrating into fibres where the activation energy is aboiit FIG. I I .- -Systems showing particle diflusion. I. Ns+ 3' C 2. 5 . PhNMe2Etf 25" C 6. NEt,-resin suspension 25' C 7 . PhXMe,CH,Ph+ 2.j' C 25" C 3 . NEt,+ 2 5 ' C . 0 bromide 13 Chlorazol Sky Blue salt 4 . T\TMe,.n-nmyl+ 2 5 O C 10-30 kcal./rnole,l2 whereas that for diffusion of most salts in water is about 5 k~al./mole.~ Similarly Boyd et aZ.2 found 4 kcal./mole for Na+ (F-mechanism) and 8 kcal./mole for P-mechanism with Amberlite IR-I. Table I shows values of the energy of activation €or several exchange reactions studied in the present work (see also Fig.10). The exchange between NH,-resin and Na+ at 25'46" (F-) shows the expected value of about 5 but it is remarkable that both Na+ at 2O-15' (P-) and NMe,-i- (P-) have the same low value. A high value of 8-2 kcal./mole is found for PhNMe,CH,Ph+. This is the largest ion yet studied which is capablc of l2 See eg. Speakman and Smith J . SOC. Djters Col. 193G 52 131. Garvie Griftillis and Neale Tmits. Farnday SOC. 1934 30 271. T. R. E. KRESSMAN AND J. A. KITCHENER 10.3 penetrating and reaching all the exchanging sites in this resin and it seems likely that the higher activation energy is a consequence of the steric difficulty which it experiences in moving through the molecular pores of the resin. On the other hand the smaller ions showing P-mechanism evidently do not experience any such " wall-effect " and move simply through the water which fills the pores.It is not invariably true that diffusion in solids requires large energies of activation and results in agreement with the present picture have been obtained by other workers. The high value with fibres may reflect distortion of a flexible macromolecular structure by the dye molecule. With rigid porous solids the activation energy for the diffusion of molecules smaller tha% the $ores may be almost normal for an aqueous medium. For example Tiselius l 3 found 5-4 kcal./mole for the diffusion of water normal to the (201) face in the zeolite heulandite. On the other hand diffusion normal to the (001) face required 9.1 kcal/mole ; presumably the interstitial holes in this direction are much smaller and of about the same size as the water molecule.Diffusion of salts in z yo agar gels shows the same temperature coefficient as for free diffusion in aqueous s~lution.~ Role of the Anion.-To preserve electro-neutrality the ions leaving the resin must be replaced by others of the same total charge. If the ions were bound to specific sites the situation might arise where the rate of dissociation from sites was the factor limiting the rate of exchange and it might then be necessary to provide extra (soluble) anions by diffusion of salt from the ambient solution into the interstices of the resin before rapid cation exchange could take place This possibility was studied by the experiments (recorded above) in which the anion was provided by the dye Chlorazol Sky Blue FFS which is found to diffuse extremely slowly and to an extremely small extent into the resin.With both Na+ and NEt,+ the rate of exchange was the same as with the simple halide salts (see Fig. 5 curve 2 ; Fig. 7 curve 4 ; Fig. 11 curve 3) indicating that it is not necessary to have soluble anions within the pores for exchange to occur rapidly. The high mobility of the cations alone is proved conclusively by the experiment with finely divided NEt,-resin suspension in place of NEt,Br solution. This suspension exchanges cations with normal NH,-resin grains almost as rapidly as does the homogeneous salt solution the apparent rate constants being in the ratio of 0.7411 (cf.Fig. 5 curves 7 and IO) proving that soluble anions play no significant part at all in the kinetics of cation exchange at least with the resinous exchanger studied here. In this experiment the process starts with the interpenetration of the Gouy diffuse double layers of the large grains with those of the small suspended particles. At the very low ionic strengths prevailing the double layers spread so far from the microscopic particles that the cation distribu- tion in the liquid is similar to that in a true salt solution. Consequently the outside part of the double layer from the large granules can mix rapidly with the cations of the suspension leaving diffusion of cations within the pores of the resin as the rate-controlling process as it is with the ordinary salt solution.A similar process no doubt occurs in normal exchanges with soluble salts-the soluble cations exchange first with the resin cations in the outer diffuse double layer thus setting up a concentration gradient within the pores which is eliminated by ordinary diffusion this latter process requiring no soluble anions. l3 Tiselius Z. physik. Chew. A 1934 169 425 ; 1935 174 4 0 1 . ION EXCHANGE STUDIES 104 We are indebted to Prof. H. V. A. Briscoe for his encouragement and interest in the work and to Dr. R. F. Hudson for helpful discussion. One of us (T.R.E.K.) thanks The Permutit Co. Ltd. Chiswick W.4 for a grant in support of the work and for permission to publish it. Imperial CoLlege of Science G Technology South Kensington London S.W.7. CATION EXCHANGE WITH A SYNTHETIC PHENOLSULPHONATE RESIN Part V. Kinetics BY T. R. E. KRESSMAN AND J. A. KITCHENER Received 11th Jztly 1949 -1 study has been made of the kinetics of exchange between solutions of various simple inorganic and substituted quaternary ammonium salts and the ammonium form of the sulphonated phenol-formaldehyde resin for which equilibrium measure-ments have been reported elsewhere. Two mechanisms are observed where the rate is controlled by diffusion in the particles of the exchanger (P-mechanism) and in the bounding Kernst film (F-mechanism) respectively. These are distinguished by the form of the kinetics by interruption tests and by the influence of stirring. The factors that decide which mechanism applies in a given system are discussed.Measurements of the influence of temperature suggest that so long as the cation is small compared with the pores of the resin (e.g. Na+ and KMe,+ with the present resin) the energy of activation for diffusion is ca. 5 kcal./mole as for free diffusion in water. Larger ions which have difficulty in penetrating show a higher value e.g. 8 kcal./mole with PhNMe ,CH ,Ph +. Exchange occurs readily between a granular resin and a finely divided resin in wspension proving that soluble anions are not needed to release cations from the resin. The high mobility of the cations is consistent with a diffuse donble-layer model T. R. E. KRESSMAN AND J. A. KITCHENER 91 The potentialities of ion exchange separation techniques can be fully exploited only if consideration is given to both equilibrium and kinetic aspects.These are largely independent and they differ with different ion exchangers even though the exchange groups may be the same. There is no simple relationship between affinities and rates of exchange for a series of ions although with the phenolsulphonate resin used in the present study the rate generally decreases as the affinity increases. The possibilities of separations based on differences of rates of exchange have not yet been fully explored. Fig. I shows an example where such an effect might be employed the concentrations of H+ NH4+ and NEt4+ in a solution (initially equimolar with respect to NH,+ and NEt,+) are shown during the course. of an exchange experiment with a H-resin (contain-ing -SO,- groups) in a stirred system.It is seen that the equilibrium affinity of NEt4+ for the resin is greater than that of -NH4+ but the rate of exchange is greater for the NH4+ ion. If the experiment were stopped after 2 or 3 min. the solution would be relatively enriched in NEt,+ but at equilibrium the reverse would apply. FIG. I. Both equilibria and kinetics for a series of ions depend very greatly on the structure of the solid exchanger. The equilibrium relationships for a phenoIsulphonate resin similar to Zeo-Karb 215 have already been described elsewhere in Parts I-IV of this series of papers.1 The present paper describes studies of the kinetics of exchange with the same resin. Experiments and Results Materials.-The sample of resin was taken from the same batch of H-resin as that used in earlier studies.1 The general technique of converting the material to the salt form and of determining the equivalent weights air drying etc.is the same as described there. With the exception of the experiments illustrated in Fig. I where the H-form of the resin was used the NH,-form was used through-out the work and after preparing it from the H-resin i t was air-dried and sieved - 10 + 22 mesh. Two samples were sieved out and all the results except those illustrated in Fig. 8 were obtained with one of these samples. Those in Fig. 8 were obtained with the second sample and although the resin granules were also - 10 + 22 mesh a larger fraction of the larger mesh size was present and the velocity of exchange is thus somewhat lower with this sample than with the first.1 Kressman and Kitchener J . Cltcm. Soc. 1949 1190 92 KINETICS OF CATION EXCHANGE In the series of experiments illustrated in Fig. I the H-form of the resin \\as used and because of its slight instability,f it was prepared by thoroughly washing the free acid from a sample taken from the main batch air-drying rapidly and sieving - 10 +22 mesh ; i t was then used immediately. In this way the experi-ments were completed before the H-resin had begun to hydrolyze. Determination of Effective Sphere Radius of the Swollen Resin Granules .-A large number of particles-of the order of 500-were counted, allowed to swell in water the water removed and the resin granules mopped L)etween filter papers until the surface moisture had been removed.The swollen granules were then weighed. The density of the swollen granules \%as determined with the aid of a density bottle in the usual way. The effective sphere radius Y of the swollen granules is calculated from where p is the density of the swollen resin N the number of granules arid IT’ the weight of the N granules. 413 . 7 ~ ~ p N = W , C C m. SecLbn uf x x FIG. 2. The first sample of KH,-resin which was used for the majority of the work (see under Materials) was found in this way to have an effective sphere radius of 0.446 mm. ; and the second sample used for the experiments illustrated in Fig. 8 an effective sphere radius of 0.517 mm. The density of the swollen granules was 1-33 g./cm.3 Determination of Exchange Velocity.-The limited bath method was used throughout the work and not the column technique with infinite bath as used hy Boyd et The essential requirements in such heterogeneous studies of standardized vigorous agitation and the necessity of starting and stopping the reaction sharply were achieved by using a centrifugal type of stirrer as shown i n Fig.2 a. The resin was placed in the wire-gauze cage and as the stirrer rotated in the solution a very rapid stream of the solution passed over the resin which was forced into a roughly cylindrical “ wall ” on the insitlc of the cage (see Fig. 2 b). The constancy of speed of rotation of the stirrer x i as followed stroboscopically. H y using several stroboscope discs constant speeds of rotation from 300 rev. jmin. a t intervals to 1200 rev./min.were attained. The reaction was started by lowering the already rotating stirrer with the resin in the cage into the solution ; and i t was stopped after the appropriate interval of time by raising it. still rotat-Boyd. Adamson and RIj-ers J . A v i i e ~ . C‘kem. SOC. 1947 69 2 8 ~ 1 T. R. E. KRESSMAN AND J. A. KITCHENER 9.; ing out of the solution. An aliquot portion of the solution was then analyzed for NH,+ or H7- as appropriate. Temperatures between 14O and 45" C were thermostatically controlled to within -+ 0.2" ; outside this range the limits were The weight of resin taken in every run was such that it contained 2-5 m.equiv. o f exchangeable cation and 2.5 m.equiv. of the other cation were also present in the aqueous phase in the form of 125 ml.0.02 N solution. Fig. 3 shows that the mixing is a t a maximum when the speed of rotation is between 1000 and 1100 rev./min. ; higher speeds caused entrainment of air and accordingly an apparent decrease in velocity. The speed of rotation was, therefore standardized a t 1000 rev./min. throughout the work. Influence of Agitation.-Fig. 3 shows how the time for half-change depends upon the speed of rotation of the stirrer for two systems NH4R + NaCl and NH,R NEt,Br respectively. The time for half-change in the first system -L - 0.50. FIG. 3.-Effect of stirring on the time for half-change. r . Centrifugal stirrer Na+. 1. Centrifugal stirrer NEt,+. 3. Glass stirrer NEt4+. decreases as the speed of rotation is increased approaching a minimum at 1000 rev./min.when the stationary liquid film surrounding the particles has reached a minimum. (Above about 1100 rev./min. entrainment of air occurs and erratic results are consequently obtained.) No greater velocity was obtained even when the solution was circulated over the resin (in the form of a shallow bed) extremely rapidly by means of a pump. The time for half-change in the NEt ' system is also practically at a minimum a t 1000 re\jniin. but unlike the Na+ system it remains at a minimum at speeds down to 750 rev./min. and only then when very little relative motion of solid and liquid Is occurring does it begin to increase. Fig. 3 also shows a curve obtained in the NEt4+ system with a simple bent qlass-rod stirrer (Fig. 2 6 ) . This stirrer is seen to be more efficient a t slower ipeeds of rotation but at higher speeds the centrifugal stirrer results in a thinne 94 KINETICS OF CATION EXCHA4NGE liquid film than does the simple stirrer.Evidently so long as the glass stirrer keeps the particles in suspension increased agitation has an immeasurable effect upon the thickness of the liquid film. Preparation of NEt4-resin in Fine Suspension.-A series of rate experi-ments was carried out in which no soluble anions whatever were present in the system the exchange occurring between the ordinary NH,-resin and a fine aqueous suspension of NEt,-resin. This suspension was prepared by milling well-washed NEt,-resin with water. A sample of H-resin was washed free of acid and converted to the NEt,-form with a solution of NEt,Br. The rate of reaction is considerably less than with ammonium chloride and a lower rate of flow was accordingly used and a greatei-volume of solution was found necessary to remove the whole of the exchangeable hydrogen ions from the resin.The resin-about 40 g.-was thoroughly washed to remove all traces of electrolytes and then ground in a laboratory porcelaiil ball mill for about 24 hr. with about I 1. water. The milky suspension so obtained was allowed to settle for about I hr. and the liquid decanted. This consisted of a suspension of the finest particles the largest of which were of the order of I [A diam. It was adjusted by dilution to contain 0.02 equiv. of exchangeable SEt,+ per 1. of suspension as indicated by a Kjeldahl nitrogen determination. Care was taken when using this sus-pension to ensure that homogeneous samples were taken from it.It was 3 3 0 gNQ% vigorously kept gently shaken agitated just during prior to the and re-moval of the sample. The NI-I,-resin was washed with water before each run was started to remove any traces OH NH2 of electrolytes which might be con-:"3O "="Q3;3- taminating it. The velocity determination with this suspension was carried out exactly as with the aqueous salt FIG. 4.-Anion of Chlorazol Sky Blue FFS. solutions. The results are shown in Fig. 5 curve 10 and Fig. I I curve 6 and graphs of Qt/Q against t6 for the NEt,-resin suspension and for NEt,Br indicate that the apparent rate constants are in the ratio of 0*74/1. Preparation of the Salts of Chlorazol Sky Blue FFS.-The sodium salt was prepared from the crude dye by dissolving it in water and salting-out with sodium acetate.This was repeated three times. Adhering sodium acetate was finally removed by washing repeatedly with alcohol. The purified product was dried a t IOOO C. The NEt,-salt was prepared in solution from NEt40H and the pure dye acid. The NEt,OH was prepared in known concentration (about 0.2 N) by adding excess moist silver oxide to a solution of NEt,Br and filtering. The dye acid was prepared from the pure sodium salt by passing quantitatively a dilute solution (about 0.01 N) containing 4-96 g. of the salt through a IOO ml. column of Zeo-Karb 215 (The Permutit Co. Ltd.) containing exchangeable hydrogen ions. The solution was then evaporated on the water bath until its concentration was of the order of 0.03 N cooled and " titrated '' with the NEt,OH to pH 7 using a pH meter.The quantity of NEt,OH required was exactly that calculated from its known concentration. The solution of the NEt,-dye salt was then diluted to I 1. In the runs with the dye salts as in those with the NEt,-resin suspension, the resin was washed with water before each run was started to remove traces of electrolytes. Velocity with Different Cations .-Since the equilibrium positions are very different with the various cations and since only the kinetics are at present being considered the extent of exchange at time t is expressed as Qt/O!, i.e., the fraction of the amount of exchange occurring a t time t to that occurring at equilibrium. Fig. 5 shows the graphs of Qt/Q against t for a number of different OH NH T.R. E. KRESSMAN 4ND J. A. KITCHENER 9s cations at 25" C. The same results are plotted in Fig. 6 as - log, ( I - Qt/Qc?,) against t and Qt/Qm against t4 for analysis of mechanism-see below. Influence of Temperature .-The activation energies of the exchange reactions were determined from velocity measurements at several temperatures with the NH,-resin and respectively Na+ NMe,+ and PhNMe2CH,PhA. FIG. .=j.-Rate of exchange between NH,-resin and various cations a t 2 5 O C. I . 2. 3-4. 5. 6 . 7. 8. 9. 10. I I . 12. 13. The results K+ (chloride). Na+ 0 chloride x interrupted A Chlorazol Sky Blue salt. Li+ (chloride). Mg++ and Ba++ 0 chloride A MgSO,. NMe + (bromide).A1 3- + + (chloride). NEt,+ (bromide). NEt,' (bromide) interrupted at 5 min. for 30 min. NMe ,.n-Amyl+ (bromide). PhNMe,Et+ (bromide) and NEt,-resin suspension. Th++++ (nitrate). Th++++ (nitrate) interrupted at 2.5 min. for 20 min. PhNMe,CH ,Ph+ (chloride). - are shown in Fig. 7 8 and 9 where they are plotted in terms of -1og,,(1 - Qt/Qa.) against t and QJQ against 14 as appropriate. The logarithms of the limiting slopes of the lines so obtained are plotted against the reciprocals of the absolute temperature in Fig. 10. The values for the activation energies given in Table I are obtained from the slopes of these straight lines. TABLE I range kcal./mole 2 O- I 4- 8 Cation 5.1 Na+ (P) NMe + 1'-61" 5.1 Na+ (F) 2 5 "-45 * 7 YhNMe2CH2Phi 8. 96 KINETICS OF CATION EXCHANGE Discussion The structure of the phenolsulphonate exchange resin is a three-dimensional rigid network resembling a sponge and containing five -SO groups to every seven phenol residues (see Part I).l The cations are highly mobile inside the water-filled interstices and form a Gouy diffuse double layer round each resin granule.Consequently cation exchange does not involve any " chemical " step, since no covalent bond has to be broken and the rate of exchange depends simply on one or more of a series of consecutive transport steps-namely, forced convect ion of the fluid diffusion through the unmixed boundary (1%) FIG. 6.-Tests of mechanism. I . K+. 2. I i f . 3 . Ag+. 4. Mg++ and Ha". j. N b t + . 6. Al+++. 7-8 . 9.10. I I . 13. 13. Na + (2 5 ") . NMe,+. NEt,+. NMe,.n-Amyl+. PhNRiIe,Et+ and Th++++. PhNMe ,CH ,Pi> + . NEt,-resin suspension. layer and diffusion inside the particles. As Boyd et aL2 have shown three possible types of kinetics may therefore be encountered in a practical, dynamic exchange process : (i) Diffusion in the boundary liquid film is rate-determining (ii) Diffusion in the solid particle is rate-determining (P-mechanism). (iii) In an intermediate region the rate is influenced by diffusional The factors that may decide which mechanism applies in a given system are (I) particle size (2) degree of agitation of the solution (3) diffusion coefficients of the ions in the solution and inside the resin particles (4) temperature, (5) the equilibrium distribution coefficient and (6) solution concentra-tion.Variation of any one of these factors may produce a change of mechanism. For example Boyd et aL2 have shown that the Na+-K+ exchange (under their conditions of experiment) is F at 0.001 M ionic strength I at 0.01 M and P at 0-1 M. (F-mechanism) . resistances in both phases (I-mechanism) T. R. E. KRESSMAN AND J. A. KITCHENER 97 Criteria for Distinguishing Mechanisms. (a) Influence of Stirring.-The rate increases with the degree of mixing of the liquid as long as the F- or I-mechanisms are rate-controlling whereas with P-mechanism it should be independent. With increasing speed of rotation of a given stirrer a maximum exchange rate may be reached, beyond which the rate may remain constant or even decrease (Fig.3). Only exceptionally does such a maximum indicate the point where 1-mechanism gives place to P. More usually the rate reaches a limit for hydrodynamical reasons the stirrer having reached its limiting efficiency for mixing and it may still be possible to achieve faster rates of exchange by using a different type of stirrer (Fig. 3). However in border-line cases a change of mechanism may sometimes be detected by the influence of stirring-notably when a maximum exchange rate is reached in the stirring range slower than the known hydrodynamical limit. The exchange with NEt4+ in Fig. 3 provides an instance of this. ( a ) (b) FIG. 7.-Influence of temperature W a f . I . 2'03 c 4. 25'0°C 0 Chloride 2. 8.4' 5. 31'8" i' interrupted j. 14.8" 6.45'7" A Chlorazol Sky Blue salt (b) Form of the Kinetics.-@) FILM DIFFUSION (F-MECHANISM) .-The well-known Nernst static diffusion film theory is clearly only a crude approxi-mation for the complex situation which exists near an irregular solid surface in a stirred liquid; in particular the "thickness" of such a layer is a mathematical fiction which has proved of little use since this quantity must usually be deduced from the kinetics (not vice versa) and its dependence on stirring temperature viscosity etc. is problematical. Only in a few idealized cases has it been possible to calculate the diffusional transport up to a body in a stirred l i q ~ i d . ~ However the Nernst layer approximation is useful for treating the kinetics of exchange under fixed conditions of stirring and temperature.2 In the present system let Qo be the number of milli-equivalents of the pure A+ form of the resin which at time t = o is brought into contact with V ml.solution containing Qo milli-equivalents of a salt of the cation B+. Suppose the resin consists of n particles of mean equivalent sphere radius Y , 3Levich Acta Physicochiw 1942 17 257; 1944 19 117 133 Faraday SOG. Discussiom 1947 I 37. 98 KINETICS OF CATION EXCHANGE and that the effective Nernst diffusion layer thickness is 8. Let the amount of exchange which has occurred after time t be Qt (milli-equivalents). The diffusion process taking place in the Nernst layer is essentially a cation exchange at constant ionic strength. The rate is therefore proportional to the gradient of concentration of A+ (or B+) ; let D L be the Fick's law diffusion coefficient for the inter-diffusion of A+ and B+ under these conditions.After time t the concentration of A+ on the outer side of the Nernst layer is Qt/V. The concentration of A+ at the surface of the particle is more difficult to estimate. Boyd et aZ.2 took this as the concentration which would be found in a solution in equilibrium with the prevailing resin composition. Under their conditions this was simply proportional to the [A+] in the resin, but under the present conditions it would be necessary to introduce the mass action equilibrium constant (see Part I),l and the resulting kinetics would be mathematically very complicated. FIG. 8.-Influence of temperature NMe,+. I . Go-6' C 3.25.0' 4. 12.6~ C 2. 41.3" 5 . I ' 2 O However it will be shown that the cations in the exchanger are highly mobile ; there seems therefore to be no reason to distinguish between " free " and " bound'' cations in the resin phase-at least for simple inorganic cations. Further the number of cations present at the outer surface of the particles as gegenions to the -SO,- groups far exceeds those present as soluble electrolyte from the solution phase since soluble anions are largely repelled by the opposing zeta potential.* Consequently the number of A+ ions which are free to participate in diffusion away from the surface of the sphere at any instant is simply proportional to the number present in the resin let it be put equal to k'(Qo - Qt). See for example Verweg and Overbeek Theory of the StabWy of Lyophobic CoZZoids, (Elsevier 1948) p.31 T. R. E. KRESSMAN AND J. A. KITCHENER 99 Application of Fick's law to the Nernst layer (the diffusion gradient being assumed linear) gives Integration leads finally to the equation where Q is the amount of A+ which has passed into the solution when equilibrium has been reached and k is a constant equal to Dk'/6. This equation is of the same form as that obtained by Boyd et aL2 (for an infinite bath) which was found to fit the kinetics of their Na+-K+ exchange in dilute solutions at least up to Qt/Qw = 0.4. In the present work the equation fits the kinetics well in appropriate cases (see Fig. G a and 7 b). For example the NH,+-Na+ exchange at 25" C fits accurately from Qt/Qm = o up to at least QJQ = 0.9.Other cases which conform to these kinetics include Ag+ K+ Li+ Mg++ Ba++ Al+++. FIG. 9.-Influence of temperature PhNMe,CH,Ph+. 1. 25.0' C 2. 36-5'C 3. 50.0' C (ii) PARTICLE DIFFUSION (P-MECHANISM) .-Assuming the solution effectively uniform up t o the particle surface and the rate controlled by radid diffusion inwards with a constant diffusion coefficient the kinetics of the P-mechanism should be formally similar to those for conduction of heat into a sphere from a well-stirred bath. Boyd et aL2 have successfully applied the well-known theory for conduction of heat into a sphere from an infinite bath. The corresponding theory for a limited bath is given by Carslaw and Jaeger.5 The form of their solution however is less convenient than that of an alternative solution recently given by Paterson,G viz., 5 Carslaw and Jaeger Conduction of Heat in Solids (Oxford Univ.Press 1g47) pp. 83-6 Paterson Proc. Physic. Soc. 1947 59 50. 201 I00 KINETICS OF CATION EXCHANGE %t where 7 I= - Y being the thermal diffusivity ; and Q and p are the Y 2 ) roots of thc equation arid ID is (heat capacity of the sphere)/(heat capacity of the bath). This solution is valid up to T = 0.1 which in practice covers most of the process (e.g. up to Qt/Qco = 0.84 when w = I). In attempting to apply Paterson's solution to the ion-exchange process with e.g. NEt,+ and NH,-resin x is identified with the cation-exchange diffusion coefficient inside the resin (Dp) and Qt/Qm with the ratio (NH,+ out in time t)/(NH4+ out at equilibrium) and ze with the ratio (NEt,+ in x2 + 370'x - 3w = 0 FIG.10.-Activation energy data. I . Na+ (P). 2. NMe,+. scale). 3. Diffusion coefficient data for NaCl 4. Na-" (F) (-0.5 added t o vertical 5. PhNMe,.CH,Ph+. in water.' resin at equil.)/(NEt,+ in solution at equil.). From the value of w cc and p can be calculated and a graph of Qt/Qa as a function of z can be computed. Since the best value of (Dp/r2) can be found t o fit the theoretical curve to the experimental points (provided of course the theory gives the correct form of the kinetics) ; hence r being determined independently a value of D p can be obtained. The attempt to fit the equation in this way to the data in Fig. 11 did not lead t o satisfactory agreement when experiment and theory were fitted at Ql/Qm = 0.5 the deviations were always in the sense that the equation predicted a more rapid reaction initially than in fact occurred.T = (Dp/r2)t, Barrer 7 Int. Crit. Tables Vol. 5 . Barrer Trans. Faraday SOC. 1949 45 358 T. R. E. KRESSMAN AND J. A. KITCHENER I01 lias recently discussed possible causes for deviatioiis in the somewhat analogous system of gas sorption by zeolites. Lack of conformity in the present system could be due to (I) slow swelling of the initially dry resin, and (2) change of diffusion coefficient with resin composition. (Curiously enough good agreement was obtained when w was taken as (NH,+ in resin at equil.)/(NH,+ in solution at equil.) ; the lines in Fig. 11 were obtained in this way. Barrer 8 has shown that Paterson's solution like the well-known one for diffusion into a sphere from an infinite bath approximates to a &relation for small values of Qt/Qm the limiting expression being The significance if any of this is not clear.) However examination of such graphs shows that whereas the line €or aii infinite bath is within z yo of linear up to about Qt/Qm = 0.5 the linear range becomes progressively shorter as the bath becomes more limited.Thus at w = I the L3 graph is linear only within 4 % up to QtlQ = 0.25. Consequently in testing for P-mechanism by this method of plotting the data a straight line is to be expected only at small values of Qt'QL. Fig. 6 0 and 7 a show examples of exchange inately to these expectations. Values for the nominal diffusion coefficients reactions which conform approxi- TABLE I1 nIEFUSION COEFFICIENT yT 230 c _ _ _ ~ - ~ - - __- -obtained from these graphs arc given Cation ' Dp(cm +c.-l x 108) 2' O ' j in Table 11.NRle,+ I that the F- and P-kinetics are PhNMe,CH,Ph+ o*oor, The ta graph is used as the first test ~ ~ 3 ~ - A m y l 0'3 for P-mechanism. It is fortunate PhNMe2Et+ 1 0'1 sufficiently different in form to be distinguished readily by conformity to either log (I - Qt/Qm) against t or Qt/Qm against 0 graphs respectively. Fig. G (curves j and 7) shows how data which fit onc are clearly excluded from fitting the other. (iii) INTERMEDIATE (1-MEcHANIsM) .-The full kinetics of the 1-process, transitional between F- and P- have not yet been worked out for a limited bath.The analogous problem of heat conduction into a composite sphere has been solved for the case of a constant surface t e m p e r a t ~ r e . ~ Howevcr, the algebra is already very complicated and with a finite bath would be much more so. Crank and Godson l o have obtained approximate numerical bolutions for certain cases of composite infinite cylinders in a limited bath using the method of finite differences The same method could be used for the present problem but it would be necessary to have more explicit know-ledge of the constant k' in the F-kinetics theory than is at present available. The only example of what appears to be I-mechanism encountered in the present work is the Na+-NH,+ exchange within the range 15"-z5" C the kinetics at 2 5 O 35" and 46" fitting F-mechanism and those at IS- 8' and zo fitting P-mechanism.(c) Interruption tests l1 provide a simple criterion for the existence of a large concentration gradient inside the particles thus differentiating P- or I-mechanism froin F. In the present work such tests have in every case where applied confirmed the niechanisms already suggested from the 9 Carslaw and Jaeger ref. 5 p. 288. l o Crank and Godson Phil. M a g 1947 38 794. l1 Kunin and Myers J . Plzj~sic. Ckewi. 1947 51 I I I I 102 KINETICS OF CATION EXCHANGE form of the kinetics. Thus the Na+-NH,+ exchange at 25" gives F-kinetics and shows no discontinuity on interruption (Fig. 7 curve 4) whereas Na"-NH,+ at zo and NMe,+ and NEt,+-NH,+ at 2 5 O givc P-kinetics and show a large discontinuity in the expected direction.Like-wise Th+ $- + +-NH,+ gives P-kinetics and shows a similar discontinuity on interruption (see Fig. 5 curves 8 and 12). ( d ) Temperature coefficient might be expected to afford a criterion of mechanism since the temperature coefficient for diffusion through the solid might reasonably be supposed to be distinctly higher than for free diffusion in solution. This is certainly the case with large molecules such as dyes pmetrating into fibres where the activation energy is aboiit FIG. I I .- -Systems showing particle diflusion. I. Ns+ 3' C 2. 25" C 6. NEt,-resin suspension 25' C 3 . NEt,+ 2 5 ' C . 0 bromide 7 . PhXMe,CH,Ph+ 2.j' C 4 . T\TMe,.n-nmyl+ 2 5 O C 5 . PhNMe2Etf 25" C 13 Chlorazol Sky Blue salt 10-30 kcal./rnole,l2 whereas that for diffusion of most salts in water is about 5 k~al./mole.~ Similarly Boyd et aZ.2 found 4 kcal./mole for Na+ (F-mechanism) and 8 kcal./mole for P-mechanism with Amberlite IR-I.Table I shows values of the energy of activation €or several exchange reactions studied in the present work (see also Fig. 10). The exchange between NH,-resin and Na+ at 25'46" (F-) shows the expected value of about 5 but it is remarkable that both Na+ at 2O-15' (P-) and NMe,-i- (P-) have the same low value. A high value of 8-2 kcal./mole is found for PhNMe,CH,Ph+. This is the largest ion yet studied which is capablc of l2 See eg. Speakman and Smith J . SOC. Djters Col. 193G 52 131. Garvie Griftillis and Neale Tmits. Farnday SOC. 1934 30 271 T. R. E. KRESSMAN AND J.A. KITCHENER 10.3 penetrating and reaching all the exchanging sites in this resin and it seems likely that the higher activation energy is a consequence of the steric difficulty which it experiences in moving through the molecular pores of the resin. On the other hand the smaller ions showing P-mechanism evidently do not experience any such " wall-effect " and move simply through the water which fills the pores. It is not invariably true that diffusion in solids requires large energies of activation and results in agreement with the present picture have been obtained by other workers. The high value with fibres may reflect distortion of a flexible macromolecular structure by the dye molecule. With rigid porous solids the activation energy for the diffusion of molecules smaller tha% the $ores may be almost normal for an aqueous medium.For example Tiselius l 3 found 5-4 kcal./mole for the diffusion of water normal to the (201) face in the zeolite heulandite. On the other hand diffusion normal to the (001) face required 9.1 kcal/mole ; presumably the interstitial holes in this direction are much smaller and of about the same size as the water molecule. Diffusion of salts in z yo agar gels shows the same temperature coefficient as for free diffusion in aqueous s~lution.~ Role of the Anion.-To preserve electro-neutrality the ions leaving the resin must be replaced by others of the same total charge. If the ions were bound to specific sites the situation might arise where the rate of dissociation from sites was the factor limiting the rate of exchange and it might then be necessary to provide extra (soluble) anions by diffusion of salt from the ambient solution into the interstices of the resin before rapid cation exchange could take place, This possibility was studied by the experiments (recorded above) in which the anion was provided by the dye Chlorazol Sky Blue FFS which is found to diffuse extremely slowly and to an extremely small extent into the resin.With both Na+ and NEt,+ the rate of exchange was the same as with the simple halide salts (see Fig. 5 curve 2 ; Fig. 7 curve 4 ; Fig. 11 curve 3), indicating that it is not necessary to have soluble anions within the pores for exchange to occur rapidly. The high mobility of the cations alone is proved conclusively by the experiment with finely divided NEt,-resin suspension in place of NEt,Br solution. This suspension exchanges cations with normal NH,-resin grains almost as rapidly as does the homogeneous salt solution the apparent rate constants being in the ratio of 0.7411 (cf. Fig. 5 curves 7 and IO) proving that soluble anions play no significant part at all in the kinetics of cation exchange at least with the resinous exchanger studied here. In this experiment the process starts with the interpenetration of the Gouy diffuse double layers of the large grains with those of the small suspended particles. At the very low ionic strengths prevailing the double layers spread so far from the microscopic particles that the cation distribu-tion in the liquid is similar to that in a true salt solution. Consequently, the outside part of the double layer from the large granules can mix rapidly with the cations of the suspension leaving diffusion of cations within the pores of the resin as the rate-controlling process as it is with the ordinary salt solution. A similar process no doubt occurs in normal exchanges with soluble salts-the soluble cations exchange first with the resin cations in the outer diffuse double layer thus setting up a concentration gradient within the pores which is eliminated by ordinary diffusion this latter process requiring no soluble anions. l3 Tiselius Z. physik. Chew. A 1934 169 425 ; 1935 174 4 0 1 104 ION EXCHANGE STUDIES We are indebted to Prof. H. V. A. Briscoe for his encouragement and interest in the work and to Dr. R. F. Hudson for helpful discussion. One of us (T.R.E.K.) thanks The Permutit Co. Ltd. Chiswick W.4 for a grant in support of the work and for permission to publish it. Imperial CoLlege of Science G Technology, South Kensington London S. W.7

ISSN:0366-9033    

DOI:10.1039/DF9490700090

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

ION EXCHANGE STUDIES ION EXCHANGE STUDIES 11. The Determination of Thermodynamic Equilibrium Constants - (1) K - - . - a& BY J. F. DUNCAN AND B. A. J. LISTER Received 13th JuZy 1949 A critical examination of the postulates used by different authors to evaluate the activities of ions in an exchanger is given and an experimental test is made using the barium-hydrogen and lanthanum-ammonium exchange systems. It is shown that the assumption that the activities in the exchanger are proportional to the molar concentra- tions (in arbitrary units) gives an approximately constant value for the mass product but even on this postulate the mass product varies by about 50 yo rising to a maximum a t an equivalent fraction of about for the multivalent ion in the exchanger.It is suggested that the activities in the exchanger have been wrongly evaluated because of the influence of some secondary process such as adsorption solution of ions in the exchanger or swelling of the resin. In this paper the following symbols have been used a thermodynamic activity. K, thermodynamic equilibrium constant. c molar concentration. x molar fraction. C equivalent concentration. X equivalent fraction defined as the ratio of the number of equivalents of a given cation to the total number of equivalents of all cations in the sane phase. B%i + zHR + Ba + zHg aBaR a - a B a g ' a&E 12 number of moles. Suffixes have been used to indicate the ion and the particular phase to which reference is made thus xBaR is the molar fraction of barium ions in the resinous exchanger.uHS is the thermodynamic activity of hydrogen ions in solution. The thermodynamic equilibrium constant of an ion-exchange reaction of the type is written Whilst the activities a,,, aHs in solution may be readily evaluated in favour- able cases from known activity data some assumption must be made for the activities of the two ions in the solid phase. Hitherto at least three different assumptions have been made for aBaR and aHR as follows J. F. DUNCAN AND B. A. J. LISTER 105 (i) Boyd Schubert and Adamsonl have assumed that the ions in the solid phase exist in ideal solution and have put the activities in the solid phase equal to the molar fractions of the two species. If the x terms are molar fractions and the n terms are the number of moles of Ba and H R then QBBs @lR I( - - .A = XBaR aBag X'. a a - CBa C R .- (nBaR + %,> %BaR a& . . (3) By plotting log against log (aBa8/a:%) for the sodium-barium system a straight line was obtained by Boyd Schubert and Adamson which seemed to indicate that the concept of ideal solution in thesolid phase was a reliable hypothesis. (ii) Bauman and Eichhorn 2 on the other hand have written the equilibrium constant of a reaction of the above type in a form corresponding to CBaR a& K =-.- c L ~ aBas ' where the c terms are molar concentrations and have derived an equation of the form where cR is the total molar exchange capacity of the exchanger and cs the total molar concentration of the solution.It is difficult to see how this equation can be derived from eqn. (4) unless some assumption has been made which has not been specifically mentioned in their paper. Nevertheless these authors make a test of their equation which appears to support their postulates. (iii) Kressman and Kitchener 3 have assumed that the activities of the ions in the exchanger are proportional to the equivalent fractions defined as nBar. = - = X B n R and aHn = - CHR = XHR = I - X B a g the exchanger. Hence c, where the C terms are the number of equivalents of the ions in the exchanger C is the total capacity and the X terms are the equivalent fractions in In this treatment the activity coefficients of the ions in solution have also been neglected and the equation written The mass-law product estimated from an equation of this form was found to be constant over a range of 0 ~ 7 6 ~ X ~ ~ 0 ~ 8 8 for the barium-ammonium system (and over similar ranges for the other systems) and it was concluded that the changes occurring in yDas/y&s must be accompanied by propor- tionate changes in yBaE/&.It appears surprising that three different assumptions for the activities of the ions in the exchanger can lead to equations which are apparently Boyd Schubert and Adamson J . Amer. Chetn. SOC. 1947 Cg 2818. Bauman and Eichhorn ibid. 2830. Kressman and Kitchener J . Chem. SOC. 1949 1201. D" ION EXCHANGE STUDIES - cBaR - cR - ZCBaR cli - (8) (I - XBag)2 - cz, X% a - GR - The postulates of Boyd Schubert and Adamson and those of the other two sets of authors are however mutually exclusive for if eqn.(4) is written in terms of the number of moles of the two ions in a given weight of exchanger we arrive at This is different from eqn. (3) which has a terni + '12,~) equal to the total number of moles of the two ions in the exchanger which must vary according to the position of equilibrium. Thus it is impossible to obtain a constant mass-law product using eqn. (3) from results which give a constant with eqn. (4) and (6). In view of this confusion a thorough test of the mass-action concept has been made by estimating the mass-law product from the same results using eqn. (3) and also by assuming the activities in the exchanger to be equal to the molar concentrations (i-e.neglecting the activity coefficients of the ions in the exchanger). In order to simplify the presentation of the results eqn. (4) obtained on the latter assumption has been used in the form given by substitution of the equivalent fractions for the concentration terms. Thus K CBaR . "8 GR amg CBaX c i s $& -.-.- - YB% 2ckR CBap ZcgR CBeS CBag -.-.- - 2 G S y& yBa8 - XBaR XgS -& Cs -~ .-.-._ XgR XBea YRag C 106 equally successful in describing the results. All three approaches when applied to exchanging ions of the same valency lead to the same equations which would therefore be expected to give a constant mass-law product for homovalent exchange.We have previously shown that the mass- law product for the sodium-hydrogen system is not constant but rises to a maximum for a value of 0.12 for the molar fraction of sodium ions in the exchanger. This suggests that the simple assumptions made for the activities of ions of the same valency in the exchanger is not valid and a $riori one would expect these assumptions to be in greater error for heterovalent exchange for which the system is likely to be less ideal. The situation is further confused by the apparent contradictory nature of eqn. (3) and (4). If the capacity of the exchanger is constant the postulates of Kressman and Kitchener and of Bauman and Eichhorn are such that a constant mass-law product can be obtained from the same set of results by use of both eqn.(4) and (7) (provided the ratio of the activity coefficients of the ions in solution does not vary greatly) for Hence Duncan and Lister ibid. (in press). 107 where C is the total equivalent concentration in solution and the X terms are equivalent fractions of the respective ions in the exchanger and in solution. The ratio of the K values given by eqn. (10) and (6) is equal to the capacity of the exchanger. For the barium-hydrogen system plots of K against xBaE and against XBag were made to test eqn. (3) and (10) respectively. In both cases the mass-law product was found to vary but the curve obtained using eqn. (10) was more nearly constant and had a form similar to that obtained for the sodium-hydrogen system. The same was true of the exchange of lanthanum and ammonium ions the value of K being evaluated according to the eq uat ion J.F. DUNCAN AND B. A. J. LISTER Experimental The position of equilibrium between exchanger and solution has been deter- mined (i) by batch equilibration methods and (ii) by determining the break- through volume necessary for a column of exchanger in the hydrogen form to be saturated by a given mixture of metal ions and hydrogen ions. The apparatus was in principle the same as that used for investigating the sodium-hydrogen system.4 Radiochemical tracer methods were used to study both the barium and the lanthanum systems the two tracers being 139Ba (half-life 86 min. p-energy 2.3 MeV y-energy 0.6 MeV) and 1*OLa (half-life 40 hr. P-energy 1-45 MeV y-energies 0.87 0.49 and 0.33 MeV) respectively.In the lanthanum-ammonium batcli equilibration method volumes of radioactive lanthanum nitrate solution from IOO ml. to 5 1. were made u p with ammonium nitrate to a total concentration of 0.1 N the lanthanum concentration being varied from I O - ~ to I O - ~ N. To these solutions known weights of Dowex 50 (0.15 to 0.007 g.) were added the exchanger (in the ammonium form) having been previously dried in an oven a t IIOO C to constant weight and the capacity having been determined by the column method using lanthanum. A similar batch equilibration method was also used for the barium-hydrogen system. In the breakthrough volume method a mixture o€ barium chloride and hydrochloric acid a t a total concentration of 0.2 N or of lanthanum nitrate and ammonium nitrate a t a total concentration of 0-1 N was passed down the column saturated with the univalent ion.Since the multivalent ion is held more strongly a sharp boundary was obtained and was easily observed by measuring the radioactivity of the liquid leaving the column by means of a Geiger-Muller counter of the liquid flow type.$ After correcting for the paralysis time of the counting assembly and the decay of the radio-tracer used the coilcentration of the barium or lanthanum entering the column was equated to the activity observed after the column was saturated. By measuring the shaded area shown in Fig. I the amount of the multivalent ion taken up by the column for a given concentration in solution may be estimated and hence the niass-law product determined.In order to estimate the mass-law product for values of XBaR and XLaR below about 0-2 it is necessary to reduce the concentration of the respective ions in solution to I O - ~ M or less. With such low concentrations very large break- through volumes are necessary unless the capacity of the column is kept low. In one experiment a column containing about 10 mg. Dowes j o of capacity 0.0446 m. equiv. required 1000 ml. of a solution containing 1.8 x I O - ~ g./l. of lanthanum nitrate to saturate the column a t a value of XLaR = 0-417. To study the equilibrium a t very low concentrations of the multivalent ion it was necessary to use batch methods since the flow velocity becomes too fast for reasonably sharp boundaries to be obtained if the breakthrough volume is to be reached within a time during which the radio-tracer is still active.The values of XBaR XBsg Cp and Cs obtained experimentally were substi- 108 ION EXCHANGE STUDIES tuted in eqn. (10). Corrections for the activity coefficients y& and yk& for mixtures of these two ions are necessary and may be estimated from the activities of the electrolytes in mixtures of barium chloride and hydrochloric acid for where y k ~ ~ ~ (BaClZ) and y*Braz~cl) are the activity coefficients of the eleckobtes in mixtures of the two. Now i t is possible to estimate y*HCl(Bauz) data of Randall and Breckenridge 6 by use of the equation log YklXCl (Boc11) = log YZHCl + a12CB&Clzgi from the where yoHcl is the activity coefficient of hydrochloric acid in solutions of the same ionic strength.From these data plots were made of logy*Ha against cBaPg for solutions of constant ionic strengths from which aI2 was estimated and used to determine yfHCl(BrC1s) in the solutions. 'C'nfortunately data were not available to enable ykBaClr (HCI) to be estimated but since y~Hcl(Baclz) was found to differ from y s c1 by less than I % it was considered reliable to use y&aCIz as y*BaC1z(HCI) in solutions of the same ionic strength. The expression corresponding to eqn. (12) for the lanthanum-ammonium svstem is tration for the same range was always less than 0.01 N. For values of XUR below 0.8 the ammonium ion concentration was 0-ogg-0.1 N whilst the lanthanum ion concentration was less than 0.001 N.Although the shape of the mass-law product plot may be slightly in error for XLaR > 0.8 the activity correction is almost certainly constant within about I yo for lower values of XL%. FIG. I .-Diagrammatic representation of concentration volume plot for breakthrough experiment. In this case the activity data even for pure lanthanum nitrate are not available and no correction could be made. The results given below therefore do not represent the true value of the mass-law product but the general variation will by more than 0.09 to 0.10 N for XLaa < 0.9 whilst the lanthanum nitrate concen- be qualitatively correct. The concentration of ammonium nitrate did not vary 6 Randall and Breckenridge J .Amev. Chcm. SOL 1927 49 1435. 109 J. F. DUNCAN AND B. A. J. LISTER FIG. 2.-Mass-law product' plot for barium-hydrogen system according to eqn. (10). A.-Room temperature (with no correction for activity coefficients in solution). B.-Room temperature (with activity coefficient correction). c.-87O C (with no correction for activity coefficients in solution). 0 CD 0 Breakthrough experiments. 0 0 9 Batch equilibration experiments. FIG. 3.-Mass-law product plot for barium-hydrogen system a t room temperature according to eqn. (3) (with no correction for activity coefficients in solution) 0 Breakthrough experiments. 0 Batch equilibration experiments. I10 ION EXCHANGE STUDIES FIG. q.-lblass-law product plot for lanthanum-ammonium system according to eqn.(11) (with no correction for activity coefficients in solution). 0 Breakthrough experiments. Points interpolated from XLaB-XLag curve constructed from batch equilibration data. FIG. 5 .-Mass-law product plot for lan thanurn-ammonium system according to solid solution concept (corresponding to eqn. (3) with no correction for activity coeffi- cients in solution). 0 Breakthrough experiments. Points interpolated from XLaR-XLaS curve constructed from batch equilibration data. The variation in the mass-law product calculated according to eqn. (10) from the equilibrium data given in Table I and I1 for the barium chloride- hydrochloric acid system is shown in Fig. 2 . It will be seen that a t room temperature the mass-law product varies by about 50 yo from high values of X B ~ ~ to a maximum at a value of X B ~ ~ = O - ~ I .By comparison Fig. 3 shows that the mass-law product derived from the same experimental results by use of eqn. (3) varies bya factor of 2 or more over the same range,the maximum being obtained at X ~ ~ ~ = 0 - 1 7 corresponding to X B ~ ~ = O . ~ I . From this it was concluded that neither approach was strictly valid although the activities of the ions in the exchanger were more nearly approximated by the use of molar concentrations in the exchanger expressed in arbitrary units. Fig. 2 may be used if desired to estimate the variation of the activity coefficients of the ions in the exchanger although the values obtained depend on the value of K which is assumed to be valid and on the units of concentration.For the lanthanum-ammonium system the equilibrium data are given in Table 111 and the curve obtained by use of eqn. (11) without making corrections for the activities of the ions in solution is shown in Fig. 4. The curve calculated assuming the validity of the solid solution concept (corre- sponding t o eqn. (3)) is shown in Fig. 5. Much greater variations of K are obtained thus supporting the conclusion that eqn. (11) (which assumes Results and Discussion TABLE I 0.210 0.207 0'209 0.209 0.208 0 ' 2 1 ~ 0.200 EXPERIMENTAL DATA FOK BA~+-H+ SYSTEM (TOTAL CONCENTRATIOK Total Equivalent Concentrat ion (C4 0.210 Breakthrough Experiments 0'220 0.01254 0.0055 0.00254 0'00 I g I 0.00135 0.000767 0~000658 0.000542 0.000272 0~000231 0~000163 0-0001025 0~000063 0.0000436 0~0000263 0~0000225 0.2 N) Equivalent Concentration of Barium Capacity of Exchanger (m.equiv.) (CBas) 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.130 0.500 No. of m. equiv. of Barium in Exchanger 1-15 0.414 1.15 1-15 1-15 0.340 0.315 0.380 0.289 0.248 0.2xo 0.210 0'210 0.195 0.204 0.195 0'200 0'2OU J. I;. DUNCAN AND €3. A. J. LISTER 0~00002 I I 0.130 0.549 1.360 0.995 I11 0.237 0.224 0.168 0.156 0.121 0.02 I I 0*0481 I. 15 1.15 1-15 1-15 1-15 1-15 1-14 x.15 1-14 1-14 0-0106 0'0201 0.0239 1-14 1-14 0.0253 1-14 112 TABLE I1 EXPERIMENTAL DATA FOR BA~+-H+ SYSTEMS AT 87' C (TOTAL CON- Total Equivalent Concentration 0.206 (CS) o m 8 No.of 111. equiv. of Barium in Exchanger 0.212 0.213 0'2x2 0'210 0.208 0.206 CENTRATION 0.2 N) Equivalent Concentration of Barium Capacity of Exchanger (m. equiv.) (CBa,) 0'0201 0-0105 0.0080 0*00320 0*001gg 0.000629 0.000365 0~0012 I 0-0004g8 0.000443 0.206 0.206 0.206 0.206 0.206 0.206 0-000334 0~000217 0~000135 0~0000812 TABLE I11 0.1 N ) Equivalent Concentration Capacity of Exchanger of Lanthanum 0~00000625 0-00000575 EXPERIMENTAL DATA FOR LA'+-NH 2 SYSTEM (TOTAL CONCENTRATION Total Equivalent Concentration (m.equiv.) (CLa,) (CS) 3.107 3.107 0.410 0.0446 0.0446 0.0639 0.0446 Breakthrough Experiments 0.500 0'1000 O*I002 O*IOOO 0.0998 0~1000 O*IOOO O.IOO0 0.01860 Batch Equilibration Exjvriments ('Typical results from a series of 70 experi- men t s) 0*IOOO 0'0102 0~00100 0.000676 o~ooo119 0.0000364 0~0000186 0.0000797 0.0000339 0~0000203 0*00000936 0 ~ 0 0 0 0 0 ~ 6 ~ O*IOoO 0.1000 0'0999 0-1000 O'IOOO 0'1000 0'1000 0.151 0.0709 0.032 I 0.0426 0.0353 0.0606 0.0697 0-0378 No. of m. equiv. of La3+ in Exchanger 3.000 2.810 0.3250 0.03424 0.02.778 0.03208 0.08800 0'03470 0~01400 0~00000 I2 I 0.6573 095890 0*5029 0.4260 0.5760 0.3258 0.2560 0.6077 0.225% 0-1657 0.2476 0.3060 0.1333 0.1672 0*01485 0.01 I I0 0~01800 0.01950 0.00375 J.F. DUNCAN AND B. A. J. LISTER 113 proportionality between activities and molar concentrations) is more nearly valid. The fact that variations in the mass-law product are obtained here for systems shown by other authors to give constants is evidently attributable to the fact that most of the earlier mass-law product determinations were made over quite small ranges. For example Kressman and Kitchener worked within the ranges 0.76 < X B ~ ~ <0-88 and 0.71 < X A ~ ~ <0-88 over which the mass-law product is seen not to vary by more than about 10 yo and Boyd Schubert and Adamson similarly used restricted ranges (estimated for the barium-sodium system as 0.64 <o.g~ and for the lanthanum- sodium system as 0.87 <XLaR < 0.92).(Reproduced by peririission of the Ckeinical Society.) FIG. 6.-Mass-law product for the sodium-hydrogen system (C = 0.2 N). A.-Room temperature. ~ . - - 8 7 O C (correction for activity coefficient in solution). a Breakthrough experiments. Batch equilibration experiments. The use of eqn. (10) and (11) is seen to give results for the exchange of uni-bi and uni-tervalent ions which are very similar to those obtained for uni-univalent exchange. For comparison the mass-law product of the sodium-hydrogen system at room temperature is shown in Fig.6. Since the concepts of Kressman and Kitchener and of Bauman and Eichhorn must lead to mass-law products which vary in the same way as those shown in Fig. 2 4 and 6 no method yet proposed is satisfactory in evaluating the activities of the ions in the exchanger other than to a first approximation. Whilst it is possible that some other approach may be satisfactory it seems to us that a fundamental feature may have been omitted in all the methods listed. For instance whilst it is undoubtedly true that to a first approxima- tion the total capacity of an exchanger is constant and it is probably correct to assume the exchange capacity to be constant it is possible that the resin takes up non-exchangeable material by adsorption on the surface by solution in the exchanger or by other processes.In this respect there is a little experimental evidence ; the height of the maximum obtained in the mass-law product plots for the sodium-hydrogen and the barium-hydrogen systems decreases as the temperature increases (see Fig. z and 6) as would be expected for the contribution of a non-exchange process (e.g. an adsorp- tion phenomenon) which usually decreases with increasing temperature. Differences in mass-law product may also be caused by swelling phenomena as has been suggested by Gregors on thermodynamic grounds. It is already Gregor ibid. 1948 70 1293. 114 EXCHANGE EQUILIBRIA I N POROUS ANION EXCHANGERS known that ion exchange material swells and contracts in solutions of different electrolytes or in solutions of the same electrolyte in different con- centrations and it is possible that swelling may also be different according to the ratio of the ions in the exchanger.Further experimental work is required to ascertain whether such secondary processes hitherto uncon- sidered have an effect on the exchange equilibrium. We should like to acknowledge the assistance of Mr. P. E. Brown with the experimental work and Mr. M. A. Hewitt with the computing. We are indebted to Dr. E. Glueckauf for his interest in this investigation and we should like to thank the Director of the Atomic Energy Research Establish- ment for permission to publish this paper. ADDENDuM.-Some confusion seems to have arisen concerning the method of evaluating K by use of eqn.(10) from the experimental results. The figures shown in column 3 of Table 1-111 were not used as C, as this would lead to a value of K which depends on the amount of exchanger used. In order that K should refer to the same quantity of exchanger C was taken as the capacity of T g. of exchanger (4-3 m. e q u i ~ . / g . ) ~ . Th.e Atomic Energy Research Establishment Harwell Berkshire. ION EXCHANGE STUDIES ION EXCHANGE STUDIES 11. The Determination of Thermodynamic Equilibrium Constants BY J. F. DUNCAN AND B. A. J. LISTER Received 13th JuZy 1949 A critical examination of the postulates used by different authors to evaluate the activities of ions in an exchanger is given and an experimental test is made using the barium-hydrogen and lanthanum-ammonium exchange systems.It is shown that the assumption that the activities in the exchanger are proportional to the molar concentra-tions (in arbitrary units) gives an approximately constant value for the mass product, but even on this postulate the mass product varies by about 50 yo rising to a maximum a t an equivalent fraction of about for the multivalent ion in the exchanger. It is suggested that the activities in the exchanger have been wrongly evaluated because of the influence of some secondary process such as adsorption solution of ions in the exchanger or swelling of the resin. In this paper the following symbols have been used : a thermodynamic activity. K, thermodynamic equilibrium constant. c molar concentration. x molar fraction.C equivalent concentration. X equivalent fraction defined as the ratio of the number of equivalents of a given cation to the total number of equivalents of all cations in the sane phase. 12 number of moles. Suffixes have been used to indicate the ion and the particular phase to which reference is made thus xBaR is the molar fraction of barium ions in the resinous exchanger. uHS is the thermodynamic activity of hydrogen ions in solution. The thermodynamic equilibrium constant of an ion-exchange reaction of the type is written B%i + zHR + Ba + zHg - (1) aBaR a&, a&E a B a g ' K - - . - a -Whilst the activities a,,, aHs in solution may be readily evaluated in favour-able cases from known activity data some assumption must be made for the activities of the two ions in the solid phase.Hitherto at least three different assumptions have been made for aBaR and aHR as follows J. F. DUNCAN AND B. A. J. LISTER 105 (i) Boyd Schubert and Adamsonl have assumed that the ions in the solid phase exist in ideal solution and have put the activities in the solid phase equal to the molar fractions of the two species. If the x terms are molar fractions and the n terms are the number of moles of Ba and H R , then . . (3) I( - - . A = .- XBaR (nBaR + %,> %BaR a& a -X'. aBag @lR QBBs By plotting log against log (aBa8/a:%) for the sodium-barium system a straight line was obtained by Boyd Schubert and Adamson, which seemed to indicate that the concept of ideal solution in thesolid phase was a reliable hypothesis.(ii) Bauman and Eichhorn 2 on the other hand have written the equilibrium constant of a reaction of the above type in a form corresponding to CBaR a& K =-.-a c L ~ aBas ' where the c terms are molar concentrations and have derived an equation of the form where cR is the total molar exchange capacity of the exchanger and cs the total molar concentration of the solution. It is difficult to see how this equation can be derived from eqn. (4) unless some assumption has been made which has not been specifically mentioned in their paper. Nevertheless these authors make a test of their equation which appears to support their postulates. (iii) Kressman and Kitchener 3 have assumed that the activities of the ions in the exchanger are proportional to the equivalent fractions defined as CBa CHR = X B n R and aHn = - = XHR = I - X B a g c, C R nBar.= -where the C terms are the number of equivalents of the ions in the exchanger, C is the total capacity and the X terms are the equivalent fractions in the exchanger. Hence In this treatment the activity coefficients of the ions in solution have also been neglected and the equation written The mass-law product estimated from an equation of this form was found to be constant over a range of 0 ~ 7 6 ~ X ~ ~ 0 ~ 8 8 for the barium-ammonium system (and over similar ranges for the other systems) and it was concluded that the changes occurring in yDas/y&s must be accompanied by propor-tionate changes in yBaE/&. It appears surprising that three different assumptions for the activities of the ions in the exchanger can lead to equations which are apparently Boyd Schubert and Adamson J .Amer. Chetn. SOC. 1947 Cg 2818. Bauman and Eichhorn ibid. 2830. Kressman and Kitchener J . Chem. SOC. 1949 1201. D 106 ION EXCHANGE STUDIES equally successful in describing the results. All three approaches when applied to exchanging ions of the same valency lead to the same equations, which would therefore be expected to give a constant mass-law product for homovalent exchange. We have previously shown that the mass-law product for the sodium-hydrogen system is not constant but rises to a maximum for a value of 0.12 for the molar fraction of sodium ions in the exchanger. This suggests that the simple assumptions made for the activities of ions of the same valency in the exchanger is not valid and, a $riori one would expect these assumptions to be in greater error for heterovalent exchange for which the system is likely to be less ideal.The situation is further confused by the apparent contradictory nature of eqn. (3) and (4). If the capacity of the exchanger is constant the postulates of Kressman and Kitchener and of Bauman and Eichhorn are such that a constant mass-law product can be obtained from the same set of results by use of both eqn. (4) and (7) (provided the ratio of the activity coefficients of the ions in solution does not vary greatly) for cBaR - cR ZCBaR cli - (8) - X% -(I - XBag)2 - cz, GR -The postulates of Boyd Schubert and Adamson and those of the other two sets of authors are however mutually exclusive for if eqn.(4) is written in terms of the number of moles of the two ions in a given weight of exchanger we arrive at This is different from eqn. (3) which has a terni + '12,~) equal to the total number of moles of the two ions in the exchanger which must vary according to the position of equilibrium. Thus it is impossible to obtain a constant mass-law product using eqn. (3) from results which give a constant with eqn. (4) and (6). In view of this confusion a thorough test of the mass-action concept has been made by estimating the mass-law product from the same results using eqn. (3) and also by assuming the activities in the exchanger to be equal to the molar concentrations (i-e. neglecting the activity coefficients of the ions in the exchanger).In order to simplify the presentation of the results eqn. (4) obtained on the latter assumption has been used in the form given by substitution of the equivalent fractions for the concentration terms. Thus K CBaR . "8 a - GR amg CBaX c i s $& 2ckR CBap YB% CBag 2 G S y& ZcgR CBeS yBa8 XBaR XgS -& Cs XgR XBea YRag C, -.-.- --.-.- --~ .-.-._ -Hence Duncan and Lister ibid. (in press) J. F. DUNCAN AND B. A. J. LISTER 107 where C is the total equivalent concentration in solution and the X terms are equivalent fractions of the respective ions in the exchanger and in solution. The ratio of the K values given by eqn. (10) and (6) is equal to the capacity of the exchanger. For the barium-hydrogen system plots of K against xBaE and against XBag were made to test eqn.(3) and (10) respectively. In both cases the mass-law product was found to vary but the curve obtained using eqn. (10) was more nearly constant and had a form similar to that obtained for the sodium-hydrogen system. The same was true of the exchange of lanthanum and ammonium ions the value of K being evaluated according to the eq uat ion, Experimental The position of equilibrium between exchanger and solution has been deter-mined (i) by batch equilibration methods and (ii) by determining the break-through volume necessary for a column of exchanger in the hydrogen form to be saturated by a given mixture of metal ions and hydrogen ions. The apparatus was in principle the same as that used for investigating the sodium-hydrogen system.4 Radiochemical tracer methods were used to study both the barium and the lanthanum systems the two tracers being 139Ba (half-life 86 min.p-energy 2.3 MeV y-energy 0.6 MeV) and 1*OLa (half-life 40 hr. P-energy 1-45 MeV, y-energies 0.87 0.49 and 0.33 MeV) respectively. In the lanthanum-ammonium batcli equilibration method volumes of radioactive lanthanum nitrate solution from IOO ml. to 5 1. were made u p with ammonium nitrate to a total concentration of 0.1 N the lanthanum concentration being varied from I O - ~ to I O - ~ N. To these solutions known weights of Dowex 50 (0.15 to 0.007 g.) were added the exchanger (in the ammonium form) having been previously dried in an oven a t IIOO C to constant weight and the capacity having been determined by the column method using lanthanum.A similar batch equilibration method was also used for the barium-hydrogen system. In the breakthrough volume method a mixture o€ barium chloride and hydrochloric acid a t a total concentration of 0.2 N or of lanthanum nitrate and ammonium nitrate a t a total concentration of 0-1 N was passed down the column saturated with the univalent ion. Since the multivalent ion is held more strongly, a sharp boundary was obtained and was easily observed by measuring the radioactivity of the liquid leaving the column by means of a Geiger-Muller counter of the liquid flow type.$ After correcting for the paralysis time of the counting assembly and the decay of the radio-tracer used the coilcentration of the barium or lanthanum entering the column was equated to the activity observed after the column was saturated.By measuring the shaded area shown in Fig. I the amount of the multivalent ion taken up by the column for a given concentration in solution may be estimated and hence the niass-law product determined. In order to estimate the mass-law product for values of XBaR and XLaR below about 0-2 it is necessary to reduce the concentration of the respective ions in solution to I O - ~ M or less. With such low concentrations very large break-through volumes are necessary unless the capacity of the column is kept low. In one experiment a column containing about 10 mg. Dowes j o of capacity 0.0446 m. equiv. required 1000 ml. of a solution containing 1.8 x I O - ~ g./l. of lanthanum nitrate to saturate the column a t a value of XLaR = 0-417.To study the equilibrium a t very low concentrations of the multivalent ion it was necessary to use batch methods since the flow velocity becomes too fast for reasonably sharp boundaries to be obtained if the breakthrough volume is to be reached within a time during which the radio-tracer is still active. The values of XBaR XBsg Cp and Cs obtained experimentally were substi 108 ION EXCHANGE STUDIES tuted in eqn. (10). Corrections for the activity coefficients y& and yk& for mixtures of these two ions are necessary and may be estimated from the activities of the electrolytes in mixtures of barium chloride and hydrochloric acid for where y k ~ ~ ~ (BaClZ) and y*Braz~cl) are the activity coefficients of the eleckobtes in mixtures of the two.Now i t is possible to estimate y*HCl(Bauz) from the data of Randall and Breckenridge 6 by use of the equation where yoHcl is the activity coefficient of hydrochloric acid in solutions of the same ionic strength. From these data plots were made of logy*Ha against cBaPg for solutions of constant ionic strengths from which aI2 was estimated and used to determine yfHCl(BrC1s) in the solutions. log YklXCl (Boc11) = log YZHCl + a12CB&Clzgi FIG. I .-Diagrammatic representation of concentration volume plot for breakthrough experiment. 'C'nfortunately data were not available to enable ykBaClr (HCI) to be estimated, but since y~Hcl(Baclz) was found to differ from y s c1 by less than I % it was considered reliable to use y&aCIz as y*BaC1z(HCI) in solutions of the same ionic strength.The expression corresponding to eqn. (12) for the lanthanum-ammonium svstem is In this case the activity data even for pure lanthanum nitrate are not available and no correction could be made. The results given below therefore do not represent the true value of the mass-law product but the general variation will be qualitatively correct. The concentration of ammonium nitrate did not vary by more than 0.09 to 0.10 N for XLaa < 0.9 whilst the lanthanum nitrate concen-tration for the same range was always less than 0.01 N. For values of XUR below 0.8 the ammonium ion concentration was 0-ogg-0.1 N whilst the lanthanum ion concentration was less than 0.001 N. Although the shape of the mass-law product plot may be slightly in error for XLaR > 0.8 the activity correction is almost certainly constant within about I yo for lower values of XL%.6 Randall and Breckenridge J . Amev. Chcm. SOL 1927 49 1435 J. F. DUNCAN AND B. A. J. LISTER 109 FIG. 2.-Mass-law product' plot for barium-hydrogen system according to eqn. (10). A.-Room temperature (with no correction for activity coefficients in solution). B.-Room temperature (with activity coefficient correction). c.-87O C (with no correction for activity coefficients in solution). 0 CD 0 Breakthrough experiments. 0 0 9 Batch equilibration experiments. FIG. 3.-Mass-law product plot for barium-hydrogen system a t room temperature according to eqn. (3) (with no correction for activity coefficients in solution), 0 Breakthrough experiments.0 Batch equilibration experiments I10 ION EXCHANGE STUDIES FIG. q.-lblass-law product plot for lanthanum-ammonium system according to eqn. (11) (with no correction for activity coefficients in solution). 0 Breakthrough experiments. Points interpolated from XLaB-XLag curve constructed from batch equilibration data. FIG. 5 .-Mass-law product plot for lan thanurn-ammonium system according to solid solution concept (corresponding to eqn. (3) with no correction for activity coeffi-cients in solution). 0 Breakthrough experiments. Points interpolated from XLaR-XLaS curve constructed from batch equilibration data J. I;. DUNCAN AND €3. A. J. LISTER I11 Results and Discussion The variation in the mass-law product calculated according to eqn.(10) from the equilibrium data given in Table I and I1 for the barium chloride-hydrochloric acid system is shown in Fig. 2 . It will be seen that a t room temperature the mass-law product varies by about 50 yo from high values of X B ~ ~ to a maximum at a value of X B ~ ~ = O - ~ I . By comparison Fig. 3 shows that the mass-law product derived from the same experimental results by use of eqn. (3) varies bya factor of 2 or more over the same range,the maximum being obtained at X ~ ~ ~ = 0 - 1 7 corresponding to X B ~ ~ = O . ~ I . From this it was concluded that neither approach was strictly valid although the activities of the ions in the exchanger were more nearly approximated by the use of molar concentrations in the exchanger expressed in arbitrary units.Fig. 2 may be used if desired to estimate the variation of the activity coefficients of the ions in the exchanger although the values obtained depend on the value of K which is assumed to be valid and on the units of concentration. For the lanthanum-ammonium system the equilibrium data are given in Table 111 and the curve obtained by use of eqn. (11) without making corrections for the activities of the ions in solution is shown in Fig. 4. The curve calculated assuming the validity of the solid solution concept (corre-sponding t o eqn. (3)) is shown in Fig. 5. Much greater variations of K are obtained thus supporting the conclusion that eqn. (11) (which assumes TABLE I EXPERIMENTAL DATA FOK BA~+-H+ SYSTEM (TOTAL CONCENTRATIOK 0.2 N) Total Equivalent Concentrat ion (C4 Equivalent Concentration of Barium (CBas) Capacity of Exchanger (m.equiv.) No. of m. equiv. of Barium in Exchanger Breakthrough Experiments 0'220 0.210 0.210 0.207 0'209 0.209 0.208 0 ' 2 1 ~ 0.2xo 0.210 0'210 0.195 0.204 0.195 0.01254 0.0055 0.00254 0'00 I g I 0.00135 0.000767 0~000658 0.000272 0~000231 0~000163 0-0001025 0~000063 0.000542 0.0000436 0.200 0'200 0'2OU 0~0000263 0~0000225 0~00002 I I 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.500 0.130 0.500 0.130 0.549 1.360 0.995 0.414 0.380 0.340 0.315 0.289 0.248 0.237 0.224 0.168 0.156 0*0481 0-0106 0.121 0.02 I I 0'0201 0.0253 0.0239 1-15 1.15 1-15 1-15 I.15 1.15 1-15 1-15 1-15 1-15 x.15 1-14 1-14 1-14 1-14 1-14 1-1 112 TABLE I1 EXPERIMENTAL DATA FOR BA~+-H+ SYSTEMS AT 87' C (TOTAL CON-CENTRATION 0.2 N) Total Equivalent Concentration (CS) Equivalent Capacity of No. of 111. Exchanger equiv. of Barium of Barium in Exchanger (m. equiv.) Concentration (CBa,) 0.206 o m 8 0.213 0'210 0.212 0'2x2 0.208 0.6573 095890 0.5760 0*5029 0.4260 0.3258 0.2560 0.225% 0-1657 0.6077 0.2476 0.3060 0.1333 0.1672 0'0201 0-0105 0.0080 0*00320 0*001gg 0.000629 0.000365 TABLE I11 0.206 0.206 0.206 0.206 0.206 0.206 0.206 EXPERIMENTAL DATA FOR LA'+-NH 2 SYSTEM (TOTAL CONCENTRATION 0.1 N ) 0~0012 I 0-0004g8 0.000443 0-000334 0~000217 0~000135 0~0000812 Total Equivalent Concentration (CS) Breakthrough Experiments Equivalent Capacity of No.of m. Exchanger equiv. of La3+ in Exchanger Concentration of Lanthanum (m. equiv.) (CLa,) 0'1000 O*I002 O*IOOO 0.0998 0~1000 O*IOOO O.IOO0 0.500 0.000676 o~ooo119 0.0000364 0~0000186 0'0102 0~00100 3.107 3.107 0.410 0.0446 0.0446 0.0639 0.0446 3.000 2.810 0.03424 0.02.778 0.03208 0.01860 0.3250 Batch Equilibration Exjvriments ('Typical results from a series of 70 experi-men t s) 0*IOOO O*IOoO 0.1000 0'0999 0-1000 O'IOOO 0'1000 0'1000 0.0000797 0.0000339 0~0000203 0*00000936 0 ~ 0 0 0 0 0 ~ 6 ~ 0~00000625 0-00000575 0~00000 I2 I 0.151 0.0709 0.032 I 0.0426 0.0353 0.0606 0.0697 0-0378 0.08800 0'03470 0~01400 0*01485 0.01 I I0 0~01800 0.01950 0.0037 J.F. DUNCAN AND B. A. J. LISTER 113 proportionality between activities and molar concentrations) is more nearly valid. The fact that variations in the mass-law product are obtained here for systems shown by other authors to give constants is evidently attributable to the fact that most of the earlier mass-law product determinations were made over quite small ranges. For example Kressman and Kitchener worked within the ranges 0.76 < X B ~ ~ <0-88 and 0.71 < X A ~ ~ <0-88 over which the mass-law product is seen not to vary by more than about 10 yo, and Boyd Schubert and Adamson similarly used restricted ranges (estimated for the barium-sodium system as 0.64 <o.g~ and for the lanthanum-sodium system as 0.87 <XLaR < 0.92).(Reproduced by peririission of the Ckeinical Society.) FIG. 6.-Mass-law product for the sodium-hydrogen system (C = 0.2 N). A.-Room temperature. ~ . - - 8 7 O C (correction for activity coefficient in solution). a Breakthrough experiments. Batch equilibration experiments. The use of eqn. (10) and (11) is seen to give results for the exchange of uni-bi and uni-tervalent ions which are very similar to those obtained for uni-univalent exchange. For comparison the mass-law product of the sodium-hydrogen system at room temperature is shown in Fig. 6. Since the concepts of Kressman and Kitchener and of Bauman and Eichhorn must lead to mass-law products which vary in the same way as those shown in Fig.2 4 and 6 no method yet proposed is satisfactory in evaluating the activities of the ions in the exchanger other than to a first approximation. Whilst it is possible that some other approach may be satisfactory it seems to us that a fundamental feature may have been omitted in all the methods listed. For instance whilst it is undoubtedly true that to a first approxima-tion the total capacity of an exchanger is constant and it is probably correct to assume the exchange capacity to be constant it is possible that the resin takes up non-exchangeable material by adsorption on the surface by solution in the exchanger or by other processes. In this respect there is a little experimental evidence ; the height of the maximum obtained in the mass-law product plots for the sodium-hydrogen and the barium-hydrogen systems decreases as the temperature increases (see Fig. z and 6) as would be expected for the contribution of a non-exchange process (e.g. an adsorp-tion phenomenon) which usually decreases with increasing temperature. Differences in mass-law product may also be caused by swelling phenomena as has been suggested by Gregors on thermodynamic grounds. It is already Gregor ibid. 1948 70 1293 114 EXCHANGE EQUILIBRIA I N POROUS ANION EXCHANGERS known that ion exchange material swells and contracts in solutions of different electrolytes or in solutions of the same electrolyte in different con-centrations and it is possible that swelling may also be different according to the ratio of the ions in the exchanger. Further experimental work is required to ascertain whether such secondary processes hitherto uncon-sidered have an effect on the exchange equilibrium. We should like to acknowledge the assistance of Mr. P. E. Brown with the experimental work and Mr. M. A. Hewitt with the computing. We are indebted to Dr. E. Glueckauf for his interest in this investigation and we should like to thank the Director of the Atomic Energy Research Establish-ment for permission to publish this paper. ADDENDuM.-Some confusion seems to have arisen concerning the method of evaluating K by use of eqn. (10) from the experimental results. The figures shown in column 3 of Table 1-111 were not used as C, as this would lead to a value of K which depends on the amount of exchanger used. In order that K, should refer to the same quantity of exchanger C was taken as the capacity of T g. of exchanger (4-3 m. e q u i ~ . / g . ) ~ . Th.e Atomic Energy Research Establishment, Harwell Berkshire

ISSN:0366-9033    

DOI:10.1039/DF9490700104

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

EXCHANGE EQUILIBRIA IN ANION-EXCHANGE RESINS POROUS EXCHANGERS BY ROBERT KUNIN AND ROBERT J. MYERS Received 22nd August 1949 The exchange equilibria of strong-base and weak-basc anion-exchange resins prcpared with various degrees of cross-linking were examined t o determine the effect of this quantity upon the behaviour of the products. By suitable improvements in “ porosity ” the exchange capacity for large molecules such as penicillin or coloured bodies in beet molasses can be increased. The swelling and hydraulic characteristics are also affected by changes in cross-linking. The results are quite analogous t o the molecular-sievr effect noted for the adsorption of gases by zeolites and may therefore be considered an ‘‘ ionic-sieve ” effect. Chromatographic techniques that have employed ion-exchange substances as adsorbents have aided considerably in the analysis of rare earth mixtures,l solutions of amino acids,2 purine bases nucleic acids 3 and other systems The principles involved in these separations are similar to those involved in the chromatographic analysis of non-ionic constituents ; however superimposed on these principles are several added ones deriving from the ion-exchange process.Notably successful separations have been achieved through the application of variations in elution agents degree of column loading flow-rates during development and particle size of adsorbent. Differences in exchange equilibria have been called into service through searches for adsorbents with specific adsorptive properties or investigations 1 Boyd Schubert and Adamson J .Awcr. Chem. Soc. 1948 69 2818. Winters and Kunin Igzd. Eizg. Cheni. 1949 41 460. Cohn Science 1949 109 377. ROBERT KUNIN AND ROBERT J. MYERS into the effects of complexing agents and of pH on elution. Further extension of the application of exchange equilibria principles in chromatographic analysis has been made through the use of the newer synthetic resin exchange adsorbents that contain various functional groups and possess various degrees of selectivity based upon differences in ionic size. Equilibria in anion exchangers have been studied extensively by Wik- lander Myers Eastes and Urquhart Bhatnagar and co-workers Griess- bach and Kunin and Myers8 Most of these studies were conducted on weak-base type anion exchangers that is the synthetic resins employed were based upon alkylene polyamines or aromatic diamines.Kunin and McGarvey examined the equilibria relationships in a strong-base or quaternary-t ype anion exchanger. Titration curves of anion exchangers with strong acids clearly indicate the strength of the basic group. The order of decreasing exchange ability differs for the weakly basic resins as compared with the strong base chiefly with respect to the position of the hydroxyl ion. In the case of the strongly basic resin the order was found by Kunin and McGarvey 9 to be as follows for the strong-base exchanger Amberlite IRA-400 citrate > sulphate 7 oxalate > iodide > nitrate > chromate > bromide > thiocyanate > chloride >formate>hydroxyl>fluoride>acetate.The order of replacing ability is determined by means of “ symmetry ” studies in which the percentage exchange effected is plotted against the symmetry value or milli-equivalents of ion added per milli-equivalent of dry resin. Exchange equilibria relation- ships such as the Freundlich Langmuir or Rothmund-Kornfeld equations hold only over certain ranges of concentration and are best applied when ions of similar size and charge are involved. In general symmetry plots are readily obtained and are more meaningful in discussions of practical problems in chromatography. The availability of both weakly and strongly basic anion exchangers offers a range of operations to the laboratory investigator.Such an operating range is applicable mainly in cases of relatively small ions. When larger molecules such as complex colour bodies penicillin and ascorbic acid are under consideration it is desirable to employ modified exchangers that exhibit greater capacities for large ions. Recent advances in synethtic resin tech- nology have led to the development of ion-exchange resins with different degrees of “porosity” such that larger ions may be adsorbed and eluted effectively. Since the successful application of chromatographic techniques requires operation at or near equilibrium conditions the porosity of the exchanger becomes a vital factor since rates of diffusion throughout the resin particle vary with both the porosity of the resin and size of the ion.Samuelson l o and Bauman l1 have studied the effects of swelling and porosity upon the equilibria of cation exchange in sulphonic acid cation- exchange resins. It is an object of this paper to present some preliminary data on the performance of laboratory preparations of porous anion exchangers of the weak-base and strong-base type. Bhatnagar et al. J . Indian C h e w SOC. 1941 18 447. 7 Gricssbach 2. Ver. Detitsch. Cheudzer. Beih. 1939 31 I. 8 Kunin and Myers J . Anzer. Chenz. Sos. 1947 69 2874. Experimental The resins selected for this study were modifications of the Amberlite IRA-400 (Series A) studied by Kunin and McGarvey in which the density was altered 4 Wiklander -4 i m . Roy. Agric. Colt?. Swedcw 1946 14 I. 5 Myers Eastes and Urquhart Ilzd.Eng. Claeuz. 1941 33 1270. 9 Kunin and IlcGarvey l n d . En:. Chenz. 1949. 41 1265. lo Samuelson Diss. (Valhallavagen Sweden 1944). Bauman Abstr. Div. Call. Chenz. (Amer. Cltem. SOC.) 1949 14. SERIES A Decrease in Resin Volume Swelling dry+wet Moisture Capacity P = O'I Relative Degree of Cross- Cl' O f iO % Total Exch. Capacity Capacity m.-equiv./g. m. -equiv. /g. L4pparent Density ;. (dry)/ml. !J- ;!Iao 1- linking O/ /O % I 0.15 0.19 0.26 0.13 3 - .- 2 3 3'2 3'1 3'1 2.6 2'3 P =oo'ol / O 26 I 0 0 0 74 7 6 66 81 540 180 I 60 2-6 F' % 44 56 59 52 66 / O 64 23 I 0 5 4 51 67 7 3 74 7 6 40 5 3 0 0 so 83 3'2 1.8 0 ' 1 Decolor .Cap. Beet Sugar g. sugar/g. 22 j I2 j 220 525 150 3'9 7'7 7'2 6.3 6.4 5.1 4 s 3 4 I 2 S 0.14 0'20 0.26 0.32 - I75 I I0 ROBERT KUNIN AND ROBERT J. MYERS hl- varying the degree of cross-linking and Series B a similar set of resins in which the basic groups were of the weak-base type. The total exchange capacities and exchange equilibria were determined in the manner described by Kunin and Myers * and by Kunin and McGarvey.O The total capacities of the strong-base resin were also determined by the extent of splitting of sodium chloride. The percentage swelling was determined by a measurement of the difference in volume between two equal weights of resin one of which was put in the chloride form the other in the hydroxyl form.The apparent densities and moisture contents were determined as described The capacity for large ions and decolorizing capacity were determined in column experiments. The capacity for decolorizing beet molasses \vas determined by passing a 15 Brix sugar solution over a 20 ml. bed of resin at a rate of 0.067 ml. per ml./min. The end-point was taken as 20 yo colour leakage. The capacity for penicillin was determined by passing dilute solutions (usually about 0-01 N) of the compound over a 2 ml. bed in a 2 ml. pipette a t a slow rate with contact times from three to four minutes. Appropriate chemical tests were employed to detect breakthrough.The moisture capacity was determined by equilibrating the salt (chloride) form of the resin in water and determining the moisture content of the equili- hrated resin. Results physical cliarac teristics of both resin series are described in Table I in which the variation of moisture capacity apparent density and swelling characteristics are reported as a function of the degree of resin cross-linking. It is quite obvious that the ability to imbibe moisture and swell decreases markedly upon increasing the degree of cross- linking. Similarly the degree of resin de-swelling upon electrolyte addition Physical Characteristics .-The decreases as the degree of cross-linking increases. CAPACITIES.-The total capacities of the two resin series are shown in Table I.An increase in the degree of cross-linking is accompanied by a decrease in total exchange capacity for both resin series. However i t is interesting to note that the weak-base resins decrease more rapidly than the corresponding strong bases. The capacities for penicillin G as determined by a column procedure are Chemical CharaCteriStiCS.-EXCHANGE shown as a function of degree of cross-linking in Table I. The adsorption of penicillin was studied using 0.01 N solutions of pure penicillin G. Nessler’s reagent was used to detect leakage and breakthrough capacity. The much greater capacity for penicillin in the porous exchangers is shown clearly by the data. As the degree of cross-linking increases the fraction of the total resin capacity that is available to penicillin decreases markedly.In fact a t high degrees of cross-linking practically no adsorption of the large penicillin anion can be detected. Similar results were obtained with the Series B resins for the adsorption of the weakly acidic coloured bodies in beet molasses. EXCHANGE EQUxLIBRIA.-The effect of the degree of cross-linking upon anion- exchange equilibria was determined solely with the Series A strong-base type resins. The extent of exchange of various halogen anions with the hydroxyl ions of the resin A series is described in Table I. It is quite obvious that the degree of cross-linking has a marked effect upon the exchange equilibria. The results appear to indicate that as the degree of cross-linking increases the equilibrium constant approaches unity.Discussion The results indicate another parameter of importaiice in chromatography with ionic adsorbents namely the degree of porosity of the resin structure inaddition t o those of pH and basicity or acidity of the adsorbent is to bc considered in applications of ion exchangers. By suit able modifications of presently available ion-exchange resins the exchange capacity for large ions can be increased. Such ‘‘ porous ” exchangers are modified also with respect to swelling and hydraulic characteristics but certain sacrifices can CALCIUM-SODIUM ION EXCHANGE I18 be made in these matters to achieve appreciably improved capacities for large anions. Finite limitations on the size and shape of large ions that may be adsorbed are to be expected and will be determined by the structure characteristics of the resin particle.As the degree of cross-linking is increased the ability to imbibe water and swell is decreased thereby limiting the penetration and exchange capacity towards large ions. These results are quite analogous to the molecular-sieve effect noted by Barrer l2 for the adsorption of gases by the zeolites and may therefore be considered an “ ionic-sieve ” effect. The lack of ionic selectivity as the internal pore space increased due to a low degree of cross-linking has been noted by Samuelson lo and Wa1t0n.l~ Walton attributes this effect to the fact that the environment inside of the resin particle tends to approach that of the external solution thereby decreasing the difference between the adsorption of the various ions.Conclusions .-“ Porous ” modifications of conventional anion-exchange resins have been prepared and examined for their exchange and physical characteristics. It has been found that as the degree of cross-linking decreases the ability to adsorb large anions increases. I t has also been noted that decreasing the degree of cross-linking markedly affects the exchange equilibria decreasing ion selectivity. The resins examined were synthesized by Dr. Charles McBurney and Dr. Fred Boettner in our Philadelphia laboratories. Miss Ruth Barry and Mr. Frank McGarvey assisted in the determination of exchange capacities. Rohm and Haas Company Plziladelphia Pennsylvania U.S.A. EXCHANGE EQUILIBRIA IN ANION-EXCHANGE RESINS : POROUS EXCHANGERS BY ROBERT KUNIN AND ROBERT J.MYERS Received 22nd August 1949 The exchange equilibria of strong-base and weak-basc anion-exchange resins prcpared with various degrees of cross-linking were examined t o determine the effect of this quantity upon the behaviour of the products. By suitable improvements in “ porosity ” the exchange capacity for large molecules such as penicillin or coloured bodies in beet molasses can be increased. The swelling and hydraulic characteristics are also affected by changes in cross-linking. The results are quite analogous t o the molecular-sievr effect noted for the adsorption of gases by zeolites and may therefore be considered an ‘‘ ionic-sieve ” effect. Chromatographic techniques that have employed ion-exchange substances as adsorbents have aided considerably in the analysis of rare earth mixtures,l solutions of amino acids,2 purine bases nucleic acids 3 and other systems, The principles involved in these separations are similar to those involved in the chromatographic analysis of non-ionic constituents ; however, superimposed on these principles are several added ones deriving from the ion-exchange process.Notably successful separations have been achieved through the application of variations in elution agents degree of column loading flow-rates during development and particle size of adsorbent. Differences in exchange equilibria have been called into service through searches for adsorbents with specific adsorptive properties or investigations 1 Boyd Schubert and Adamson J .Awcr. Chem. Soc. 1948 69 2818. Winters and Kunin Igzd. Eizg. Cheni. 1949 41 460. Cohn Science 1949 109 377 ROBERT KUNIN AND ROBERT J. MYERS into the effects of complexing agents and of pH on elution. Further extension of the application of exchange equilibria principles in chromatographic analysis has been made through the use of the newer synthetic resin exchange adsorbents that contain various functional groups and possess various degrees of selectivity based upon differences in ionic size. Equilibria in anion exchangers have been studied extensively by Wik-lander Myers Eastes and Urquhart Bhatnagar and co-workers Griess-bach and Kunin and Myers8 Most of these studies were conducted on weak-base type anion exchangers that is the synthetic resins employed were based upon alkylene polyamines or aromatic diamines.Kunin and McGarvey examined the equilibria relationships in a strong-base or quaternary-t ype anion exchanger. Titration curves of anion exchangers with strong acids clearly indicate the strength of the basic group. The order of decreasing exchange ability differs for the weakly basic resins as compared with the strong base chiefly with respect to the position of the hydroxyl ion. In the case of the strongly basic resin the order was found by Kunin and McGarvey 9 to be as follows for the strong-base exchanger Amberlite IRA-400 citrate > sulphate 7 oxalate > iodide > nitrate > chromate > bromide > thiocyanate > chloride >formate>hydroxyl>fluoride>acetate.The order of replacing ability is determined by means of “ symmetry ” studies in which the percentage exchange effected is plotted against the symmetry value or milli-equivalents of ion added per milli-equivalent of dry resin. Exchange equilibria relation-ships such as the Freundlich Langmuir or Rothmund-Kornfeld equations hold only over certain ranges of concentration and are best applied when ions of similar size and charge are involved. In general symmetry plots are readily obtained and are more meaningful in discussions of practical problems in chromatography. The availability of both weakly and strongly basic anion exchangers offers a range of operations to the laboratory investigator. Such an operating range is applicable mainly in cases of relatively small ions.When larger molecules such as complex colour bodies penicillin and ascorbic acid are under consideration it is desirable to employ modified exchangers that exhibit greater capacities for large ions. Recent advances in synethtic resin tech-nology have led to the development of ion-exchange resins with different degrees of “porosity” such that larger ions may be adsorbed and eluted effectively. Since the successful application of chromatographic techniques requires operation at or near equilibrium conditions the porosity of the exchanger becomes a vital factor since rates of diffusion throughout the resin particle vary with both the porosity of the resin and size of the ion. Samuelson l o and Bauman l1 have studied the effects of swelling and porosity upon the equilibria of cation exchange in sulphonic acid cation-exchange resins.It is an object of this paper to present some preliminary data on the performance of laboratory preparations of porous anion exchangers of the weak-base and strong-base type. Experimental The resins selected for this study were modifications of the Amberlite IRA-400 in which the density was altered (Series A) studied by Kunin and McGarvey 4 Wiklander -4 i m . Roy. Agric. Colt?. Swedcw 1946 14 I. 5 Myers Eastes and Urquhart Ilzd. Eng. Claeuz. 1941 33 1270. Bhatnagar et al. J . Indian C h e w SOC. 1941 18 447. 7 Gricssbach 2. Ver. Detitsch. Cheudzer. Beih. 1939 31 I. 8 Kunin and Myers J . Anzer. Chenz. Sos. 1947 69 2874. 9 Kunin and IlcGarvey l n d . En:. Chenz. 1949. 41 1265.lo Samuelson Diss. (Valhallavagen Sweden 1944). Bauman Abstr. Div. Call. Chenz. (Amer. Cltem. SOC.) 1949 14 Relative Degree of Cross-linking I 2 3 4 s L4pparent Density ;. (dry)/ml. 0.15 0.19 0.26 0.13 3 -.-I 2 3 Decrease in Resin Volume !J- ;!Iao P = O'I P =oo'ol 64 40 26 23 5 I I 0 3 0 0 0 0 0 / O % / O 5 4 4 S Swelling dry+wet O f iO Moisture Capacity % Total Exch. Capacity m.-equiv./ SERIES A 540 22 j 180 I 60 I2 j 3'2 3'1 3'1 3'9 2-6 0.14 0.26 0.32 0'20 -525 7'7 220 7'2 I75 6.3 150 6.4 I I0 5. ROBERT KUNIN AND ROBERT J. MYERS hl- varying the degree of cross-linking and Series B a similar set of resins in which the basic groups were of the weak-base type.The total exchange capacities and exchange equilibria were determined in the manner described by Kunin and Myers * and by Kunin and McGarvey.O The total capacities of the strong-base resin were also determined by the extent of splitting of sodium chloride. The percentage swelling was determined by a measurement of the difference in volume between two equal weights of resin, one of which was put in the chloride form the other in the hydroxyl form. The apparent densities and moisture contents were determined as described The capacity for large ions and decolorizing capacity were determined in column experiments. The capacity for decolorizing beet molasses \vas determined by passing a 15 Brix sugar solution over a 20 ml. bed of resin at a rate of 0.067 ml.per ml./min. The end-point was taken as 20 yo colour leakage. The capacity for penicillin was determined by passing dilute solutions (usually about 0-01 N) of the compound over a 2 ml. bed in a 2 ml. pipette a t a slow rate with contact times from three to four minutes. Appropriate chemical tests were employed to detect breakthrough. The moisture capacity was determined by equilibrating the salt (chloride) form of the resin in water and determining the moisture content of the equili-hrated resin. Results Physical Characteristics .-The physical cliarac teristics of both resin series are described in Table I in which the variation of moisture capacity, apparent density and swelling characteristics are reported as a function of the degree of resin cross-linking.It is quite obvious that the ability to imbibe moisture and swell decreases markedly upon increasing the degree of cross-linking. Similarly the degree of resin de-swelling upon electrolyte addition decreases as the degree of cross-linking increases. Chemical CharaCteriStiCS.-EXCHANGE CAPACITIES.-The total capacities of the two resin series are shown in Table I. An increase in the degree of cross-linking is accompanied by a decrease in total exchange capacity for both resin series. However i t is interesting to note that the weak-base resins decrease more rapidly than the corresponding strong bases. The capacities for penicillin G as determined by a column procedure are shown as a function of degree of cross-linking in Table I. The adsorption of penicillin was studied using 0.01 N solutions of pure penicillin G.Nessler’s reagent was used to detect leakage and breakthrough capacity. The much greater capacity for penicillin in the porous exchangers is shown clearly by the data. As the degree of cross-linking increases the fraction of the total resin capacity that is available to penicillin decreases markedly. In fact a t high degrees of cross-linking practically no adsorption of the large penicillin anion can be detected. Similar results were obtained with the Series B resins for the adsorption of the weakly acidic coloured bodies in beet molasses. EXCHANGE EQUxLIBRIA.-The effect of the degree of cross-linking upon anion-exchange equilibria was determined solely with the Series A strong-base type resins. The extent of exchange of various halogen anions with the hydroxyl ions of the resin A series is described in Table I.It is quite obvious that the degree of cross-linking has a marked effect upon the exchange equilibria. The results appear to indicate that as the degree of cross-linking increases the equilibrium constant approaches unity. Discussion The results indicate another parameter of importaiice in chromatography with ionic adsorbents namely the degree of porosity of the resin structure inaddition t o those of pH and basicity or acidity of the adsorbent is to bc considered in applications of ion exchangers. By suit able modifications of presently available ion-exchange resins the exchange capacity for large ions can be increased. Such ‘‘ porous ” exchangers are modified also with respect to swelling and hydraulic characteristics but certain sacrifices ca I18 CALCIUM-SODIUM ION EXCHANGE be made in these matters to achieve appreciably improved capacities for large anions.Finite limitations on the size and shape of large ions that may be adsorbed are to be expected and will be determined by the structure characteristics of the resin particle. As the degree of cross-linking is increased the ability to imbibe water and swell is decreased thereby limiting the penetration and exchange capacity towards large ions. These results are quite analogous to the molecular-sieve effect noted by Barrer l2 for the adsorption of gases by the zeolites and may therefore be considered an “ ionic-sieve ” effect. The lack of ionic selectivity as the internal pore space increased due to a low degree of cross-linking has been noted by Samuelson lo and Wa1t0n.l~ Walton attributes this effect to the fact that the environment inside of the resin particle tends to approach that of the external solution thereby decreasing the difference between the adsorption of the various ions. Conclusions .-“ Porous ” modifications of conventional anion-exchange resins have been prepared and examined for their exchange and physical characteristics. It has been found that as the degree of cross-linking decreases the ability to adsorb large anions increases. I t has also been noted that decreasing the degree of cross-linking markedly affects the exchange equilibria decreasing ion selectivity. The resins examined were synthesized by Dr. Charles McBurney and Dr. Fred Boettner in our Philadelphia laboratories. Miss Ruth Barry and Mr. Frank McGarvey assisted in the determination of exchange capacities. Rohm and Haas Company, Plziladelphia, Pennsylvania U.S.A

ISSN:0366-9033    

DOI:10.1039/DF9490700114

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

I18 FULLY SWOLLEN ALGINATE GELS AS PERMUTITES KINETICS OF CALCIUM-SODIUM ION EXCHANGE BY J. L. MONGAR AND A. WASSERMANN Received 20th July 1949 On addition of sodiuni chloride solution to fully swollcil cylindrical calcium alginate gels a stationary sodium chloride concentration in the gel phase is established before appreciable conversion of the calciuln alginate into the sodium salt has taken ylacc. These observations together with the results of other experiments make i t possible to estimate approximate initial values of third-order velocity coefficients characterizing a homogeneous ionic replaceincnt reaction not a diffusion step. The overall rate of previously investigated ion-exchange reactions appears to depend on the rate of a diffusion stage.l Some authors interpreted the results of their kinetic measurements as indicating that the ion exchange itself is the slowest step ; but so far as we are aware thesc experiments 1 See e.g, Boyd Adamson and Neyers J .Anier. Chem. Soc. 1947 69. 2836. 2 For references see Duncan and Lister Qunrf. Rcv. 1948 2 307. J. L. MONGAR AND A. WASSERMANN 119 were done with adsorbents of non-uniform shape and no attempt was made to find out how the velocity coefficients computed from the analytical data are influenced by the surface/volume ratio of the exchange material. The work now to be described deals with the following reversible meta- thesis calcium alginate + sodium chloride + k mixed sodium-calcium alginate -/- calcium chloride. . * k 2 Ca" + z Na' + k z Na' + Ca" k2 where the species with the subscript g are confined to the gel phase viz.they are ' I non-permeant," and k and k2 are third order velocity coefficients relating respectively to the forward and to the reverse process. The results of the experiments are compatible with the assumption that the initial rate of the forward reaction relates to the homogeneous ion exchange itself not to a diffusion step and approximate numerical values of the velocity coefficient k are estimated. The measurements were done with cylindrical alginate gels of high water-content the surface-volume ratio of the test pieces being varied in the range 34-140 cm.'l. Temperature Gel " C 20 20 . . CaCl I8 0 (1) The chloride ions do not participate stoichiometrically and therefore the cation exchange proper can simply be represented by (14 Diffusion coef.x 10' (ern. x sec.-I) I8 2 6.5 4 0.7 8 5 2 8.0 & 0.8* 7 i- It _ _ _ _ ~ I * This value was calculated from observations in which only the earliest stage of the diffusion of calcium chloride from the gel to the solution mas taken into account. . Initial conc. of Permeant Electrolyte in Gel (g . -mole per 1. Gel) 2-30 1-00 1 0'500 Results The alginate gels containing gz yo of water were prepared surface-dried and analyzed by methods described in a paper which we hope to publish in the J . Chem. SOG. (referred to below as Part I). Table I shows the numerical values of diffusion-coefficients D of some simple electrolytes dissolved in these gels.* These measurements were made by using a method similar to that outlined by Eggleton Eggleton and Hill,3 the D value in the last line being calciilated with the help of an equation taken from R a r s e ~ .~ TABLE I DIFFUSION COEFFICIENTS OF SIMPLE ELECTROLYTES IN ALGINATE GELS Permeant Electrolyte Surface/ volume Ratio of Gel (cm.-l) NaOH Na2C03 5 60 i 3 t This value relates to the latest stage of the diffusion. . . 18 3 Proc. ROJ~. SOC. 13 1928 103 620. Diffusion in and Through Solids (Cambridge 1941) ; cf. also Hill PYOG. ROJI. SOG. B 1928 104 39- * Sodium alginate is soluble in water but if the sol is 2 N with respect to sodium hydroxide or carbonate a gradual transition into a gel occurs.CALCIUM-SODIUM ION EXCHANGE TABLE I1 CHLORIDE CONCENTRATION IN THE GEL g. equiv. ;< 104 Rate of Flow of NaCl Soln. (cm."min.) Weight of Surface- dry Fibre (m&) Chloride Calciuni Released Released from Fibre* from Fibrei Stationary NaCl concn. in Gel (g . -mole/ 1. Gel) I 1'0 ~ I20 A coniparisoii with the figures listed in Landolt-Boer~zsteirz's Tables shows that the D values of NaOH Na,CO and CaC1 in water are similar to those listed in the last column of Table I ; this is of interest because the diffusion coefficients of other simple electrolytes dissolved in a less highly swollen cation eschange material are 5-10 times smaller.1 In carrying o u t the kinetic measurements fully swollen calcium alginate fibres of diameter varying between 0.24 and 1.2 mm.were hung over a hook and rinsed first with water and then with sodium chloride solution of the same temperature thereby obtaining an effluent el. After known time intervals the flow of the sodium chloride solution was interrupted the fibre was quickly surface-dried weighed and rinsed with water until permeant electrolytes were removed from the gel the effluent thus obtained being designated e,. The two effluents were analyzed for calcium chloride determinations in e2 being RESULTS 01; EXPERIMENTS WITH FULLY SWOLLEN CALCIUM RLGINATE FIBRES OF 0.48 MM. DIAM. AT zoo SHOWING THE RAPID ATTAINMENT OF A STATIONARY SODIUM Time of Treatment of Calcium ,Ilginate Fibres with N NaCl Soln.(sec . ) 5 1 I 40 IS 178 I79 1 1.88 1.4 1.8 1'0 & 0'1 ' I 0.03 0.03 0.07 0.07 0.07 40 40 IS IS8 j * . ' * . . 10 I 0 . . I0 . . 20 . . I . . I I . . . . ' I ! I 186 180 1 I 86 1 0.14 0'20 0'1 0.8 f 0'1 1'0 f 0'1 0.9 & 0'1 1'1 1'0 f * 0'1 0'1 1'1 f 0'1 1.8 1.6 2'1 2'0 2'0 2'3 * Into effluent e, during the final rinsing with water. t Into effluent el during the treatment with the NaCl solution ; IOO yo ion exchange corresponds to a release of 0-72 x IO-* g.-equiv. calcium. + Using the appropriate equation for cylindrical diffusion (sec ref.,4 a diffusion coeffi- cient of 12 x 1o-O cm.2 sec.-l and assuming that the NaCl diffused into the gel is not used up i t can easily be shown that after 10 sec.the NaCl concentration should be about 0.9 g.-molc/l. gel. TABLE 111 KINETICS OF CALCIUM-SODIUM 103 EXCII.%NGE TYPICAL RESULTS Temperature 37 $1 ; Initial diam. of fully swollen calcium alginate fibre 1-15 mm. ; initial conccntration of non-pernieant calcium in gel 0-2 g.-mole/l. fully swollen gel ; rate of flow of sodium chloride solution 5 cm.3/min. ; stationary sodiuni chloride con- centration in gel o*oSoo g.-mole/l. fully smollen gel. 40 40 40 . . I' 30 90 60 _.__ 192 159 192 I ! 0.35 0.42 ___ I Goc ' 840 1'1 4 0'1 I I020 I Time of treating of fibre with sodium chloride solution (sec.) .. . . ISO I 300 :& Non-permeant calcium converted into CaC1 (which was analyzed in 35 4'2 . . . . . . . . effluent el) . . . . . . 101 x I., :.' (A? (SCC.'). . . . Rauman and Eichhorn J . Anzer. C h e w SOL 1947 69 2832. J. L. MONGAR AND A. WASSERMANN is unity. __ Run Temp. also carried out ; it was possible therefore to estimate the sodium chloride concentration in the alginate gel and to establish that the quantity of calcium chloride in e2 was negligibly small. The results of typical measurements are in Table I1 and 111. The figures in the last line of Table I11 were obtained by integration of the following rate equation 121 (2) i; (12 g.-mole-2 sec. -l) " C Initial Diam.Fibre* (mm.) where a is the stationary sodium chloride concentration in the gel b and b - x are the concentration values of the non-permeant calcium respectively a t zero time and time t. The latter concentration could be estimated by taking into account the figures in the second line of Table I11 and the results of control tests (Part I) which showed that no significant storage of CaCl in the fibre occurs that there is no membrane hydrolysis and that the ratio No. of g.-equiv. non-permeant calcium removed from gel as result of ion exchange No. of g.-equiv. non-permeant sodium incorporated into gel KINETICS OF CALCIUM-SODIUM ION EXCHANGE SUMhlARY OF RESULTS Stationary NaCl conc. Rate of Flow of NaCl Soln. - . No.- - ~ (0.28 I 3 2 4 5 6 20 0'45 0.45 7 9 8 I 0 I1 I2 13 I4 I5 16 0.28 0.40 ! 5 5 0'43 0.55 0.55 0.5 5 0.96 0.28 0.50 1'1 1'1 0.28 0.5 5 1'1 1'0 1'1 0'45 30 37 20 20 1'1 I 7 I8 I9 20 2I$ solution containing carbonate-free sodium hydroxide. I 0 20 1-00 1-00 1'00 1-00 0.800 2 2 2 2 2 10-20 40-45 0.800 0.400 0.130 0'100 0'100 0'100 0.08 0.08 0.07 0'10 0.08 0.08 0.08 * The diameter ZY of all fibres was much smaller than their length ; the surface/ volume ratio is therefore 2 / ~ . t Each figure listed in this column was obtained from the result of tests similar to those indicated by the data in Table 11.In these experiments the calcium alginate was rinsed with a sodium chloride Graphical extrapolation of kla2 t o small conversion ratios leads to an initial value of 8-5 0.5 x 10-4 sec.-1 corresponding to an initial velocity coefficient of hy = 0.13 & 0.04 1.2 g.-mole-2 sec.-1 Similar experiments were carried out under conditions indicated by the figures in column 2-6 of Table IV the relevant initial velocity coefficients being listed in the last column. The accuracy of these hy values is & 30 % in most runs but in a few experi- ments e.g. No. 15 it is &- 45 yo. Such errors are considerably larger than 5 ___ -~ 30 2 40 5 5 5 45 5 0'10 0.08 dx/dt = K,a'(b -z) TABLE IV (cm.a/min .) in Gelt (g .-mole/ 1. Gel) -__- 1'10 2 2 2 ~~ Soln. * 0.0 I 6 0.0 I 4 0.014 0'020 0.014 0.0 I 6 0.016 0.030 0.14 0.16 0.16 0.1s 0.16 0-16 0-12 0.14 pH of 5-6 I 0 I 0 0'12 0.16 0.13 0'1 2 0'12 122 CALCIUM-SODIUM IOX EXCHANGE thosc 01 most published rate measurements relating t o one-phase systems or to heterogeneous reactions involving less highly swollen adsorbents. It should also be noted t h a t the fibre diameter could only be varied within a rather narrow range. Discussion The kinetic significance of velocity coefficients deduced from eqn. ( 2 ) is based on the validity of the following assumptions (i) the diffusion of NaCl from the outside solution into the gel is fast compared t o the rate of formation of CaCl ; (ii) the volume of the alginate gel remains constant throughout the course of observation; (iii) the diffusion of CaC1 from the gel to the outside solution is fast compared to the rate of the reverse reaction (I).The figures in Table I1 and the results of similar tests show that a stationary NaCl concentration in the gel is rapidly attained before an appreciable conversion of calcium alginate has occurred ; and it appears therefore that as a first approximation (i) is valid. It is knownJ6 on the other hand that the conversion of calcium alginate into the mixed salt is accompanied by considerable swelling so that the second assumption is not compatible with the actual facts.The swelling of the alginate gels must give rise to a decrease of b - x in eqn. ( 2 ) which in turn should bring about an increase of k,n2. The figures in the last line of Table 111 FIG. I .--Concentration dependence of velocity and equilibrium coefficients. Tcmpera - ture 20" (.4) black circles left ordinate ; (u) open circles right ordinate. and the results of other experiments show however that the k,a2 values decrease with increasing conversion and we believe that under these conditions the influence of the back reaction over-compensates the effect due to the swelling. The instantaneous rate of the reverse reaction (I) which is probably proportional to the square of the concentration of the non-permeant sodium ions must decrease rapidly with decreasing conversion into the mixed alginate and it is reasonable therefore to postulate the validity of (iii) for the initial stage of the interaction between calcium alginate and sodium chloride.Such a hypothesis is in accordance with experiments showing that the ki values are not detectably dependent on the rate of flow of the NaCl solution or on the fibre diameter (cf. Runs Nos. 1-7 and 12-16 of Table IV). If it should be possible in future t o carry out more accurate measurements it may be found that the initial velocity coefficients increase somewhat with increasing surface/volume ratio of the test piece. Observations of this kind would not invalidate any essential conclusions to be derived from this work ; but the correct initial velocity coefficients would have to be estimated by plotting the fibre diameter r against A,"") and by extrapolating to Y = 0.6 hlongar and Wassermann J . Physiol. 1947 106 32P. MacA4rthur Mongar and Wassermann Nrrtzzre 1949 164 I 10. J. L. MONGAR AND A. WASSERMANN I 2.7 The k," values in Table IV increase with decreasing NaCl concentration as shown by curve A in the Figure. The trend can be taken to be a solvent effect the dielectric constant and other properties of a medium being dependent on the salt concentration. Curve B represents the concentration dependence of the equilibrium coefficient k,/k, the later values having been determined by measurements done in a static system (Part I). These graphs provide confirmatory evidence for the correct interpretation of the kinetic measurements it being well known that in many cases an alteration of chemical conditions gives rise t o similar changes of rate and equilibrium coefficients.The association of certain simple electrolytes involves covalent forces 8 and it is not impossible that similar effects play a role to a certain extent in the salts of alginic acid. Moreover the solvation of permeant sodium or calcium ions is probably not identical with that of these cations if they are stoichiometrically combined with the alginate residues thereby being held in proximity of a highly charged poly-anion. The ion exchange may be accompanied therefore by an alteration of both covalent and solvation forces and this should give rise to an appreciable activation energy or to a small probability of reaction.The observed temperature dependence of the velocity coefficient k; (see Runs Nos. 16-19 in Table IV) is not compatible with an overall activation energy exceeding a few kcal. It has to be taken into account however that the calcium-sodium ion exchange like other homogeneous ionic processes of the stoichiometric type 3 reactants-+product can take place as a result of consecutive bimolecular processes. These can bc formulated in the present case as follows \ Ca" + Na' +- k; {Ca**g}+Na*g) kI (3) (Ca,) + Na' k" -+ Ca" + Na' where {Ca",) is the symbol for an intermediate in which the non-permeant calcium is combined with only one carboxylate grouping of an alginate residue ; k ki and k" are bimolecular velocity coefficients and the significance of the other symbols is the same as in (ra).It can be shown with the help of the stationary state approximation that k; is proportional to k"ki/kl, provided k!,[Na',] > K"[Na']. The overall activation energy E of the forward reaction (I) would be given therefore by E = E" -t Ei - EL, where the various terms on the right-hand side relate to the corresponding velocity coefficient in (3). Thus a small overall activation energy does not necessarily imply a small activation energy of elementary bimolecular replacement steps. J ) . * The work described in this paper forms part of an investigation into the influence of ion exchange on the molecular shape of chain-like poly- electrolytes. h grant from the Department of Scientific and Industrial Research is gratefully acknowledged.Pharmacology Department afcd the Sir William Ramsay nnd Ralph Forster Laboratories U ~ I iversity Coll ege London. 7 For a theoretical treatment of such correlations see e.g. Evans and Polanyi Twits. Faraday SOG. 1935 31 492. 8 Bell and Prue J . Chem. SOG. 1949 362 where references to previous work will be found. E.g. Wejss J . Ckewz. SOC. 1944 309. I18 FULLY SWOLLEN ALGINATE GELS AS PERMUTITES : KINETICS OF CALCIUM-SODIUM ION EXCHANGE BY J. L. MONGAR AND A. WASSERMANN Received 20th July 1949 On addition of sodiuni chloride solution to fully swollcil cylindrical calcium alginate gels a stationary sodium chloride concentration in the gel phase is established before appreciable conversion of the calciuln alginate into the sodium salt has taken ylacc.These observations together with the results of other experiments make i t possible to estimate approximate initial values of third-order velocity coefficients characterizing a homogeneous ionic replaceincnt reaction not a diffusion step. The overall rate of previously investigated ion-exchange reactions appears to depend on the rate of a diffusion stage.l interpreted the results of their kinetic measurements as indicating that the ion exchange itself is the slowest step ; but so far as we are aware thesc experiments Some authors 1 See e.g, Boyd Adamson and Neyers J . Anier. Chem. Soc. 1947 69. 2836. 2 For references see Duncan and Lister Qunrf. Rcv. 1948 2 307 J. L. MONGAR AND A.WASSERMANN 119 were done with adsorbents of non-uniform shape and no attempt was made to find out how the velocity coefficients computed from the analytical data are influenced by the surface/volume ratio of the exchange material. The work now to be described deals with the following reversible meta-thesis : k , k 2 calcium alginate + sodium chloride + mixed sodium-calcium alginate -/- calcium chloride. . (1) The chloride ions do not participate stoichiometrically and therefore the cation exchange proper can simply be represented by Ca" + z Na' + z Na' + Ca" where the species with the subscript g are confined to the gel phase viz., they are ' I non-permeant," and k and k2 are third order velocity coefficients relating respectively to the forward and to the reverse process.The results of the experiments are compatible with the assumption that the initial rate of the forward reaction relates to the homogeneous ion exchange itself not to a diffusion step and approximate numerical values of the velocity coefficient k are estimated. The measurements were done with cylindrical alginate gels of high water-content the surface-volume ratio of the test pieces being varied in the range 34-140 cm.'l. Results * (14 k , k2 . The alginate gels containing gz yo of water were prepared surface-dried and analyzed by methods described in a paper which we hope to publish in the J . Chem. SOG. (referred to below as Part I). Table I shows the numerical values of diffusion-coefficients D of some simple electrolytes dissolved in these gels.* These measurements were made by using a method similar to that outlined by Eggleton Eggleton and Hill,3 the D value in the last line being calciilated with the help of an equation taken from R a r s e ~ .~ TABLE I DIFFUSION COEFFICIENTS OF SIMPLE ELECTROLYTES IN ALGINATE GELS Temperature " C Gel Surface/ volume Ratio of Gel (cm.-l) Permeant Electrolyte Initial conc. of Permeant Electrolyte in Gel (g . -mole per 1. Gel) Diffusion coef. x 10' (ern. x sec.-I) 5 60 20 . . 20 . . I8 18 0 NaOH Na2C03 3 CaCl i * This value was calculated from observations in which diffusion of calcium chloride from the gel to the solution t This value relates to the latest stage of the diffusion. 2-30 I8 2 1-00 8 5 2 8.0 & 0.8* 6.5 4 0.7 1 0'500 7 i- It I _ _ _ _ ~ only the earliest stage of the mas taken into account.3 Proc. ROJ~. SOC. 13 1928 103 620. Diffusion in and Through Solids (Cambridge 1941) ; cf. also Hill PYOG. ROJI. SOG. B , 1928 104 39-* Sodium alginate is soluble in water but if the sol is 2 N with respect to sodium hydroxide or carbonate a gradual transition into a gel occurs I20 CALCIUM-SODIUM ION EXCHANGE A coniparisoii with the figures listed in Landolt-Boer~zsteirz's Tables shows that the D values of NaOH Na,CO and CaC1 in water are similar to those listed in the last column of Table I ; this is of interest because the diffusion coefficients of other simple electrolytes dissolved in a less highly swollen cation eschange material are 5-10 times smaller.1 In carrying o u t the kinetic measurements fully swollen calcium alginate fibres of diameter varying between 0.24 and 1.2 mm.were hung over a hook and rinsed first with water and then with sodium chloride solution of the same temperature thereby obtaining an effluent el. After known time intervals the flow of the sodium chloride solution was interrupted the fibre was quickly surface-dried weighed and rinsed with water until permeant electrolytes were removed from the gel the effluent thus obtained being designated e,. The two effluents were analyzed for calcium chloride determinations in e2 being TABLE I1 RESULTS 01; EXPERIMENTS WITH FULLY SWOLLEN CALCIUM RLGINATE FIBRES OF 0.48 MM. DIAM. AT zoo SHOWING THE RAPID ATTAINMENT OF A STATIONARY SODIUM CHLORIDE CONCENTRATION IN THE GEL Time of Treatment of Calcium ,Ilginate Fibres with N NaCl Soln.(sec . ) Rate of Flow of NaCl Soln. (cm."min.) Weight of Surface-dry Fibre (m&) g. equiv. ;< 104 Chloride Released from Fibre* Calciuni Released from Fibrei Stationary NaCl concn. in Gel (g . -mole/ 1. Gel) I I j * . ~ 40 5 ' * . . IS 10 . . . . ' 40 1 20 . . I 40 ! 40 I I0 . . . . I IS I 30 . . 40 I 60 . . . . 40 90 . . . . I 0 . . 178 1 1.88 I79 I 1.4 186 ' 1.8 180 1 1.8 IS8 1.6 I 86 2'1 192 1 2'0 159 2'0 192 2'3 0.03 0.03 0.07 0.07 0.07 0.14 0.35 0.42 0'20 1'0 0'1 0.8 f 0'1 1'0 & 0'1 1'0 f 0'1 0.9 & 0'1 1'1 f 0'1 1'0 * 0'1 1'1 f 0'1 1'1 4 0'1 * Into effluent e, during the final rinsing with water.t Into effluent el during the treatment with the NaCl solution ; IOO yo ion exchange corresponds to a release of 0-72 x IO-* g.-equiv. calcium. + Using the appropriate equation for cylindrical diffusion (sec ref.,4 a diffusion coeffi-cient of 12 x 1o-O cm.2 sec.-l and assuming that the NaCl diffused into the gel is not used up i t can easily be shown that after 10 sec. the NaCl concentration should be about 0.9 g.-molc/l. gel. TABLE 111 KINETICS OF CALCIUM-SODIUM 103 EXCII.%NGE TYPICAL RESULTS Temperature 37 $1 I' ; Initial diam. of fully swollen calcium alginate fibre 1-15 mm. ; initial conccntration of non-pernieant calcium in gel 0-2 g.-mole/l. fully swollen gel ; rate of flow of sodium chloride solution 5 cm.3/min.; stationary sodiuni chloride con-centration in gel o*oSoo g.-mole/l. fully smollen gel. _.__ ___ I I I ! I Time of treating of fibre with sodium :& Non-permeant calcium converted into CaC1 (which was analyzed in chloride solution (sec.) . . . . effluent el) . . . . . . 101 x I., :.' (A? (SCC.'). . . . ISO I 300 Goc ' 840 I020 35 4'2 Rauman and Eichhorn J . Anzer. C h e w SOL 1947 69 2832 J. L. MONGAR AND A. WASSERMANN 121 also carried out ; it was possible therefore to estimate the sodium chloride concentration in the alginate gel and to establish that the quantity of calcium chloride in e2 was negligibly small. The results of typical measurements are in Table I1 and 111. The figures in the last line of Table I11 were obtained by integration of the following rate equation, where a is the stationary sodium chloride concentration in the gel b and b - x are the concentration values of the non-permeant calcium respectively a t zero time and time t.The latter concentration could be estimated by taking into account the figures in the second line of Table I11 and the results of control tests (Part I) which showed that no significant storage of CaCl in the fibre occurs that there is no membrane hydrolysis and that the ratio No. of g.-equiv. non-permeant calcium removed from gel as result of ion exchange No. of g.-equiv. non-permeant sodium incorporated into gel is unity. dx/dt = K,a'(b -z) . * (2) TABLE IV KINETICS OF CALCIUM-SODIUM ION EXCHANGE SUMhlARY OF RESULTS __ Run No.- - ~ I 2 3 4 5 6 7 8 9 I 0 I1 I2 13 I4 I5 16 I 7 I8 I9 20: 2I$ Temp. " C 20 5 5 30 37 20 20 Initial Diam. Fibre* (mm.) (0.28 0'43 0.55 0.55 0.5 5 0.96 0.28 0.40 0.28 0'45 0.45 0.50 1'1 1'1 0.28 0.5 5 1'1 1'0 1'1 0'45 1'1 ! Stationary NaCl conc. Rate of Flow in Gelt of NaCl Soln. (g . -mole/ (cm.a/min .) 1. Gel) , -__- -1'10 1-00 1-00 1'00 1-00 0.800 0.800 0.400 0.130 0'100 0'100 0'100 0.08 0.08 0.07 0.08 0.08 0.08 0.08 0'10 0'10 2 2 2 I 0 20 2 2 2 2 2 10-20 40-45 5 30 40 5 5 5 45 5 2 ~~ ___ pH of Soln. 5-6 I 0 I 0 -~ i; (12 g.-mole-2 sec.-l) 0'020 0.014 0.0 I 6 0.0 I 4 0.014 0.0 I 6 0.016 0.030 0.14 0.16 0.16 0.14 0.1s 0.16 0-16 0.16 0.13 0-12 0'12 0'1 2 0'12 * The diameter ZY of all fibres was much smaller than their length ; the surface/ t Each figure listed in this column was obtained from the result of tests similar to In these experiments the calcium alginate was rinsed with a sodium chloride volume ratio is therefore 2 / ~ . those indicated by the data in Table 11. solution containing carbonate-free sodium hydroxide. Graphical extrapolation of kla2 t o small conversion ratios leads to an initial value of 8-5 0.5 x 10-4 sec.-1 corresponding to an initial velocity coefficient of hy = 0.13 & 0.04 1.2 g.-mole-2 sec.-1 Similar experiments were carried out under conditions indicated by the figures in column 2-6 of Table IV the relevant initial velocity coefficients being listed in the last column.The accuracy of these hy values is & 30 % in most runs but in a few experi-ments e.g. No. 15 it is &- 45 yo. Such errors are considerably larger tha 122 CALCIUM-SODIUM IOX EXCHANGE thosc 01 most published rate measurements relating t o one-phase systems or to heterogeneous reactions involving less highly swollen adsorbents. It should also be noted t h a t the fibre diameter could only be varied within a rather narrow range. Discussion The kinetic significance of velocity coefficients deduced from eqn. ( 2 ) is based on the validity of the following assumptions (i) the diffusion of NaCl from the outside solution into the gel is fast compared t o the rate of formation of CaCl ; (ii) the volume of the alginate gel remains constant throughout the course of observation; (iii) the diffusion of CaC1 from the gel to the outside solution is fast compared to the rate of the reverse reaction (I).The figures in Table I1 and the results of similar tests show that a stationary NaCl concentration in the gel is rapidly attained before an appreciable conversion of calcium alginate has occurred ; and it appears, therefore that as a first approximation (i) is valid. It is knownJ6 on the other hand that the conversion of calcium alginate into the mixed salt is accompanied by considerable swelling so that the second assumption is not compatible with the actual facts. The swelling of the alginate gels must give rise to a decrease of b - x in eqn.( 2 ) which in turn should bring about an increase of k,n2. The figures in the last line of Table 111, FIG. I .--Concentration dependence of velocity and equilibrium coefficients. Tcmpera -ture 20" (.4) black circles left ordinate ; (u) open circles right ordinate. and the results of other experiments show however that the k,a2 values decrease with increasing conversion and we believe that under these conditions the influence of the back reaction over-compensates the effect due to the swelling. The instantaneous rate of the reverse reaction (I), which is probably proportional to the square of the concentration of the non-permeant sodium ions must decrease rapidly with decreasing conversion into the mixed alginate and it is reasonable therefore to postulate the validity of (iii) for the initial stage of the interaction between calcium alginate and sodium chloride.Such a hypothesis is in accordance with experiments showing that the ki values are not detectably dependent on the rate of flow of the NaCl solution or on the fibre diameter (cf. Runs Nos. 1-7 and 12-16 of Table IV). If it should be possible in future t o carry out more accurate measurements it may be found that the initial velocity coefficients increase somewhat with increasing surface/volume ratio of the test piece. Observations of this kind would not invalidate any essential conclusions to be derived from this work ; but the correct initial velocity coefficients would have to be estimated by plotting the fibre diameter r, against A,"") and by extrapolating to Y = 0.6 hlongar and Wassermann J . Physiol. 1947 106 32P. MacA4rthur Mongar and Wassermann Nrrtzzre 1949 164 I 10 J. L. MONGAR AND A. WASSERMANN I 2.7 The k," values in Table IV increase with decreasing NaCl concentration, as shown by curve A in the Figure. The trend can be taken to be a solvent effect the dielectric constant and other properties of a medium being dependent on the salt concentration. Curve B represents the concentration dependence of the equilibrium coefficient k,/k, the later values having been determined by measurements done in a static system (Part I). These graphs provide confirmatory evidence for the correct interpretation of the kinetic measurements it being well known that in many cases an alteration of chemical conditions gives rise t o similar changes of rate and equilibrium coefficients.The association of certain simple electrolytes involves covalent forces 8 and it is not impossible that similar effects play a role to a certain extent, in the salts of alginic acid. Moreover the solvation of permeant sodium or calcium ions is probably not identical with that of these cations if they are stoichiometrically combined with the alginate residues thereby being held in proximity of a highly charged poly-anion. The ion exchange may be accompanied therefore by an alteration of both covalent and solvation forces and this should give rise to an appreciable activation energy or to a small probability of reaction. The observed temperature dependence of the velocity coefficient k; (see Runs Nos.16-19 in Table IV) is not compatible with an overall activation energy exceeding a few kcal. It has to be taken into account however that the calcium-sodium ion exchange like other homogeneous ionic processes of the stoichiometric type 3 reactants-+product, can take place as a result of consecutive bimolecular processes. These can bc formulated in the present case as follows : k; Ca" + Na' +- {Ca**g}+Na*g) kI, k" (Ca,) + Na' -+ Ca" + Na', where {Ca",) is the symbol for an intermediate in which the non-permeant calcium is combined with only one carboxylate grouping of an alginate residue ; k ki and k" are bimolecular velocity coefficients and the significance of the other symbols is the same as in (ra). It can be shown with the help of the stationary state approximation that k; is proportional to k"ki/kl, provided k!,[Na',] > K"[Na']. The overall activation energy E of the forward reaction (I) would be given therefore by E = E" -t Ei - EL, where the various terms on the right-hand side relate to the corresponding velocity coefficient in (3). Thus a small overall activation energy does not necessarily imply a small activation energy of elementary bimolecular replacement steps. The work described in this paper forms part of an investigation into the influence of ion exchange on the molecular shape of chain-like poly-electrolytes. h grant from the Department of Scientific and Industrial Research is gratefully acknowledged. \ (3) J ) . * Pharmacology Department afcd the Sir William Ramsay nnd Ralph Forster Laboratories, U ~ I iversity Coll ege, London. 7 For a theoretical treatment of such correlations see e.g. Evans and Polanyi, 8 Bell and Prue J . Chem. SOG. 1949 362 where references to previous work will Twits. Faraday SOG. 1935 31 492. be found. E.g. Wejss J . Ckewz. SOC. 1944 309

ISSN:0366-9033    

DOI:10.1039/DF9490700118

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

SOME METHODS FOR EXTENDING THE SCOPE OF PARTITION CHROMATOGRAPHY €31. ALFRED A. LEVI Received r5tJ2 July 1949 Three methods are described for extendiiig partition chromatography to suhatanccs with limited solubilities in water. Although in a large majority of cases adsorption chromatography is a very convenient and efficient process partition chromatography off crs certain advantages especially for colourless substances and for quantitative work. Thus (i) Partition chromatography as described by Martin and Synge 1. offcrs scope for the separation of acids since few adsorbents have been found which will give satisfactory chroinatograms with acidic substances. (ii) Substances left as bands on the column after a partition chromato- gram are always easily eluted by simple extraction with more of the stationary phase.The quantitative elution of adsorbed material is some- t imes difficult or impossible. (iii) Conditions on a part it ion chromatograni are generally easily reproducible. This is not always the case with adsorption since it is difficult to make successive batches of adsorbents with reproducible properties. (iv) Distribution * isotherms for partition are conveniently measured in tap funnels or on columns. Hence the data required for calculations based on chromatographic theory are easily obtained. (v) Experience has shown that with silica gel as support little or 110 “ coning” occurs in carefully prepared columns and the fronts of the bands are substantially horizontal even in quite large columns.On the other hand the original method of partition chromatography is applicable only to the comparatively few substances which have at least a moderate solubility in water. The present paper reviews some methods by which this limitation can be to some extent overcome. In the first method water is replaced by another solvent immiscible or only partially miscible with the flowing phase. This method has given some excellent separations but is again severely limited by the relatively small number of solvent pairs which can be used. Examples are (i) For the separation of isomeric hexachlorocyclohexanes stationary phase nitromethane ; flowing phase n-hexane. (ii) For the separation of fatty acids stationary phase methanol ; flowing phase low-boiling liquid paraffins.In the few examples so far published isotherms for substances separated appear to be linear or nearly so. This method should therefore he worth 1 Martin and Syngc L3iocliciii. J . 1941 35 13jS. 2 Ramsay and Patterson J . Ass. 08. Agric. Clacna. 194G 09 337. 3 Kamsay and Pattersor- J . A s s . Of. Agric. C’liem. 1948 31 139 164 * I t is suggested that the terms “ distribution ratio clistribution isotherm,” ctc. be used to cover all cases where a substance distributes itself betwen two phascs whether by adsorption partition ion exchange or other mechanism. ALFRED A. LEV1 considering whenever 3 mixture with suitable physical properties is encountered. In the second method water as stationary phase is replaced by a solution of some substance which reacts reversibly with some or all of the constituents of a mixture.The most generally useful system consists of a solution of a suitable buffer as stationary phase. An immiscible solvent forms the flowing phase. By this means mixtures of acids or bases can be separated. This technique which requires no unusual practical procedure offers the following advantages (i) By the choice of a suitable solute in the stationary phase a partition chromatographic technique becomes available for substances which have a relatively low solubility in water. (ii) Another variable for the control of the rate of movement of the bands becomes available in the case of acids and bases. (iii) Extremes of acidity and alkalinity can be avoided when handling sensitive material.(iv) By the use of concentrated solutions as stationary phase relatively large quantities of material can be separated. (v) Impurities in the support such as traces of alkali or foreign metals in silica gel become of less importance in the presence of the buffer. This technique was successful in the separation of the different penicillins,4 and has also given some useful results in the analysis of alkaloid mixtures.5 It suffers from two disadvantages (i) With colourless substances indicators on the column can no longer be used as in the Martin and Synge technique. This means that observa- tions must be largely confined to the effluent solution. There are now a number of procedures available for this and in view of the flexibility of the system this is no great disadvantage.(ii) In general the distribution of substances between buffer and solvent seems to follow isotherms of the Freundlich type viz. Q = aCn. The bands obtained are unsymmetrical and with large loads much overlapping of bands may occur. It has been shown that the position and shape of a band of phenylacetic acid on such a column with ether as solvent can be predicted from chromatographic theory and from the isotherms determined in tap funnels. The difficulty of " tailing " was serious with the penicillins and in all cases chromatographic column fractionation has had to be supplemented by crystallization to obtain homogeneous material. It is worth pointing out that curved isotherms are by no means confined to buffered solutions.Many instances are known where substances are distributed between solvent and water in conformity with Freundlich distribution isotherms and in some cases the value of the exponent is quite low (ca. 0.5). This difficulty can be turned to advantage by making use of the technique of displacement development,6 since for this procedure to be successful distribution ratio must vary with concentration i.e. the isotherms must be curved. The study of displacement development on partition columns in these laboratories has brought to light some points of interest not previously discussed and these are described below. For partition chromatography isotherms can be determined by three methods (i) By simple partition experiments in tap funnels.(ii) By passing sufficient of a solution of known concentration C through 4 Levi et al. Biochem. J. 1948 43 257 262. 5 Evans and Partridge Quart. J . Plzarm. Pharmacol. 1948 21 126 ; 31 441. 6 Tiselius Arkiv. Kenzi Min. Geol. A 1943 16 I. Q o/ ‘C 0 - - aCdL-’ = (V - S ) / P dQ,/dC = aicC2l-I = (v - I/ - S)/P ?.z = (v - vo - S)/(V - S) . In practice 7; is liable to be somewhat too low with a corresponding effect on n but the procedure is convenient and sufficiently accurate for such purposes as choice of developer for a displacement chromatogram. (iii) By cletermination of specific retention volumes at a series of known initial concentrations. This is the most accurate method and for partition is convenient if some device is available for detecting the issuing zones.10 Concenfiaiin in so/venIp/7os.r (GWVA~U;~. I/> 5 100 126 EXTENSIOS OF SCOPE OF PARTITION C ~ ~ I ~ O ~ ~ A ~ a column to give iin iiicompletely developed band. From measurements of the volumes of effluent and of the concentration variation of the tail the isotherm can be calculated. For the Freundlich isotherm it is only necessary to find the retention volume V of the front of the constant coil- centration zone and the volume v at which the constant concentration zone just finishes. If Qo is the concentration in the stationary phase Vo the initial volume S the pore volume of the column and I> thtl weight of stationary phase then 540- From the family of isotherms so obtained a developer at suitable concentrd- tion is chosen and the characteristic step concentrations are read off froin thc graphs (see Fig.I) or are calculated. Thus for the Freundlich isotherin if Cu is the developer concentration and C, C, . . . . . .the final step concentrations of the components 01- log a $- (7’21 - I) log c = log a2 + (122 - I) log c2 - . . . . . . . . log l t ] ) Ql/C == Q2/C2 = . . . . . . . QD/CD + (nu - I) log C D . Thc following interesting deduction can be made from this equation. If we assume for simplicity that n is the same for all the substances we haw log C2 - log C = (log a - log az)/(iz - I). Since $2 usually has values lying between 0.5 and 1.0 it is seen that concentra- tion differences between steps may be quite large for very small differences in the values of a.In fact the only limit to the resolving power of this technique is the sensitivity to small changes in concentration of the device used to follow the issuing zones. I27 *4LFRED A. LEV1 If an approximate knowledge of the composition of the mixture to be analyzed is available the length of column occupied by the completed dis- placement chromatogram can be calculated. This is useful as a guide to the amount of charge which can be put on a given column. Experience has shown that the column should be some three to four times the above calculated length. If the front of the developer moves a distance XD when unit volume passes through the column and there are m, m2 . . . . . . . . . . milli-equivalents of the components present in the initial charge then the length of column occupied by the completed chromatogram will be XD 12 + E2 + c2 If from a previous frontal analysis or by other means x, the distance travelled per unit volume by the front of the original mixed zone before development commences is known then the volume of developer required - .. * . 1 to complete the displacement chromatogram can be calculated. FIG. 2 .-Tracings of partition chromatograms developed by the displacenlent technique. Thus assume that until displacement is complete the front of the whole system continues to move at a rate x per unit volume. When displacement is completed the front of the new zones will have moved the same distance. where V is the volume of the original solution. Hence x (Vo 4- VD) = XUI’D -1 XD [jfll/c~ + fiz2/C2 + .. . . ] Since in most cases the front of the undisplaced zones will move a distance less than x,Vu this expression will give a maximum value for VD. From what has been said above it can be seen that this technique offer- very great scope for the quantitative analysis of complex mixtures of closely related materials (e.g. homologous series position isomerides and stereo- nique of coupled filters or ‘‘ front straighteners ” ’ has not been found isomerides). By working on a small scale with narrow columns the tech- necessary in partition chromatograms (see Fig. 2). These would doubtless 7 Claesson Arkiv. Kcmi Min. Geol. A 1947 24 I. T 28 QUAKTITATIVE PAPER CHROMATOGRAPHY be advantageous on a preparative scale but this possibility has not yet been adequately explored.The great sensitivity of the above method of displacement development to small differences in partition characteristics involves some restriction on the range of substances which can be separated under any given set of conditions. Hence for a very crude mixture some preliminary separation is desirable. For this purpose partition chromatography offers another device in the use of a stationary phase consisting of a solution or suspension of some substance which reacts irreversibly with the components of a mixture.8 The chromato- gram which results from passing a mixture through such a column resembles a frontal analysis in that a number of zones are formed equal to the number of substances present.Each new zone from the bottom to the top of the column contains a new component (in addition to those already present) in the order of increasing afhity for the stationary phase. The picture differs from frontal analysis in that the total molecular concentration in each step is the same (except in so far as it may be modified by physical partition between solvent and water alone). Separation cannot be complete in such a system and the method is of little quantitative value. The products being in solution are readily recovered from the silica or other support by simple extraction with water. This procedure was valuable in the case of crude penicillin especially for the earlier samples before the fermentation process was under adequate control.Large quantities of material could be considerably improved in quality without serious loss by treatment in ether solution on silica gel columns impregnated with caustic soda solution. A partial separation of bases (e.g. a crude extract of alkaloids) could obviously be similarly effected. Research Defiartment Imperial Chemical Industries Limited Dyestufs Division Hexagon House Blackley Manchestev 9. 8 Catch Cook and Heilbron Natzwe 1942 150 163. SOME METHODS FOR EXTENDING THE SCOPE OF PARTITION CHROMATOGRAPHY €31. ALFRED A. LEVI Received r5tJ2 July 1949 Three methods are described for extendiiig partition chromatography to suhatanccs with limited solubilities in water. Although in a large majority of cases adsorption chromatography is a very convenient and efficient process partition chromatography off crs certain advantages especially for colourless substances and for quantitative work.Thus : (i) Partition chromatography as described by Martin and Synge 1. offcrs scope for the separation of acids since few adsorbents have been found which will give satisfactory chroinatograms with acidic substances. (ii) Substances left as bands on the column after a partition chromato-gram are always easily eluted by simple extraction with more of the stationary phase. The quantitative elution of adsorbed material is some-t imes difficult or impossible. (iii) Conditions on a part it ion chromatograni are generally easily reproducible. This is not always the case with adsorption since it is difficult to make successive batches of adsorbents with reproducible properties.(iv) Distribution * isotherms for partition are conveniently measured in tap funnels or on columns. Hence the data required for calculations based on chromatographic theory are easily obtained. (v) Experience has shown that with silica gel as support little or 110 “ coning” occurs in carefully prepared columns and the fronts of the bands are substantially horizontal even in quite large columns. On the other hand the original method of partition chromatography is applicable only to the comparatively few substances which have at least a moderate solubility in water. The present paper reviews some methods by which this limitation can be to some extent overcome. In the first method water is replaced by another solvent immiscible or only partially miscible with the flowing phase.This method has given some excellent separations but is again severely limited by the relatively small number of solvent pairs which can be used. Examples are (i) For the separation of isomeric hexachlorocyclohexanes stationary phase, nitromethane ; flowing phase n-hexane. (ii) For the separation of fatty acids stationary phase methanol ; flowing phase low-boiling liquid paraffins. In the few examples so far published isotherms for substances separated appear to be linear or nearly so. This method should therefore he worth 1 Martin and Syngc L3iocliciii. J . 1941 35 13jS. 2 Ramsay and Patterson J . Ass. 08. Agric. Clacna. 194G 09 337. 3 Kamsay and Pattersor- J . A s s . Of.Agric. C’liem. 1948 31 139 164 * I t is suggested that the terms “ distribution ratio clistribution isotherm,” ctc., be used to cover all cases where a substance distributes itself betwen two phascs, whether by adsorption partition ion exchange or other mechanism ALFRED A. LEV1 considering whenever 3 mixture with suitable physical properties is encountered. In the second method water as stationary phase is replaced by a solution of some substance which reacts reversibly with some or all of the constituents of a mixture. The most generally useful system consists of a solution of a suitable buffer as stationary phase. An immiscible solvent forms the flowing phase. By this means mixtures of acids or bases can be separated. This technique which requires no unusual practical procedure offers the following advantages : (i) By the choice of a suitable solute in the stationary phase a partition chromatographic technique becomes available for substances which have a relatively low solubility in water.(ii) Another variable for the control of the rate of movement of the bands becomes available in the case of acids and bases. (iii) Extremes of acidity and alkalinity can be avoided when handling sensitive material. (iv) By the use of concentrated solutions as stationary phase relatively large quantities of material can be separated. (v) Impurities in the support such as traces of alkali or foreign metals in silica gel become of less importance in the presence of the buffer. This technique was successful in the separation of the different penicillins,4 and has also given some useful results in the analysis of alkaloid mixtures.5 It suffers from two disadvantages : (i) With colourless substances indicators on the column can no longer be used as in the Martin and Synge technique.This means that observa-tions must be largely confined to the effluent solution. There are now a number of procedures available for this and in view of the flexibility of the system this is no great disadvantage. (ii) In general the distribution of substances between buffer and solvent seems to follow isotherms of the Freundlich type viz. Q = aCn. The bands obtained are unsymmetrical and with large loads much overlapping of bands may occur. It has been shown that the position and shape of a band of phenylacetic acid on such a column with ether as solvent can be predicted from chromatographic theory and from the isotherms determined in tap funnels.The difficulty of " tailing " was serious with the penicillins and in all cases chromatographic column fractionation has had to be supplemented by crystallization to obtain homogeneous material. It is worth pointing out that curved isotherms are by no means confined to buffered solutions. Many instances are known where substances are distributed between solvent and water in conformity with Freundlich distribution isotherms and in some cases the value of the exponent is quite low (ca. 0.5). This difficulty can be turned to advantage by making use of the technique of displacement development,6 since for this procedure to be successful distribution ratio must vary with concentration i.e.the isotherms must be curved. The study of displacement development on partition columns in these laboratories has brought to light some points of interest not previously discussed and these are described below. For partition chromatography isotherms can be determined by three methods : (i) By simple partition experiments in tap funnels. (ii) By passing sufficient of a solution of known concentration C through 4 Levi et al. Biochem. J. 1948 43 257 262. 5 Evans and Partridge Quart. J . Plzarm. Pharmacol. 1948 21 126 ; 31 441. 6 Tiselius Arkiv. Kenzi Min. Geol. A 1943 16 I 126 EXTENSIOS OF SCOPE OF PARTITION C ~ ~ I ~ O ~ ~ A ~ Concenfiaiin in so/venIp/7os.r (GWVA~U;~. I/> 5 10 100 540-a column to give iin iiicompletely developed band.From measurements of the volumes of effluent and of the concentration variation of the tail, the isotherm can be calculated. For the Freundlich isotherm it is only necessary to find the retention volume V of the front of the constant coil-centration zone and the volume v at which the constant concentration zone just finishes. If Qo is the concentration in the stationary phase Vo the initial volume S the pore volume of the column and I> thtl weight of stationary phase then : Q o/ ‘C 0 - - aCdL-’ = (V - S ) / P , dQ,/dC = aicC2l-I = (v - I/ - S)/P , In practice 7; is liable to be somewhat too low with a corresponding effect on n but the procedure is convenient and sufficiently accurate for such purposes as choice of developer for a displacement chromatogram.(iii) By cletermination of specific retention volumes at a series of known initial concentrations. This is the most accurate method and for partition is convenient if some device is available for detecting the issuing zones. ?.z = (v - vo - S)/(V - S) . From the family of isotherms so obtained a developer at suitable concentrd-tion is chosen and the characteristic step concentrations are read off froin thc graphs (see Fig. I) or are calculated. Thus for the Freundlich isotherin if Cu is the developer concentration and C, C, . . . . . .the final step concentrations of the components, Ql/C == Q2/C2 = . . . . . . . QD/CD , 01- log a $- (7’21 - I) log c = log a2 + (122 - I) log c2 - . . . . . . . . log l t ] ) + (nu - I) log C D .Thc following interesting deduction can be made from this equation. If we assume for simplicity that n is the same for all the substances we haw log C2 - log C = (log a - log az)/(iz - I). Since $2 usually has values lying between 0.5 and 1.0 it is seen that concentra-tion differences between steps may be quite large for very small differences in the values of a. In fact the only limit to the resolving power of this technique is the sensitivity to small changes in concentration of the device used to follow the issuing zones *4LFRED A. LEV1 I27 If an approximate knowledge of the composition of the mixture to be analyzed is available the length of column occupied by the completed dis-placement chromatogram can be calculated. This is useful as a guide to the amount of charge which can be put on a given column.Experience has shown that the column should be some three to four times the above calculated length. If the front of the developer moves a distance XD when unit volume passes through the column and there are m, m2 . . . . . . . . . . milli-equivalents of the components present in the initial charge then the length of column occupied by the completed chromatogram will be c2 - . . * . 1 XD 12 + E2 + If from a previous frontal analysis or by other means x, the distance travelled per unit volume by the front of the original mixed zone before development commences is known then the volume of developer required to complete the displacement chromatogram can be calculated. FIG. 2 .-Tracings of partition chromatograms developed by the displacenlent technique.Thus assume that until displacement is complete the front of the whole system continues to move at a rate x per unit volume. When displacement is completed the front of the new zones will have moved the same distance. Hence x (Vo 4- VD) = XUI’D -1 XD [jfll/c~ + fiz2/C2 + . . . . ] , where V is the volume of the original solution. Since in most cases the front of the undisplaced zones will move a distance less than x,Vu this expression will give a maximum value for VD. From what has been said above it can be seen that this technique offer-very great scope for the quantitative analysis of complex mixtures of closely related materials (e.g. homologous series position isomerides and stereo-isomerides).By working on a small scale with narrow columns the tech-nique of coupled filters or ‘‘ front straighteners ” ’ has not been found necessary in partition chromatograms (see Fig. 2). These would doubtless 7 Claesson Arkiv. Kcmi Min. Geol. A 1947 24 I T 28 QUAKTITATIVE PAPER CHROMATOGRAPHY be advantageous on a preparative scale but this possibility has not yet been adequately explored. The great sensitivity of the above method of displacement development to small differences in partition characteristics involves some restriction on the range of substances which can be separated under any given set of conditions. Hence for a very crude mixture some preliminary separation is desirable. For this purpose partition chromatography offers another device in the use of a stationary phase consisting of a solution or suspension of some substance which reacts irreversibly with the components of a mixture.8 The chromato-gram which results from passing a mixture through such a column resembles a frontal analysis in that a number of zones are formed equal to the number of substances present.Each new zone from the bottom to the top of the column contains a new component (in addition to those already present) in the order of increasing afhity for the stationary phase. The picture differs from frontal analysis in that the total molecular concentration in each step is the same (except in so far as it may be modified by physical partition between solvent and water alone). Separation cannot be complete in such a system and the method is of little quantitative value. The products being in solution are readily recovered from the silica or other support by simple extraction with water. This procedure was valuable in the case of crude penicillin especially for the earlier samples before the fermentation process was under adequate control. Large quantities of material could be considerably improved in quality without serious loss by treatment in ether solution on silica gel columns impregnated with caustic soda solution. A partial separation of bases (e.g. a crude extract of alkaloids) could obviously be similarly effected. Research Defiartment, Imperial Chemical Industries Limited, Dyestufs Division, Hexagon House, Blackley Manchestev 9. 8 Catch Cook and Heilbron Natzwe 1942 150 163

ISSN:0366-9033    

DOI:10.1039/DF9490700124

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

T 28 PARTITION CHROMATOGRAPHY ON PAPER WITH SPECIAL REFERENCE TO QUANTITATIVE SEPARATIONS BY A. H. GORDON - Received 5tJz JuZy 1949 Recent improvements in the method of partition chromatography on paper are reviewed. By modification of the solvents used new groups of substances can be separated. I n some cases (e.g. esters of the fatty acids) reversed phase chromatograms are most convenient. Many attempts have also been made to employ the paper chromatogram for quantitative estimations ; considerable success has been achieved in the case of the amino acids which after elution from the paper can be converted into Cu salts suitable for colorimetric determination. Some of the factors making for low recoveries such as the impurities present in the filter paper are also discussed.The method of partition chromatography on paper is at present being developed in two different directions. New groups of substances are being brought within the scope of the method by the introduction of new solvent mixtures and modified techniques for the detection of the spots on the A. H. GORDON chromatogram. In addition efforts are being made to obtain quantitative recoveries of those materials whose separation is already well established. No attempt will be made here to list all the groups of substances to which the method has already been applied its scope has already been reviewed.l The purpose of the present work is to consider first the conditions necessary for successful separations on paper chromatograms and second some of the various techniques which have been used for the quantitative estimation of the separated substances.For separations to take place the following factors need to be considered. (a) Choice of Solvent.-A solvent must be found partially miscible with water giving sufficiently different partition coefficients for the various substances of the mixture to be separated. In practice this can be taken to mean differences of at least 10 %. Where no single solvent is available giving adequate differences of partition coefficient for all the components of the mixture a second solvent can be subsequently utilized by means of the two-dimensional technique. Although many substances can be separated in systems of this kind in which the solvent held stationary in the filter paper is water for substances of a very non-polar character it may be necessary to work in non-aqueous systems.Thus Boldingh ti by pre-treating filter paper with latex has been able to separate esters of fatty acids by means of methanol. In this system the rubber itself acts as the immobile solvent. Boldingh has also successfully used mixed organic solvents such as methanol benzene and 111 methanol acetone on rubber paper. He remarks that the rubber acts as a carrier for the less polar solvent. This is no doubt true but as using methanol acetone the Rf values are considerably increased compared with those obtained with methanol alone the effect of the acetone in the mobile phase must also be important. The use of aqueous systems employing mixed organic solvents as the moving phase has often proved necessary.Also the partition coefficients of the substances under separation have been modified by suitable additions to the aqueous phase. Systems giving convenient partition coefficients may be obtained in one of the following ways. (i) The Rj values can be changed by the admixture of a suitable proportion of a second organic solvent. This solvent may be completely miscible with water. So long as not more than enough is used for the formation of two phases useful results may be expected. The effect of the addition of a more polar solvent is of course to increase the water content of the organic solvent phase and thus the Rj values of all types of substance. Results of this kind were obtained by Partridge 6 who measured the Rf values of the sugars in n-butanol and in n-butanol containing 10 yo ethanol.principle was used by Martin and Synge 7 * when they increased the R values Before the introduction of partition chromatography on paper this of the acetylamino acids on silica gel chromatograms by employing pro- gressively higher concentrations of n-butanol in chloroform. A second effect of the alcohol in this system was its virtual elimination of adsorption of the acetylamino acids by the gel. Even though most substances seem scarcely to be adsorbed by filter paper in those cases e.g. the chromatography Consden Nature 1948 162 359. Hais Chem. Listy 1948 42 125. Hais and RAbek Chem. Listy 1949 43 80. 4 Consden Gordon and Martin Biochem.J. 1944 38 224. Boldingh Erperientia 1948 4 270. Partridge Biochem. J. 1948 42 238. Martin and Synge Biochem. J. 1941 35 1358. Gordon Martin and Synge Biochem. J 1943 37 79. E QUANTITATIVE PAPER CHROMATOGRAPHY I30 of inorganic ions,9 where such effects may be significant they are no doubt reduced by the admixture of more polar organic solvents. (ii) If the substances being separated are acids or bases their Rf values can be greatly influenced by adjustment of the pH of the aqueous phase As an example the effect of ammonia on chromatograms developed with phenol in decreasing the Rj values of the dicarboxylic and increasing those of the basic amino acids may be mentioned. Conversely acetic acid has been used by Lugg lo to assist the separation of certain organic acids by n-butanol.Buffer solutions l1 have also been used in a rather similar way both on filter paper and on silica gel chromatograms for the separation of the penicillins. A further advantage of the addition of buffers at least on silica gel chromatograms is that partition chromatograms can be run by displacement development and thus much larger amounts can be handled. In addition the Rf values of certain compounds can be modified by the addition of substances mainly soluble in the organic solvent phase. Thus Winsten and Eigen l2 have found that the Rf values in n-butanol of the various members of the streptomycin group could be markedly increased by the addition of z yo of 9-toluenesulphonic acid. The solvent also con- tained z yo of piperidine which served partially to de-ionize the basic groups of the streptomycins.The amount of $-toluenesulphonic acid used was such that the pH of the system (determined after the addition of an equal volume of water) was on the alkaline side. However it was noted that separations could still be obtained if so much 9-toluenesulphonic acid had been used that the pH was acid. It is interesting to compare the effect of 9-toluenesulphonic acid with that of ammonia in n-butanol. Thus if it is assumed that the increased Rj values of the streptomycins were not solely due to the piperidine then 9-t oluenesulphonic acid although an acid substance must have affected the system in the same sense as would be expected for the addition of ammonia.It seems likely that additions of this type have their effect both because they increase the polarity of the organic solvent phase and because as acids they specifically modify the ability of this phase to dissolve the substances being separated. That the effect is not due only to the former cause seems to be indicated by the observation l3 that the addition of p-toluenesulphonic acid to collidine increases the Rf values of the basic preferentially to those of the neutral amino acids. (b) Adsorption by the Paper.-If the separations are to depend only on partition effects the substances in the mixture to be separated must not be adsorbed by the paper. Fortunately few substances are strongly adsorbed by untreated filter paper. Perhaps the largest group of substances for which adsorption is likely to be important are the organic dyestuffs.Attempts to separate dyes by partition chromatography on paper often lead to resolution of the mixture. However closer examination of the system may reveal that partition effects have been secondary (as is the case if the separation can be carried out equally well with the dry solvent). Adsorption chromatography on strips of filter paper of the streptomycins using a z yo aqueous solution of ammonium chloride as solvent has been described by Horne and Pollard.l* Unfortunately no resolution of the different streptomycins was obtained. 9 Arden Burstall Davies Lewis and Linstead Nature 1948 162 691 lo Lugg and Overell NUhWt? 1947 160 87. 11 Goodall and Levi Nature 1946 158 675.12 Winsten and Eigen J . Amer. Chenz. SOC. 1948 70 3333. 13 Gordon (unpublished). 14 Horne and Pollard ,I. Buct. 1948 55 231. A. H. GORDON 13= Among large molecules the proteins are also known to be adsorbed at least from salt solutions by filter paper. These effects will not be discussed further here as they may be the subject of another contribution. The examples mentioned may be enough to illustrate the possibility of adsorption in paper chromatography. This possibility should not automatically lead to the opposite view that the separations in all systems employing solvents which do not form two phases with water must necessarily be due to adsorption. As was first suggested to the author by Dr. Synge,15 a system employing for instance pyridine and water can be considered to be a partition chromatogram because the effect of the cellulose hydroxyl groups results in the existence of a very different milieu inside the fibre from that of the moving phase.(c) Chemical Reactivity of Solvents and Paper.-The substances under separation must not react either with the paper or with the solvents used. For the amino acids and many of the other substances for whose separation paper chromatography has proved applicable these criteria have proved to be rather easily fulfilled. However the recent finding of Moore and Stein,lG that using I/z/I-n-butanol-n-propanol-o.I N HC1 on starch chromatograms 6% and 7% of aspartic and glutamic acids respectively became esterified suggests that closer and more quantitative examination of paper chromatograms in which mixtures of acids and alcohols are used might reveal similar effects.Cellulose itself fortunately shows little or no reactivity towards either the substances separable on paper chromatograms or the solvents used for their separation. The problem thus becomes that of the possible reactions which may occur between the substances under separation the solvents and impurities present in the paper. This will be further discussed in the next section. ( d ) Techniques for Detection and Estimation of Spots.-Finally for qualitative work a suitable detection technique must be available. In practice the sensitivity reliability and ease of application of such techniques may be the decisive factor in determining whether paper chromatography is to be used.Certainly the remarkable sensitivity of ninhydrin by which less than I pg. of an amino acid can just be made visible has been most useful. The small scale of the method as a whole e.g. 1-20 pg. per com- ponent of the mixture under separation for the amino acids the sugars and for the purines and pyrimidines has the advantage that it is usually possible to develop as many chromatograms as there are different detection techniques available. Particularly when dealing with complex biological fluids this type of approach by which several classes of compounds can be simultaneously investigated may be of value. If quantitative estimations are to be attempted not only must the substances be located but sufficiently accurate means must be found for their estimation either in situ or after removal from the paper.For quan- titative purposes obviously the most suitable detection techniques are those which do not permanently affect the substances to be estimated. The most generally applicable means by which this can be accomplished is certainly the examination of the spots in u.-v. light. In this way spots of purines pyrimidines amino acids and peptides can be shown up. Holidayz7 has reported that as little as 0-5-1 pg. per cm.2 of the purines or pyrimidines can be detected if u.-v. light of wavelengths between 230 and 400 mp is used the most important wavelengths being in the region 1 5 Synge (personal communication). 16 Moore and Stein J . Baol.Chem. 1949 178 53. 17 Holiday and Johnson Nature 1949 163 216. QUL4XTITATIVE PAPER CHROMATOGRAPHY 132 of 254 mp. Although only amounts of more than 3 pg. per cm.2 of the amino acids can be detected l8 the use of this technique has been continued by Woiwod.19 By this means it is possible to avoid the necessity of developing parallel guide chromatograms and also the most inconvenient process of visualizing the spots as their mercury derivatives in the case of the purines and pyrimidines.20 The only chemical method so far reported which ultimately leaves the substances unchanged on the chromatogram seems to be that of Brante.?l This author has sprayed chromatograms of bases such as amino-ethanol choline and creatine with an alcoholic solution of iodine.Brown spots are formed which disappear on being allowed to stand. Unfortunately this reagent is not very sensitive for the amino acids but as noted by Brante it may prove useful in quantitative experiments. Turning now to the various methods which have been used for the quantitative estimation of substances separated by paper chromatography these may be divided into those in which the substances are estimated on the paper usually as coloured derivatives and those in which they are first eluted and then estimated in solution. The former approach seems only to have been employed for the coloured substances formed from the amino acids by ninhydrin and for substances containing radio elements. Thus several authors 22 23 have reported that useful rough estimations can be made by visually comparing the spot strengths with those of a series of standard spots.Attempts have also been made to increase the accuracy of this method by the preliminary division of the amino acids into acidic neutral and basic types.24 A spectrophotometer was used for estimating the strengths of the spots but as yet no indication has been given of the accuracy thus obtained. A spectrophotometer has been similarly used by Bull 25 who has been at pains to standardize the conditions of colour development. Under the conditions chosen arginine serine valine glutamic acid leucine threonine alanine and lysine gave the same amount of colour per mole the probable error for a single determination being almost g yo. This result is somewhat unexpected as Moore and Stein,26 who have carefully measured the amounts of colour formed in solution in presence of a reducing agent have reported for instance that threonine forms only 81 yo as much colour as does lysine.It may possibly be the case that some of the amino acids form more nearly equivalent amounts of colour in solution than on paper. On the other hand Bull 25 found that glycine and methionine give less colour on paper than does leucine whereas Moore and Stein 26 indicate that under their conditions these three amino acids yield almost identical amounts of colour per mole. Possibly more thorough investigations of the optimum conditions for the development of the ninhydrin colour on paper may lead to some reduction in the rather large errors at present involved in this technique.Work of this kind may also explain why Pratt and Auclair 27 find that glycine and glutamic acid are the amino acids giving visible spots at minimum strength whereas Moore and Stein 26 have found lysine to give the strongest colour in solution. Where substances containing radioactive isotopes have been separated 18 Phillips Nature 1948 161 153. 1 s Woiwod Biochenz. J . (in press). 20 Vischer and Chargaff J . Bid. Chetiz. 1948 176 703. 21 Brante Nature 1949 163 651. 22 Consden Gordon Martin and Synge Biochenz. J . 1947 41 596. 23 Polson Biochim. Biophys. Acta 1948 2 575. 2 4 Block Science 1948 108 608. 25 Bull Hahn and Baptist J Amw. Chem. SOG. 1943 71 550 Z 6 Moore and Stein J . BioZ. Chem.1948 176 367. 2 7 Pratt and Auclair Scieizce 1948 108 213. A. H. GORDON I33 the detection technique i.e. the exposure of successive areas of the chromato- gram to the counter is itself suitable as a quantitative estimation. An apparatus designed for the estimation of spots containing P32 has recently been described.2B In order to estimate the amino acids by means of radio- active isotopes conditions have been worked out for their quantitative conversion into 9-iodophenylsulphonyl (pipsyl) derivatives containing P2. -4fter chromatography on paper nearly quantitative recoveries were obtained for the few amino acids tested.29 Further work on this promising method will be awaited with interest. Although techniques in which the substances are eluted from the paper prior to estimation cannot have the simplicity of those already mentioned they would seem to offer several compensating advantages.Thus once the substance is in solution it can be treated with a much wider variety of reagents than while it is still on paper. This should for instance allow better control of the conditions under which the ninhydrin colour is formed from the amino acids. Unfortunately no attempt has been reported to utilize the valuable information on this subject already obtained by Moore and Stein.26 The most successful attempts to estimate quantitatively the amino acids involve their conversion into copper complexes. Thus Martin and Mittel- inann3O used what was essentially the method of Pope and Stevens,31 in which the amino acids are allowed to react with a suspension of copper phosphate in phosphate-borate buffer.After centrifugation from the excess of copper phosphate the amount of copper in solution was estimated polarographically. No attempt was made to determine the recoveries of the amino acids after elution from the paper but since the current increased linearly with the amount used estimations were shown to be possible if several comparison chromatograms of known amounts of each amino acid were developed simultaneously with each analysis. The results in a test analysis of gramicidin S clearly substantiated the theory that this peptide consists of five different amino acids in equimolar proportions. Only in the case of one of these acids leucine did the amount found deviate more than 5 yo from that expected.The copper method has also been used by Woiwod,lS who has modified the reagent by replacing sodium borate with di-sodium hydrogen phosphate. He has thus obtained a more stable suspension lower blanks and a linear relationship between amino nitrogen and copper. The copper has been estimated colorirnetrically with sodium diethyldithiocarbamate instead of polarographically. In order to obtain sufficient accuracy for those amino acids present as the weaker components of mixtures the analyses have been conducted on larger amounts (1-50 pg. amino nitrogen) than those employed by Martin and Mittelmann.30 In this way similar accuracies seem to have been achieved and because of the convenience of colorimetry compared with polarography this is probably at the moment the method of choice.Generally Woiwod like Martin and Mittelmann has employed the com- parative method but in addition he has measured the losses sustained as an amino acid moves down the chromatogram. Thus only 86 % of the original amount of glycine could be recovered after the spot had moved 10 in. down a chromatogram developed with 40 yo 12-butanol-ro yo acetic acid-rjo yo water. In partial explanation of such low recoveries \Voiwod mentions an interesting observation to the effect that fluorescent 28 Lindberg and Hummel Arlziv. Kerizi 1949 I 17. 2 9 Iceston Udenfriend and Levy J. Amer. Chenz. Sac. 1947 69 3 I -j I . 3o Martin and Mittelmann Biocherrz. J. 1948 43 3 j 3 . 31 Pope and Stevens Biochem.J. 1939 33 1070. QUANTITATIVE PAPER CHROWITOGRAPHY 1.34 material can often be seen extending right down to the front of the solvent. He considers that this may be due to failure of the amino acid to reach equilibrium with the mobile phase. However as the usual treatment with ninhydrin which is believed to be capable of revealing smaller amounts of amino acids than are visible in u.-v. light does not show up such streaks it seems more likely that the amino acid must have been converted into a form no longer capable of giving a colour with ninhydrin. The existence of some substances in certain filter papers capable of combining with the amino acids is already known from the observations of Consden et aL4 on the formation of copper complexes of the amino acids.These are pro- gressively formed as the amino acids move down the chromatogram. These authors32 have also noted the formation on paper chromatograms of a zinc salt of cysteic acid which like the copper complexes gives a pink colour with ninhydrin. The Rf values of these metal complexes are close to those of the corresponding free amino acids but there seems no reason why there may not be other impurities in the paper which couldcombine to form complexes moving with any Rf value. The more the Rf values of such impurity amino-acid complexes differ from that of the amino acid itself the more difficult it must be to obtain quantitative recoveries from the paper. The solution of these difficulties may either be similar to that adopted by Consden et aL4 who added a reagent to the paper which prevented the action of the copper by itself combining with it ; or else it may be necessary to employ special purification of the paper.If purification of the natural paper making materials proves inadequate use may in time be made of papers formed entirely from synthetic fibres. These findings naturally raise the question of the occurrence of similar losses as other substances move down paper chromatograms. Fortunately very accurate methods of analysis exist both for the purines and pyrimidines and for the sugars after elution from the filter paper. For the purines and pyrimidines it is only necessary to measure their absorption at the appropriate wavelength in u.-v. light. Both Vischer and Chargaff go and Hotchkiss 33 have used this method without finding necessary a comparative approach similar to that employed for the amino acids.They report accuracies of j - 4 yo and do not mention progressive losses. The figures quoted suggest slight losses for those spots which have moved further down the chromatograms but are hardly sufficient to be decisivc. The need for further work on this question is apparent. Even more accurate analyses of sugars after separation on paper chromato- grams have been reported,34 and again no mention of progressive losses has been made. It may not be accidental that materials similar to the matrix of the chromatogram can be recovered with least loss. Kemiska Institutionea Karolins ka Institutet Stock Jtolm .52 Consden Gordon and Martin Biochem J. 1946 40 580. 33 Hotchkiss J. Biol. Chew. 1948 175 315. 3 4 Flood Hirst and Jones J. Clzenz. SOC. 1948 1679. T 28 PARTITION CHROMATOGRAPHY ON PAPER WITH SPECIAL REFERENCE TO QUANTITATIVE SEPARATIONS BY A. H. GORDON Received 5tJz JuZy 1949 Recent improvements in the method of partition chromatography on paper are reviewed. By modification of the solvents used new groups of substances can be separated. I n some cases (e.g. esters of the fatty acids) reversed phase chromatograms are most convenient. Many attempts have also been made to employ the paper chromatogram for quantitative estimations ; considerable success has been achieved in the case of the amino acids which after elution from the paper can be converted into Cu salts suitable for colorimetric determination.Some of the factors making for low recoveries such as the impurities present in the filter paper are also discussed. -The method of partition chromatography on paper is at present being developed in two different directions. New groups of substances are being brought within the scope of the method by the introduction of new solvent mixtures and modified techniques for the detection of the spots on th A. H. GORDON chromatogram. In addition efforts are being made to obtain quantitative recoveries of those materials whose separation is already well established. No attempt will be made here to list all the groups of substances to which the method has already been applied its scope has already been reviewed.l The purpose of the present work is to consider first the conditions necessary for successful separations on paper chromatograms and second some of the various techniques which have been used for the quantitative estimation of the separated substances.For separations to take place the following factors need to be considered. (a) Choice of Solvent.-A solvent must be found partially miscible with water giving sufficiently different partition coefficients for the various substances of the mixture to be separated. In practice this can be taken to mean differences of at least 10 %. Where no single solvent is available giving adequate differences of partition coefficient for all the components of the mixture a second solvent can be subsequently utilized by means of the two-dimensional technique.Although many substances can be separated in systems of this kind in which the solvent held stationary in the filter paper is water for substances of a very non-polar character it may be necessary to work in non-aqueous systems. Thus Boldingh ti by pre-treating filter paper with latex has been able to separate esters of fatty acids by means of methanol. In this system the rubber itself acts as the immobile solvent. Boldingh has also successfully used mixed organic solvents such as methanol benzene and 111 methanol acetone on rubber paper. He remarks that the rubber acts as a carrier for the less polar solvent. This is no doubt true but as using methanol acetone the Rf values are considerably increased compared with those obtained with methanol alone the effect of the acetone in the mobile phase must also be important.The use of aqueous systems employing mixed organic solvents as the moving phase has often proved necessary. Also the partition coefficients of the substances under separation have been modified by suitable additions to the aqueous phase. Systems giving convenient partition coefficients may be obtained in one of the following ways. (i) The Rj values can be changed by the admixture of a suitable proportion of a second organic solvent. This solvent may be completely miscible with water. So long as not more than enough is used for the formation of two phases useful results may be expected. The effect of the addition of a more polar solvent is of course to increase the water content of the organic solvent phase and thus the Rj values of all types of substance.Results of this kind were obtained by Partridge 6 who measured the Rf values of the sugars in n-butanol and in n-butanol containing 10 yo ethanol. Before the introduction of partition chromatography on paper this principle was used by Martin and Synge 7 * when they increased the R values of the acetylamino acids on silica gel chromatograms by employing pro-gressively higher concentrations of n-butanol in chloroform. A second effect of the alcohol in this system was its virtual elimination of adsorption of the acetylamino acids by the gel. Even though most substances seem scarcely to be adsorbed by filter paper in those cases e.g. the chromatography Consden Nature 1948 162 359.Hais Chem. Listy 1948 42 125. Hais and RAbek Chem. Listy 1949 43 80. Boldingh Erperientia 1948 4 270. Partridge Biochem. J. 1948 42 238. Martin and Synge Biochem. J. 1941 35 1358. Gordon Martin and Synge Biochem. J 1943 37 79. 4 Consden Gordon and Martin Biochem. J. 1944 38 224. I30 QUANTITATIVE PAPER CHROMATOGRAPHY of inorganic ions,9 where such effects may be significant they are no doubt reduced by the admixture of more polar organic solvents. (ii) If the substances being separated are acids or bases their Rf values can be greatly influenced by adjustment of the pH of the aqueous phase, As an example the effect of ammonia on chromatograms developed with phenol in decreasing the Rj values of the dicarboxylic and increasing those of the basic amino acids may be mentioned.Conversely acetic acid has been used by Lugg lo to assist the separation of certain organic acids by n-butanol. Buffer solutions l1 have also been used in a rather similar way both on filter paper and on silica gel chromatograms for the separation of the penicillins. A further advantage of the addition of buffers at least on silica gel chromatograms is that partition chromatograms can be run by displacement development and thus much larger amounts can be handled. In addition the Rf values of certain compounds can be modified by the addition of substances mainly soluble in the organic solvent phase. Thus Winsten and Eigen l2 have found that the Rf values in n-butanol of the various members of the streptomycin group could be markedly increased by the addition of z yo of 9-toluenesulphonic acid.The solvent also con-tained z yo of piperidine which served partially to de-ionize the basic groups of the streptomycins. The amount of $-toluenesulphonic acid used was such that the pH of the system (determined after the addition of an equal volume of water) was on the alkaline side. However it was noted that separations could still be obtained if so much 9-toluenesulphonic acid had been used that the pH was acid. It is interesting to compare the effect of 9-toluenesulphonic acid with that of ammonia in n-butanol. Thus if it is assumed that the increased Rj values of the streptomycins were not solely due to the piperidine then 9-t oluenesulphonic acid although an acid substance must have affected the system in the same sense as would be expected for the addition of ammonia.It seems likely that additions of this type have their effect both because they increase the polarity of the organic solvent phase and because as acids they specifically modify the ability of this phase to dissolve the substances being separated. That the effect is not due only to the former cause seems to be indicated by the observation l3 that the addition of p-toluenesulphonic acid to collidine increases the Rf values of the basic preferentially to those of the neutral amino acids. (b) Adsorption by the Paper.-If the separations are to depend only on partition effects the substances in the mixture to be separated must not be adsorbed by the paper. Fortunately few substances are strongly adsorbed by untreated filter paper.Perhaps the largest group of substances for which adsorption is likely to be important are the organic dyestuffs. Attempts to separate dyes by partition chromatography on paper often lead to resolution of the mixture. However closer examination of the system may reveal that partition effects have been secondary (as is the case if the separation can be carried out equally well with the dry solvent). Adsorption chromatography on strips of filter paper of the streptomycins using a z yo aqueous solution of ammonium chloride as solvent has been described by Horne and Pollard.l* Unfortunately no resolution of the different streptomycins was obtained. 9 Arden Burstall Davies Lewis and Linstead Nature 1948 162 691 lo Lugg and Overell NUhWt? 1947 160 87.11 Goodall and Levi Nature 1946 158 675. 12 Winsten and Eigen J . Amer. Chenz. SOC. 1948 70 3333. 13 Gordon (unpublished). 14 Horne and Pollard ,I. Buct. 1948 55 231 A. H. GORDON 13= Among large molecules the proteins are also known to be adsorbed at least from salt solutions by filter paper. These effects will not be discussed further here as they may be the subject of another contribution. The examples mentioned may be enough to illustrate the possibility of adsorption in paper chromatography. This possibility should not automatically lead to the opposite view that the separations in all systems employing solvents which do not form two phases with water must necessarily be due to adsorption. As was first suggested to the author by Dr.Synge,15 a system employing for instance pyridine and water can be considered to be a partition chromatogram because the effect of the cellulose hydroxyl groups results in the existence of a very different milieu inside the fibre from that of the moving phase. (c) Chemical Reactivity of Solvents and Paper.-The substances under separation must not react either with the paper or with the solvents used. For the amino acids and many of the other substances for whose separation paper chromatography has proved applicable these criteria have proved to be rather easily fulfilled. However the recent finding of Moore and Stein,lG that using I/z/I-n-butanol-n-propanol-o.I N HC1 on starch chromatograms 6% and 7% of aspartic and glutamic acids respectively became esterified suggests that closer and more quantitative examination of paper chromatograms in which mixtures of acids and alcohols are used might reveal similar effects.Cellulose itself fortunately shows little or no reactivity towards either the substances separable on paper chromatograms or the solvents used for their separation. The problem thus becomes that of the possible reactions which may occur between the substances under separation the solvents and impurities present in the paper. This will be further discussed in the next section. ( d ) Techniques for Detection and Estimation of Spots.-Finally for qualitative work a suitable detection technique must be available. In practice the sensitivity reliability and ease of application of such techniques may be the decisive factor in determining whether paper chromatography is to be used.Certainly the remarkable sensitivity of ninhydrin by which less than I pg. of an amino acid can just be made visible has been most useful. The small scale of the method as a whole e.g. 1-20 pg. per com-ponent of the mixture under separation for the amino acids the sugars and for the purines and pyrimidines has the advantage that it is usually possible to develop as many chromatograms as there are different detection techniques available. Particularly when dealing with complex biological fluids this type of approach by which several classes of compounds can be simultaneously investigated may be of value. If quantitative estimations are to be attempted not only must the substances be located but sufficiently accurate means must be found for their estimation either in situ or after removal from the paper.For quan-titative purposes obviously the most suitable detection techniques are those which do not permanently affect the substances to be estimated. The most generally applicable means by which this can be accomplished is certainly the examination of the spots in u.-v. light. In this way spots of purines pyrimidines amino acids and peptides can be shown up. Holidayz7 has reported that as little as 0-5-1 pg. per cm.2 of the purines or pyrimidines can be detected if u.-v. light of wavelengths between 230 and 400 mp is used the most important wavelengths being in the region 1 5 Synge (personal communication). 16 Moore and Stein J . Baol. Chem. 1949 178 53.17 Holiday and Johnson Nature 1949 163 216 132 QUL4XTITATIVE PAPER CHROMATOGRAPHY of 254 mp. Although only amounts of more than 3 pg. per cm.2 of the amino acids can be detected l8 the use of this technique has been continued by Woiwod.19 By this means it is possible to avoid the necessity of developing parallel guide chromatograms and also the most inconvenient process of visualizing the spots as their mercury derivatives in the case of the purines and pyrimidines.20 The only chemical method so far reported which ultimately leaves the substances unchanged on the chromatogram seems to be that of Brante.?l This author has sprayed chromatograms of bases such as amino-ethanol, choline and creatine with an alcoholic solution of iodine. Brown spots are formed which disappear on being allowed to stand.Unfortunately this reagent is not very sensitive for the amino acids but as noted by Brante, it may prove useful in quantitative experiments. Turning now to the various methods which have been used for the quantitative estimation of substances separated by paper chromatography, these may be divided into those in which the substances are estimated on the paper usually as coloured derivatives and those in which they are first eluted and then estimated in solution. The former approach seems only to have been employed for the coloured substances formed from the amino acids by ninhydrin and for substances containing radio elements. Thus several authors 22 23 have reported that useful rough estimations can be made by visually comparing the spot strengths with those of a series of standard spots.Attempts have also been made to increase the accuracy of this method by the preliminary division of the amino acids into acidic, neutral and basic types.24 A spectrophotometer was used for estimating the strengths of the spots but as yet no indication has been given of the accuracy thus obtained. A spectrophotometer has been similarly used by Bull 25 who has been at pains to standardize the conditions of colour development. Under the conditions chosen arginine serine valine glutamic acid leucine threonine alanine and lysine gave the same amount of colour per mole the probable error for a single determination being almost g yo. This result is somewhat unexpected as Moore and Stein,26 who have carefully measured the amounts of colour formed in solution in presence of a reducing agent have reported for instance that threonine forms only 81 yo as much colour as does lysine.It may possibly be the case that some of the amino acids form more nearly equivalent amounts of colour in solution than on paper. On the other hand Bull 25 found that glycine and methionine give less colour on paper than does leucine whereas Moore and Stein 26 indicate that under their conditions these three amino acids yield almost identical amounts of colour per mole. Possibly more thorough investigations of the optimum conditions for the development of the ninhydrin colour on paper may lead to some reduction in the rather large errors at present involved in this technique.Work of this kind may also explain why Pratt and Auclair 27 find that glycine and glutamic acid are the amino acids giving visible spots at minimum strength whereas Moore and Stein 26 have found lysine to give the strongest colour in solution. Where substances containing radioactive isotopes have been separated 18 Phillips Nature 1948 161 153. 1 s Woiwod Biochenz. J . (in press). 20 Vischer and Chargaff J . Bid. Chetiz. 1948 176 703. 21 Brante Nature 1949 163 651. 22 Consden Gordon Martin and Synge Biochenz. J . 1947 41 596. 23 Polson Biochim. Biophys. Acta 1948 2 575. 2 4 Block Science 1948 108 608. 25 Bull Hahn and Baptist J Amw. Chem. SOG. 1943 71 550, Z 6 Moore and Stein J . BioZ. Chem. 1948 176 367. 2 7 Pratt and Auclair Scieizce 1948 108 213 A.H. GORDON I33 the detection technique i.e. the exposure of successive areas of the chromato-gram to the counter is itself suitable as a quantitative estimation. An apparatus designed for the estimation of spots containing P32 has recently been described.2B In order to estimate the amino acids by means of radio-active isotopes conditions have been worked out for their quantitative conversion into 9-iodophenylsulphonyl (pipsyl) derivatives containing P2. -4fter chromatography on paper nearly quantitative recoveries were obtained for the few amino acids tested.29 Further work on this promising method will be awaited with interest. Although techniques in which the substances are eluted from the paper prior to estimation cannot have the simplicity of those already mentioned, they would seem to offer several compensating advantages.Thus once the substance is in solution it can be treated with a much wider variety of reagents than while it is still on paper. This should for instance allow better control of the conditions under which the ninhydrin colour is formed from the amino acids. Unfortunately no attempt has been reported to utilize the valuable information on this subject already obtained by Moore and Stein.26 The most successful attempts to estimate quantitatively the amino acids involve their conversion into copper complexes. Thus Martin and Mittel-inann3O used what was essentially the method of Pope and Stevens,31 in which the amino acids are allowed to react with a suspension of copper phosphate in phosphate-borate buffer.After centrifugation from the excess of copper phosphate the amount of copper in solution was estimated polarographically. No attempt was made to determine the recoveries of the amino acids after elution from the paper but since the current increased linearly with the amount used estimations were shown to be possible if several comparison chromatograms of known amounts of each amino acid were developed simultaneously with each analysis. The results in a test analysis of gramicidin S clearly substantiated the theory that this peptide consists of five different amino acids in equimolar proportions. Only in the case of one of these acids leucine did the amount found deviate more than 5 yo from that expected. The copper method has also been used by Woiwod,lS who has modified the reagent by replacing sodium borate with di-sodium hydrogen phosphate.He has thus obtained a more stable suspension lower blanks and a linear relationship between amino nitrogen and copper. The copper has been estimated colorirnetrically with sodium diethyldithiocarbamate instead of polarographically. In order to obtain sufficient accuracy for those amino acids present as the weaker components of mixtures the analyses have been conducted on larger amounts (1-50 pg. amino nitrogen) than those employed by Martin and Mittelmann.30 In this way similar accuracies seem to have been achieved and because of the convenience of colorimetry compared with polarography this is probably at the moment the method of choice. Generally Woiwod like Martin and Mittelmann has employed the com-parative method but in addition he has measured the losses sustained as an amino acid moves down the chromatogram.Thus only 86 % of the original amount of glycine could be recovered after the spot had moved 10 in. down a chromatogram developed with 40 yo 12-butanol-ro yo acetic acid-rjo yo water. In partial explanation of such low recoveries, \Voiwod mentions an interesting observation to the effect that fluorescent 28 Lindberg and Hummel Arlziv. Kerizi 1949 I 17. 2 9 Iceston Udenfriend and Levy J. Amer. Chenz. Sac. 1947 69 3 I -j I . 3o Martin and Mittelmann Biocherrz. J. 1948 43 3 j 3 . 31 Pope and Stevens Biochem. J. 1939 33 1070 1.34 QUANTITATIVE PAPER CHROWITOGRAPHY material can often be seen extending right down to the front of the solvent.He considers that this may be due to failure of the amino acid to reach equilibrium with the mobile phase. However as the usual treatment with ninhydrin which is believed to be capable of revealing smaller amounts of amino acids than are visible in u.-v. light does not show up such streaks, it seems more likely that the amino acid must have been converted into a form no longer capable of giving a colour with ninhydrin. The existence of some substances in certain filter papers capable of combining with the amino acids is already known from the observations of Consden et aL4 on the formation of copper complexes of the amino acids. These are pro-gressively formed as the amino acids move down the chromatogram. These authors32 have also noted the formation on paper chromatograms of a zinc salt of cysteic acid which like the copper complexes gives a pink colour with ninhydrin.The Rf values of these metal complexes are close to those of the corresponding free amino acids but there seems no reason why there may not be other impurities in the paper which couldcombine to form complexes moving with any Rf value. The more the Rf values of such impurity amino-acid complexes differ from that of the amino acid itself, the more difficult it must be to obtain quantitative recoveries from the paper. The solution of these difficulties may either be similar to that adopted by Consden et aL4 who added a reagent to the paper which prevented the action of the copper by itself combining with it ; or else it may be necessary to employ special purification of the paper.If purification of the natural paper making materials proves inadequate use may in time be made of papers formed entirely from synthetic fibres. These findings naturally raise the question of the occurrence of similar losses as other substances move down paper chromatograms. Fortunately, very accurate methods of analysis exist both for the purines and pyrimidines and for the sugars after elution from the filter paper. For the purines and pyrimidines it is only necessary to measure their absorption at the appropriate wavelength in u.-v. light. Both Vischer and Chargaff go and Hotchkiss 33 have used this method without finding necessary a comparative approach similar to that employed for the amino acids. They report accuracies of j - 4 yo and do not mention progressive losses. The figures quoted suggest slight losses for those spots which have moved further down the chromatograms but are hardly sufficient to be decisivc. The need for further work on this question is apparent. Even more accurate analyses of sugars after separation on paper chromato-grams have been reported,34 and again no mention of progressive losses has been made. It may not be accidental that materials similar to the matrix of the chromatogram can be recovered with least loss. Kemiska Institutionea, Karolins ka Institutet, Stock Jtolm . 52 Consden Gordon and Martin Biochem J. 1946 40 580. 33 Hotchkiss J. Biol. Chew. 1948 175 315. 3 4 Flood Hirst and Jones J. Clzenz. SOC. 1948 1679

ISSN:0366-9033    

DOI:10.1039/DF9490700128

出版商:RSC    

年代:1949

数据来源: RSC

摘要:

1 Barrer J . SOC. Chcm. Ind. 1945 64 130 and 133 ; Brit. Pat. 548,905 U.S. Pad. SEPARATIONS USING ZEOLITIC MATERIALS BY R. M. BARRER Scolecite Nat rolite Edingtonite Thomsonite Heulandite Stilbite Chabazite Gmelinit e Mordenite Analcite Harmot ome Levynite 2,306,610. Received 13th June 1949 A short account has been given of aspects of zeolitic sorption which are important in separating molecular mixtures. Some natural crystalline zeolites fall into three classes of molecular-sieve sorbent each capable of separating mixtures by selective occlusion if there are sufficient differences in shape and dimensions between the molecules in the mixture. The molecules removed by occlusion must always be comparatively small but separations may be quantitative after a single exposure to the zeolite.By cation interchange and by burning out interstitial ammonium ions a diversity of modified molecular-sieve sorbents can be produced with extension of the classification as molecular sieves into numerous classes the potentialities of which as selectively sorbing media have not yet been fully investigated. Moreover by operating a t low temperatures major differences may develop in the sorption rates of gases all of which are rapidly sorbed at higher temperatures and in which there are only small differences in molecular dimensions. Zeolites have sometimes proved very suitable for resolving molecular mixtures.12 Complete separations can often be effected but to use the crystals effectively one must be acquainted with their sorbent and molecular- sieve properties.In the present summary factors determining these properties will be described for natural and artificially modified zeolites. Some General Properties of Gas -sorbing Zeolites.4rystalline zeolites act as sorbents only when the water normally present in interstices within the crystals has been removed by heat and evacuation. Not all these minerals are however capable of taking up gases and vapours in place of the interstitial water. There are three structural types of importance from the viewpoint of the sorptive property :- Bonding in two dimensions weaker than that in the third. (Fibrous zeolites.) Bonding in layers stronger than that between layers. (Laminar zeolites.) Bonding strong in all dimensions.(Robust three-dimensional net work zeolites .) Fibrous and laminar zeolites tend to shrink when interstitial water is removed and this shrinkage may be irreversible if the heating is too severe. If the heat treatment during outgassing is more gentle water may be resorbed or its place may sometimes be taken by the small polar molecule NH3. However these minerals do not occlude non-polar gases and have no general power of sorption. Robust network zeolites on the other hand may occlude both polar and non-polar gases copiously although even here Barrer and Belchetz J . SOC. Chem. Ind. 1945 64 131. I35 136 SEPARATIONS USING ZEOLITIC MATER1,ALS there is a wide variety in the accessibility of the interstitial volume. Harmo- tome and as a rule analcite sorb only small polar gases; mordenite and levynite sorb many small molecules and gmelinite and chabazite occlude a still greater variety of sorb ate^.^ No zeolite has hitherto been discovered which will occlude really large molecules and although a number still remain to be investigated it is unlikely that any crystalline zeolite will prove capable of doing so.In order for this to be the case the aluminosilicate framework would have to be so open that instability would almost certainly ensue with collapse of the whole structure into a crystal more economical of space and so of higher density. Never- theless the chabazite lattice based on a comparatively open aluminosilicate framework is surprisingly stable and at least one purely synthetic zeolite has been made by hydrothermal methods with a stable framework just as open as that of chabazite.' This zeolite has no naturally occurring counter- part and there remains the possibility that other synthetic open framework crystals may be grown with the limitation on the degree of openness noted above.As a group the gas-sorbing zeolites are therefore capable of sorbing only smaller molecules and their potential value in chromatography is for removing such smaller molecules from admixture with larger ones. Molecular - Sieve Properties of Zeolites .-Several researches have been directed towards establishing the molecular-sieve behaviour of some zeolites and also towards the artificial modification of this behavi0ur.l The molecular-sieve properties to be described refer to thoroughly out gassed finely powdered zeolites but with an outgassing temperature inside the range of thermal stability of the crystals.Variable outgassing conditions can strongly influence the molecular-sieve behaviour. Several sorbate molecules of fairly accurately known shape and dimensions were used to determine the accessibility of the crystalline interstices as sorption sites. isoButane propane ethane methane argon nitrogen and oxygen were used for this purpose and three categories of molecular-sieve crystal were characterized Class I Do not sorb iso-C,H, (and other (Chabazit e gmelinit e synthetic zeolite BaA1Si206.itH20). iso-paraffins). Sorb C,H 8 (and other n-paraffins) slowly at rooin temperature or above and sorb C&6 CH and molecules of Class I1 (Mordeni t e) .smaller cross-section rapidly. Do not sorb n- or iso-paraffins. Sorb C2H6 and CH slowly and N and molecules of smaller cross- section rapidly at room tempera- ture or above. Class I11 (Ca- and Ba-mordenites). Do not sorb $2- or iso-paraffins. Negligible sorption of C,H and CH,. Rapid sorption of N2 0 and molecules of smaller cross- section at room temperature or above. 3 Barrer PYOC. Iiuy. SOC. A 1938 167 392. 4 Barrer Tram. Faraduy SOC. 1944 40 555- 5 Barrer and Ibbitson 1944 40 195 and 206. 6 Barrer Am%. Reports 1944 31. 'Barrer J . Cheiia. SOC. 1948 127. 8 Barrer and Riley ibid. 1948 133. 9 Barrer Tmns. Furaduy SOC. 1949 45 338.10 Barrer Nutwe (in press). =37 Table I. R. M. BARRER It was demonstrated that the shaje rather than the molecular volume of the sorbate exercises the decisive influence upon the sorptive behaviour. The cross-section referred to in the above classification of several zeolites is that normal to the direction of the greatest length of the molecule. For all n-paraffins this cross-section is the same in their most extended configura- tion. Since chabazite for example slowly occludes the standard molecule propane it should slowly occlude all n-paraffins. This was verified using n-butane n-pentane and wheptane. Increasing the chain length exerted a secondary influence in further slowing down the velocity of intracrystalline diffusion but as sorption equilibrium was approached large amounts of all the n-paraffins studied were occluded.n-Heptane has a considerably larger molecular volume but a smaller cross-section than isobutane which is completely excluded from the intracrystalline sorption sites. Again the standard molecule isobutane CH(CH,) is as noted not sorbed by chabazite ; and accordingly other molecules of similar shape and size such as CHCl and CHBr are also not sorbed. On the other hand the molecules ClCH,Cl ClCH2CH, BrCH,CH and BrCHzBr which have similar shapes and dimensions to CH,CH2CH3 are like propane slowly occluded by chabazite. Many other examples of predictable behaviour were observed and the sorptive action towards a variety of species is summarized in Selective Sorption from Mixtures .-The molecular-sieve property of zeolites would suggest that clear-cut separations could be obtained in molecular mixtures of which one constituent was freely occluded by the zeolite while the other constituent having the wrong molecular dimensions was not occluded.Proof of this simple principle does not appear to have been published before 1945 but since then the possibility of such separations has been abundantly ver5ed.l Two methods of verification were employed. In one-the static method-a quiescent gas or liquid mixture was brought into contact with the outgassed zeolite and after an interval of time the non-sorbed gas or liquid was removed and its composition investigated.* The second procedure-the percolation or chromatographic method-consisted in passing a gaseous or liquid mixture through a column of outgassed zeolite and investigating the composition of the effluent.This method which was employed only in one or two instances (with C,H,-C,H and CH,C12-CHCI mixtures) can undoubtedly be extended and chroma- tograms might be developed for suit able molecular mixtures. The actual separations carried out have been given elsewhere 6 and we may therefore only make some general comments upon them. Class I molecular-sieve sorbents (chabazite and synthetic BaAlSi,O,.nH,O) can be used to separate n-paraffins of lower molecular weights from all iso-paraffins aromatic hydrocarbons or naphthenes. In these same sorbents methane derivatives in which the substituents are small groups such as -NH2 -OH -CH, -CN -Cl are rapidly occluded and similarly substituted ethanes are slowly occluded.Quantitative separations of these species are usually possible from compounds which like iso-paraffins or aromatics are excluded from the sorption sites (e.g. column 4 Table I). In mordenite a Class I1 sorbent methane derivatives in which the substi- tuents are as before -NH, -OH -CH, -CN -C1 are slowly occluded but the similarly substituted ethanes are not sorbed so that separations were possible of substituted methanes from the corresponding substituted ethanes as well as from many other species as given in Table I. In the * The time-intervals ranged from an hour to more than a week and the temperature from 20' to 250' C according to the occlusion velocity of the sorbable component and its thermal stability.E* SEPARATIONS USING ZEOLITIC MATERIALS Typical mole- cules rapidly occluded a t room tempera ture or below He Ke c o co HC1 HBr H, x2 0 cos cs H,O NO NH3 CH,OH CHaNH CH,CN HCN c12 CH,CI CH,Br CH,F CH,CI, CH,F CH, C,H, C2H2 CH CH ,S I3 138 Section (i) Class I minerals -- Section (ii) Class XI minerals Section (iii) Class 111 minerals hTOLECULES OCCLUDED OR EXCLUDED BY THREE CLASSES OF 3'10LECULAR StEVE TABLE I Typical molecules moderately rapidly or slowly occluded at room temperature or above C,H and simple higher I Aromatic hydrocarbons n-Daraffins C,H ;OH C,H ,NH C ,H ,CN C,H,SH CH,CO.OCH NH(CH,) , NH(C,H,) CH, C,H CH,OH CH,NH CH,CN CH,Cl CH,F HCN c1 H,S ._________ He Ne A H, 0, S2 -- He.Ne case of Class I11 sorbents the separations effected were of small polar inorganic molecules (H,O NH, HC1) from organic compounds. Some partial or complete separations were also effected using zeolites in which both components were sorbed but at different velocities. Using chabazite these included the pairs C2H6-C3H, CH,OH-C2H,Br CZHsOH- n-C,H,, C,H,OH-CH2Br2 CH,CN-CH,Br, the molecule first mentioned being occluded most rapidly. A number of separations were also carried out which could not be effected by direct distillation either because the mixtures were azeotropic pairs (H,O-C2H,0H CH,OH-CH,COCH, H,O-d' ioxane CS2-CH,COCH, C2H50H-n-C,Hls C2H,0H-toluene) or because the constituents had nearly the same boiling points (n-heptane-isooctane).CH,CN Artificial Modification of Molecular - Sieve Sorbents .-Water which is strongly sorbed by zeolites by occupying sorption sites which would otherwise be accessible to other sorbates can greatly diminish not only the A HC1 NH C,H,F C,H,CI C,HjBr ' I,. HI CH ,Br CH,I H.COOCH, HCOOC,H1 CH,COCH I 1 All classes of molecules in columns 3 and 4 of i section (il I I Typical molecules which are not appreciably occluded a t room temperature or above I cyclo and iso-par- affins. Derivatives of these hydrocarbons. Heterocyclic com- pounds (e.g. thio- phene pyrole pyri- dine).CHCl, CCl, CHCl = CCl, CH,CHCl, CHCl ,CCI, C ,C1 6 and analogous bromo and iodo compounds. Secondary straight chain alcohols thiols nitriles and halides. Primary amines with NH group attached to a secondary carbon atom. Tertiary am- ines. Branched chain ethers thio-ethers and secondary amines All molecules referred to in column 4 section (ii). Also CH, CzHb CN,NH, CH,OH CH,SH CH,Cl CH,F R. M. BARRER Li-mordenite Li- and Ba-mordenite l 1 Lamb and Woodhouse J . 14ir?er. Chew SoC, 1936 58 2637. l 2 Emmett and DeWitt ibid. 1943 65 1253. I39 FIG. relative sorption rates of each of several sorbates in levynite and in a series of mordenites modified by base-exchange. A great measure of control over the sorption rate can be exercised by a suitable choice of base-exchange mordenite.amount but also the velocity of sorption and the molecular-sieve character of the crystals. This behaviour has been studied in any detail only with chabazite,ll l2 which was progressively transformed from a Class I sorbent into one whose properties more nearly recalled a Class I11 sorbent. Although a small amount of water might be a useful modifying agent larger amounts by depressing the sorptive capacity of the zeolite unduly would have a poisoning effect. Indeed other strongly occluded polar molecules sometimes appear to act as poisons and limit separations of mixtures which in their I40 SEPAR-ATIONS USING ZEOLITIC MATERIALS absence could easily be resolved.In resolving mixtures of n- and iso-par- affins for instance the gases should be free of moisture or other small polar molecules capable of preferential sorption and so of acting as poisons. Other methods of modifying the zeolites have now been evolved which are free from objections. The first of these is by cation interchange.s13 In inany zeolites exchange occurs freely although higher temperatures are usually needed than is the case with amorphous or gel zeolites such as Permutit. Thus exchange is best carried out by hydrothermal methods * in the temperature range 150~-250~ C using an excess of the exchanging salt. In this way for instance the ions in chabazite have been largely replaced by Ca Sr Ba NH, K Cs Na and Li ; and similarly mordenites enriched in Ca Ba NH, K Na and Li have been prepared.In analcite eschange is more specific the replacements Na + K + NH occurring readily whereas exchanges which aimed to introduce Ca Ba Cs and Li occnrred at best only to a very limited extent. FIG. 2.-Examples of the molecular-sieve action of Ba- and Ca-morctenites towards different simple gases as expressed by selectivity in the rates of sorption a t -19g" C and -78" C. A study of the cation-exchanged niordenites established that within limits set by the anionic aluminosilicate framework of the crystal the exchange process yields a group of sorbents with a range of molecular-sieve properties. The powdered crystals gave very diverse sorption rates as Fig. I indicates. Although these rates are liable t o be modified by a mean particle size which may have differed from one of these sorbents to another the range in values of the constant D/a2 (a = mean radius of particle D =1' diffusion coefficient) is greater than anything to be anticipated on this count.In general in any one sorbent the rate sequence which tended to be preserved was Kr<A<N,<O, Ne<H,<He while the affinity between sorbent and sorbate was approximately in the converse order. By cooling the sorbents to low temperatures (- 78" - 186" - ~ g j ' C ) one may often greatly alter the sorption rate of one gas relative to another,914 and also augment the amounts of each gas occluded at equili- l3 Barrcr J . Cheni. SOC. 1948 21 jS. 14 Idem Nature 1947 159 50s. * Except in thc case of NH,' v-hen vapour-phase eschangc is the simplest procedure using NH,C1 as exchanging salt.Exchange by heating the zeolite with fused low- mclting snlts is also often very satisfactory. 141 R. M. BARRER brium. Examples of the differences in sorption rates are indicated in Fig. 2 which illustrates the magnitude of the molecular-sieve effect even where the dimensions of the diffusing molecules differ only slightly. A second method of altering the dimensions of the interstitial channels and hence of modifying the molecular-sieve behaviour of zeolites is limited to ammonium oxidize gas.1° with ammonium oxygen It has gas ion-exchanged been at - found 350" possible C crystals slowly which to are capable the of occluding oxygen ions 4- 0 .. . . NH,f) + 302 = 4(- OH) + 6H20 + z&. Since the - OH group occupies a smaller volume than the - 0 . . . . NH,+ group this method as far as it has been investigated has led to crystals with more readily accessible or more open intracrystalline channels than occurred in the parent NHpeolites. A great deal of study remains to be done before the potentialities of the ion-exchange and oxidation methods are fully evaluated in the production of special purpose sorbents for effecting molecular-sieve separations of particular mixtures. It has however already been shown possible to produce a great diversity of graded molecular sieves and that quite small modifications may result in important variations in relative sorption rates variations which may often be augmented still more by an appropriate choice of the experimental temperature.The production of modified zeolites already seems to have reached a point permitting separa- tions of species in which only minor differences in dimensions arise (N2-O2 ; A-0% ; A-N ; Ne-H, etc.) although actual separations have not yet been carried out.* Activation and Poisoning of Zeolites .-It has already been Ijointed out that the zeolites should be used as finely divided powders which are well outgassed because the interstitial water may modify the amounts and velocities of occlusion. Several attempted separations using chabazite and involving acetone proved to be unsuccessful perhaps because acetone which was sorbed only excessively slowly nevertheless blocked the entrance to the interstitial channels and prevented uptake of species which in absence of acetone would have been fully occluded.Little systematic work has yet been done on interference effects which may arise between individual pairs of molecules both of which are sorbed but at different velocities. In general one would expect such interference to be less important if the rapidly diffusing species were more polar than the slowly diffusing molecule because the polar species should show a higher affinity for the intracrystalline sites and so a greater interstitial concentration. The zeolites can also be poisoned when one constituent of the molecular mixture tends to decompose at the temperature of the experiment giving carbonaceous deposits on the surface of the sorbent particles.Such deposits can be burnt off using air or oxygen and the sorptive power of the zeolite restored at least in part. Clzm&ry Department Mnrischal College The Upziversily A berdeen. * Work is now in progress concerned with this aspect. SEPARATIONS USING ZEOLITIC MATERIALS BY R. M. BARRER Received 13th June 1949 A short account has been given of aspects of zeolitic sorption which are important in separating molecular mixtures. Some natural crystalline zeolites fall into three classes of molecular-sieve sorbent each capable of separating mixtures by selective occlusion if there are sufficient differences in shape and dimensions between the molecules in the mixture.The molecules removed by occlusion must always be comparatively small, but separations may be quantitative after a single exposure to the zeolite. By cation interchange and by burning out interstitial ammonium ions a diversity of modified molecular-sieve sorbents can be produced with extension of the classification as molecular sieves into numerous classes the potentialities of which as selectively sorbing media have not yet been fully investigated. Moreover by operating a t low temperatures major differences may develop in the sorption rates of gases all of which are rapidly sorbed at higher temperatures and in which there are only small differences in molecular dimensions. Zeolites have sometimes proved very suitable for resolving molecular mixtures.12 Complete separations can often be effected but to use the crystals effectively one must be acquainted with their sorbent and molecular-sieve properties.In the present summary factors determining these properties will be described for natural and artificially modified zeolites. Some General Properties of Gas -sorbing Zeolites.4rystalline zeolites act as sorbents only when the water normally present in interstices within the crystals has been removed by heat and evacuation. Not all these minerals are however capable of taking up gases and vapours in place of the interstitial water. There are three structural types of importance from the viewpoint of the sorptive property :-Bonding in two dimensions weaker than that in Scolecite the third. Nat rolite (Fibrous zeolites.) Edingtonite Thomsonite Bonding in layers stronger than that between Heulandite (Laminar zeolites.) Bonding strong in all dimensions.(Robust three-dimensional net work zeolites .) layers. Stilbite Chabazite Gmelinit e Mordenite Analcite Harmot ome Levynite Fibrous and laminar zeolites tend to shrink when interstitial water is removed and this shrinkage may be irreversible if the heating is too severe. If the heat treatment during outgassing is more gentle water may be resorbed or its place may sometimes be taken by the small polar molecule NH3. However these minerals do not occlude non-polar gases and have no general power of sorption. Robust network zeolites on the other hand, may occlude both polar and non-polar gases copiously although even here 1 Barrer J .SOC. Chcm. Ind. 1945 64 130 and 133 ; Brit. Pat. 548,905 U.S. Pad. 2,306,610. Barrer and Belchetz J . SOC. Chem. Ind. 1945 64 131. I3 136 SEPARATIONS USING ZEOLITIC MATER1,ALS there is a wide variety in the accessibility of the interstitial volume. Harmo-tome and as a rule analcite sorb only small polar gases; mordenite and levynite sorb many small molecules and gmelinite and chabazite occlude a still greater variety of sorb ate^.^ No zeolite has hitherto been discovered which will occlude really large molecules and although a number still remain to be investigated it is unlikely that any crystalline zeolite will prove capable of doing so. In order for this to be the case the aluminosilicate framework would have to be so open that instability would almost certainly ensue with collapse of the whole structure into a crystal more economical of space and so of higher density.Never-theless the chabazite lattice based on a comparatively open aluminosilicate framework is surprisingly stable and at least one purely synthetic zeolite has been made by hydrothermal methods with a stable framework just as open as that of chabazite.' This zeolite has no naturally occurring counter-part and there remains the possibility that other synthetic open framework crystals may be grown with the limitation on the degree of openness noted above. As a group the gas-sorbing zeolites are therefore capable of sorbing only smaller molecules and their potential value in chromatography is for removing such smaller molecules from admixture with larger ones.Molecular - Sieve Properties of Zeolites .-Several researches have been directed towards establishing the molecular-sieve behaviour of some zeolites and also towards the artificial modification of this behavi0ur.l The molecular-sieve properties to be described refer to thoroughly out gassed finely powdered zeolites but with an outgassing temperature inside the range of thermal stability of the crystals. Variable outgassing conditions can strongly influence the molecular-sieve behaviour. Several sorbate molecules of fairly accurately known shape and dimensions were used to determine the accessibility of the crystalline interstices as sorption sites. isoButane propane ethane methane argon nitrogen and oxygen were used for this purpose and three categories of molecular-sieve crystal were characterized : Class I Do not sorb iso-C,H, (and other (Chabazit e gmelinit e iso-paraffins).Sorb C,H 8 (and synthetic zeolite BaA1Si206.itH20). other n-paraffins) slowly at rooin temperature or above and sorb C&6 CH and molecules of smaller cross-section rapidly. Do not sorb n- or iso-paraffins. Sorb C2H6 and CH slowly and N and molecules of smaller cross-section rapidly at room tempera-Class I1 (Mordeni t e) . ture or above. Class I11 (Ca- and Ba-mordenites). Do not sorb $2- or iso-paraffins. Negligible sorption of C,H and CH,. Rapid sorption of N2 0, and molecules of smaller cross-section at room temperature or above. 3 Barrer PYOC. Iiuy. SOC. A 1938 167 392. 4 Barrer Tram.Faraduy SOC. 1944 40 555-5 Barrer and Ibbitson 1944 40 195 and 206. 6 Barrer Am%. Reports 1944 31. 'Barrer J . Cheiia. SOC. 1948 127. 8 Barrer and Riley ibid. 1948 133. 9 Barrer Tmns. Furaduy SOC. 1949 45 338. 10 Barrer Nutwe (in press) R. M. BARRER =37 It was demonstrated that the shaje rather than the molecular volume of the sorbate exercises the decisive influence upon the sorptive behaviour. The cross-section referred to in the above classification of several zeolites is that normal to the direction of the greatest length of the molecule. For all n-paraffins this cross-section is the same in their most extended configura-tion. Since chabazite for example slowly occludes the standard molecule propane it should slowly occlude all n-paraffins.This was verified using n-butane n-pentane and wheptane. Increasing the chain length exerted a secondary influence in further slowing down the velocity of intracrystalline diffusion but as sorption equilibrium was approached large amounts of all the n-paraffins studied were occluded. n-Heptane has a considerably larger molecular volume but a smaller cross-section than isobutane which is completely excluded from the intracrystalline sorption sites. Again the standard molecule isobutane CH(CH,) is as noted not sorbed by chabazite ; and accordingly other molecules of similar shape and size such as CHCl and CHBr are also not sorbed. On the other hand the molecules ClCH,Cl ClCH2CH, BrCH,CH and BrCHzBr which have similar shapes and dimensions to CH,CH2CH3 are like propane slowly occluded by chabazite.Many other examples of predictable behaviour were observed and the sorptive action towards a variety of species is summarized in Table I. Selective Sorption from Mixtures .-The molecular-sieve property of zeolites would suggest that clear-cut separations could be obtained in molecular mixtures of which one constituent was freely occluded by the zeolite while the other constituent having the wrong molecular dimensions, was not occluded. Proof of this simple principle does not appear to have been published before 1945 but since then the possibility of such separations has been abundantly ver5ed.l Two methods of verification were employed. In one-the static method-a quiescent gas or liquid mixture was brought into contact with the outgassed zeolite and after an interval of time the non-sorbed gas or liquid was removed and its composition investigated.* The second procedure-the percolation or chromatographic method-consisted in passing a gaseous or liquid mixture through a column of outgassed zeolite and investigating the composition of the effluent.This method which was employed only in one or two instances (with C,H,-C,H, and CH,C12-CHCI mixtures) can undoubtedly be extended and chroma-tograms might be developed for suit able molecular mixtures. The actual separations carried out have been given elsewhere 6 and we may therefore only make some general comments upon them. Class I molecular-sieve sorbents (chabazite and synthetic BaAlSi,O,.nH,O) can be used to separate n-paraffins of lower molecular weights from all iso-paraffins, aromatic hydrocarbons or naphthenes.In these same sorbents methane derivatives in which the substituents are small groups such as -NH2 -OH, -CH, -CN -Cl are rapidly occluded and similarly substituted ethanes are slowly occluded. Quantitative separations of these species are usually possible from compounds which like iso-paraffins or aromatics are excluded from the sorption sites (e.g. column 4 Table I). In mordenite a Class I1 sorbent methane derivatives in which the substi-tuents are as before -NH, -OH -CH, -CN -C1 are slowly occluded, but the similarly substituted ethanes are not sorbed so that separations were possible of substituted methanes from the corresponding substituted ethanes as well as from many other species as given in Table I.In the * The time-intervals ranged from an hour to more than a week and the temperature from 20' to 250' C according to the occlusion velocity of the sorbable component and its thermal stability. E 138 SEPARATIONS USING ZEOLITIC MATERIALS --Section (ii) Class XI minerals TABLE I hTOLECULES OCCLUDED OR EXCLUDED BY THREE CLASSES OF 3'10LECULAR StEVE ._________ He Ne A H, 0, S2 Section (i) Class I minerals Typical mole-cules rapidly occluded a t room tempera ture or below He Ke, H, x2 0, c o co, cos cs, H,O NH3 CHaNH, HC1 HBr NO CH,OH CH,CN HCN c12 CH,CI CH,Br CH,F CH,CI, CH,F CH, C,H,, C2H2 CH H,S CH ,S I3 Section (iii) Class 111 minerals --He. Ne Typical molecules moderately rapidly or slowly occluded at room temperature or above Typical molecules which are not appreciably occluded a t room temperature or above C,H and simple higher I n-Daraffins I C,H ;OH C,H ,NH I C,H,F C,H,CI C,HjBr ' I,.HI CH ,Br , CH,I C ,H ,CN C,H,SH H.COOCH, HCOOC,H1 CH,COCH, CH,CO.OCH, NH(CH,) , NH(C,H,) Aromatic hydrocarbons cyclo and iso-par-affins. Derivatives of these hydrocarbons. Heterocyclic com-pounds (e.g. thio-phene pyrole pyri-dine). CHCl, CCl,, CHCl = CCl,, CH,CHCl,, CHCl ,CCI, C ,C1 6, and analogous bromo and iodo compounds. Secondary straight chain alcohols thiols, nitriles and halides. Primary amines with NH group attached to a secondary carbon atom. Tertiary am-ines.Branched chain ethers thio-ethers and secondary amines CH, C,H, CH,OH CH,NH i section (il CH,CN CH,Cl CH,F HCN 1 All classes of molecules in columns 3 and 4 of c1, A HC1 NH, All molecules referred to in column 4 section I (ii). Also CH, CzHb, I CH,OH CN,NH,, CH,SH CH,CN, CH,Cl CH,F case of Class I11 sorbents the separations effected were of small polar inorganic molecules (H,O NH, HC1) from organic compounds. Some partial or complete separations were also effected using zeolites in which both components were sorbed but at different velocities. Using chabazite these included the pairs C2H6-C3H, CH,OH-C2H,Br CZHsOH-n-C,H,, C,H,OH-CH2Br2 CH,CN-CH,Br, the molecule first mentioned being occluded most rapidly. A number of separations were also carried out which could not be effected by direct distillation either because the mixtures were azeotropic pairs (H,O-C2H,0H CH,OH-CH,COCH, H,O-d' ioxane, CS2-CH,COCH, C2H50H-n-C,Hls C2H,0H-toluene) or because the constituents had nearly the same boiling points (n-heptane-isooctane).Artificial Modification of Molecular - Sieve Sorbents .-Water which is strongly sorbed by zeolites by occupying sorption sites which would otherwise be accessible to other sorbates can greatly diminish not only th R. M. BARRER I39 Li-mordenite Li- and Ba-mordenite FIG. relative sorption rates of each of several sorbates in levynite and in a series A great measure of control over the of mordenites modified by base-exchange. sorption rate can be exercised by a suitable choice of base-exchange mordenite.amount but also the velocity of sorption and the molecular-sieve character of the crystals. This behaviour has been studied in any detail only with chabazite,ll l2 which was progressively transformed from a Class I sorbent into one whose properties more nearly recalled a Class I11 sorbent. Although a small amount of water might be a useful modifying agent larger amounts by depressing the sorptive capacity of the zeolite unduly would have a poisoning effect. Indeed other strongly occluded polar molecules sometimes appear to act as poisons and limit separations of mixtures which in their l 1 Lamb and Woodhouse J . 14ir?er. Chew SoC, 1936 58 2637. l 2 Emmett and DeWitt ibid. 1943 65 1253 I40 SEPAR-ATIONS USING ZEOLITIC MATERIALS absence could easily be resolved.In resolving mixtures of n- and iso-par-affins for instance the gases should be free of moisture or other small polar molecules capable of preferential sorption and so of acting as poisons. Other methods of modifying the zeolites have now been evolved which are free from objections. The first of these is by cation interchange.s13 In inany zeolites exchange occurs freely although higher temperatures are usually needed than is the case with amorphous or gel zeolites such as Permutit. Thus exchange is best carried out by hydrothermal methods * in the temperature range 150~-250~ C using an excess of the exchanging salt. In this way for instance the ions in chabazite have been largely replaced by Ca Sr Ba NH, K Cs Na and Li ; and similarly mordenites enriched in Ca Ba NH, K Na and Li have been prepared.In analcite eschange is more specific the replacements Na + K + NH occurring readily, whereas exchanges which aimed to introduce Ca Ba Cs and Li occnrred at best only to a very limited extent. FIG. 2.-Examples of the molecular-sieve action of Ba- and Ca-morctenites towards different simple gases as expressed by selectivity in the rates of sorption a t -19g" C and -78" C. A study of the cation-exchanged niordenites established that within limits set by the anionic aluminosilicate framework of the crystal the exchange process yields a group of sorbents with a range of molecular-sieve properties. The powdered crystals gave very diverse sorption rates as Fig. I indicates. Although these rates are liable t o be modified by a mean particle size which may have differed from one of these sorbents to another the range in values of the constant D/a2 (a = mean radius of particle D =1' diffusion coefficient) is greater than anything to be anticipated on this count.In general in any one sorbent the rate sequence which tended to be preserved was Kr<A<N,<O, Ne<H,<He while the affinity between sorbent and sorbate was approximately in the converse order. By cooling the sorbents to low temperatures (- 78" - 186" - ~ g j ' C ) one may often greatly alter the sorption rate of one gas relative to another,914 and also augment the amounts of each gas occluded at equili-l3 Barrcr J . Cheni. SOC. 1948 21 jS. 14 Idem Nature 1947 159 50s. * Except in thc case of NH,' v-hen vapour-phase eschangc is the simplest procedure, Exchange by heating the zeolite with fused low- using NH,C1 as exchanging salt.mclting snlts is also often very satisfactory R. M. BARRER 141 brium. Examples of the differences in sorption rates are indicated in Fig. 2, which illustrates the magnitude of the molecular-sieve effect even where the dimensions of the diffusing molecules differ only slightly. A second method of altering the dimensions of the interstitial channels, and hence of modifying the molecular-sieve behaviour of zeolites is limited to ammonium ion-exchanged crystals which are capable of occluding oxygen gas.1° It has been found possible slowly to oxidize the ammonium ions with oxygen gas at - 350" C : 4- 0 . . . . NH,f) + 302 = 4(- OH) + 6H20 + z&.Since the - OH group occupies a smaller volume than the - 0 . . . . NH,+ group this method as far as it has been investigated has led to crystals with more readily accessible or more open intracrystalline channels than occurred in the parent NHpeolites. A great deal of study remains to be done before the potentialities of the ion-exchange and oxidation methods are fully evaluated in the production of special purpose sorbents for effecting molecular-sieve separations of particular mixtures. It has however already been shown possible to produce a great diversity of graded molecular sieves, and that quite small modifications may result in important variations in relative sorption rates variations which may often be augmented still more by an appropriate choice of the experimental temperature.The production of modified zeolites already seems to have reached a point permitting separa-tions of species in which only minor differences in dimensions arise (N2-O2 ; A-0% ; A-N ; Ne-H, etc.) although actual separations have not yet been carried out.* Activation and Poisoning of Zeolites .-It has already been Ijointed out that the zeolites should be used as finely divided powders which are well outgassed because the interstitial water may modify the amounts and velocities of occlusion. Several attempted separations using chabazite and involving acetone proved to be unsuccessful perhaps because acetone which was sorbed only excessively slowly nevertheless blocked the entrance to the interstitial channels and prevented uptake of species which in absence of acetone would have been fully occluded. Little systematic work has yet been done on interference effects which may arise between individual pairs of molecules both of which are sorbed but at different velocities. In general one would expect such interference to be less important if the rapidly diffusing species were more polar than the slowly diffusing molecule because the polar species should show a higher affinity for the intracrystalline sites and so a greater interstitial concentration. The zeolites can also be poisoned when one constituent of the molecular mixture tends to decompose at the temperature of the experiment giving carbonaceous deposits on the surface of the sorbent particles. Such deposits can be burnt off using air or oxygen and the sorptive power of the zeolite restored at least in part. Clzm&ry Department, Mnrischal College, A berdeen. The Upziversily, * Work is now in progress concerned with this aspect

ISSN:0366-9033    

DOI:10.1039/DF9490700135

出版商:RSC    

年代:1949

数据来源: RSC