Search

In the adsorption isotherm approach, ion exchange should be considered as a competitive adsorption of two sorbates. This would lead to adsorption isotherms equivalent to the law of mass action. The simple Langmuir equation, representing a single-sorbate system, should not be used to represent ion exchange even at low occupancy values of an ideal reaction. The application of this equation generally leads to erroneous values of selectivity constants. In case of heterovalent exchange, it also leads to erroneous capacity values. These facts have been revealed by the isotherm of Boyd et al. [G.E. Boyd, J. Schubert, A.W. Adamson. J. Am. Chem. Soc. 69 (1947) 2816] for homovalent exchange, and by the approximate isotherm of Misak for heterovalent exchange. Cases of constancy of the equilibrium quotient of equivalent ion exchange are numerous. Adsorption isotherms can also be applied to the more frequent cases where such constancy is only observed over a certain composition range of the solid. Binding constants represented by the simple Langmuir equation are actually not constants, being dependent on solution concentration. Except under certain particular conditions, such constants are generally meaningless.

Hydrous ferric oxide (HFO), stannic oxide (HSO) and mixed ferric/stannic oxide (HMO), obtained by precipitation with ammonia from metal chloride solutions, were all found to be microporous. Analysis (X-ray, infra-red, thermal) has shown that HFO and HMO are amorphous, whereas HSO is microcrystalline. It also showed that all the samples are oxides containing adsorbed H 2 O molecules and surface OH groups. In addition, it revealed that HMO is not a simple mixture of HFO and HSO, a conclusion which was also supported by the data on porous texture, intrinsic surface ionization constants, pH pzc and apparent capacity. For all the oxides, the apparent Zn 2+ capacity is generally higher than the Co 2+ capacity, in agreement with a stronger interaction of Zn 2+ with O 2− or OH − on the surface. Equilibria measurements point to the adsorption of both Co 2+ and Zn 2+ on all oxides with the release of two protons.

Thermal analysis has shown that contrary to the case of Zn 2+ , considerable amounts of water enter in HFO (hydrous ferric oxide), HSO (hydrous stannic oxide), and HMO (hydrous mixed oxide) when Co 2+ is adsorbed. A large amount of Co 2+ and Zn 2+ in all the oxides cannot participate in isotopic exchange, seemingly because of being trapped inside the micropores, a fact hindering oxide capacity determination by isotopic dilution. The % adsorption–pH curves for tracer and 10 −3 M transition metal chloride solutions show a general preference to Zn 2+ over Co 2+ by all the oxides, in agreement with thermal analysis data. Large shifts of these curves to the left occur with tracer experiments on many samples on adding 0.1 M NaCl, probably due to Cl − adsorption accompanied by a change in the adsorption mechanism. Conclusions from log K d –pH plots may be erroneous, probably due to changes in the activity coefficients of surface species even at sufficiently low surface coverages. More adsorption of Co 2+ and Zn 2+ generally occurs on the oxide of lower pH PZC (higher acidity).

Thermal analysis has shown that contrary to the case of Zn{sup 2+}, considerable amounts of water enter in HFO (hydrous ferric oxide), HSO (hydrous tannic oxide), and HMO (hydrous mixed oxide) when Co{sup 2+} is adsorbed. A large amount of Co{sup 2+} and Zn{sup 2+} in all the oxides cannot participate in isotopic exchange, seemingly because of being trapped inside the micropores, a fact hindering oxide capacity determination by isotopic dilution. The % adsorption-pH curves for tracer and 10{sup {minus}3} M transition metal chloride solutions show a general preference to Zn{sup 2+} over Co{sup 2+} by all the oxides, in agreement with thermal analysis data. Large shifts of these curves to the left occur with tracer experiments on many samples on adding 0.1 M NaCl, probably due to Cl{sup {minus}} adsorption accompanied by a change in the adsorption mechanism. Conclusions from log K{sub d}-pH plots may be erroneous, probably due to changes in the activity coefficients of surface species even at sufficiently low surface coverages. More adsorption of Co{sup 2+} and Zn{sup 2+} generally occurs on the oxide of lower pH{sub PZC} (higher acidity).

The thermodynamics of Li/Na and Li/Cs exchange on hydrous stannic oxide (α-stannic acid) was studied in aqueous and methanolic solutions. With respect to Δ G °, the results in the aqueous medium are qualitatively similar to those for a β-stannic acid sample. For Δ H ° and Δ S °, contrary results are obtained, possible reasons for which are discussed. The addition of methanol generally leads to a decrease in Δ G ° due to the changed entropy term. When comparison could be made (consistent standard states), the data of transfer thermodynamics showed that the effect of this addition on ion–solvent interactions is the major effect. However, the excess thermodynamic functions of mixing, calculated at different loadings, showed that upon the addition of methanol, the excess enthalpy and entropy of mixing decrease considerably at relatively low loadings and increase at high loadings. As these functions express the departure of the exchange reaction from ideality due to interactions (ion-fixed site and ion–solvent) in the solid phase, it was concluded that alcohol addition affects these interactions. This was not revealed in the transfer data where the whole-exchanger composition range was considered.

Self-diffusion coefficients of Na + , D i , inside hydrous stannic oxide have been measured in aqueous and alcoholic solutions at different Na + -capacities. D i always displays the usual decrease with capacity, which seems to be much less profound at the relatively high capacities. There is apparently almost no effect of addition of methanol on D i , whereas a decrease of D i , which disappears towards relatively high capacities, occurs on addition of propan-1-ol. These results are in marked difference with previous results of the effective diffusion coefficients of Na + H + systems. This difference seems to indicate that the effect of alcohol addition in increasing the strength of electrostatic interaction of counter ions with the exchanger surface (a factor which, when predominating, decreases mobility) is higher with protons than with the Na + ions.

In a previous work, the author gave an adsorption isotherm for the general case of single-site heterovalent exchange and simpler approximate equations having linear forms involving the same plot used for the Langmuir isotherm. The present work gives a substantiation of these equations by calculations involving some potential practical examples. It also extends the equations to the case of two-site (multisite) heterovalent exchange and considers the evaluation of the different ion exchange parameters. Further discussion is given on the hazards of the application of the Langmuir equation to ion exchange reactions. For homovalent exchange, instead of the Langmuir isotherm, use of the single-site isotherm or two-site isotherm of Boyd et al., as treated in our previous work, is recommended. This isotherm is graphically represented by the same plot used for the Langmuir isotherm. It is more instructive, avoids confusion and allows for equilibrium constant calculations. In the case of heterovalent exchange, the equations given in the previous work for single-site adsorbents and in the present work for two-site (multisite) ones should invariably be used to obtain correct results.

A ferric oxide gel (Fe-I), prepared by adding ammonia to ferric chloride and washing with dilute ammonia and water, and the gel (Fe-II), obtained by further washing of Fe-I with more concentrated ammonia (which removed more of the Cl− ions from the gel) and water were studied. It was found that the apparent capacity for Cl− ion adsorption by ligand exchange with surface OH groups is much higher for Fe-II than for Fe-I. The FE-II displayed a complex variation of this capacity with alcohol addition, whereas Fe-I displayed a continuous increase in sorption with alcohol concentration. The latter increase is in the order propan-1-ol > ethanol > methanol, and may be attributed to a stronger interaction of Cl− ions with the surface. For FE-II, the complex behaviour may in part be connected with the development of micropores. The latter effect changed the rate-controlling mechanism from chemical for Fe-I to particle diffusion for FE-II, and may be accompanied by an increase in the surface area, leading to an increase in the capacity. For the Cl− ion exchange, the values of the rate constant for Fe-I and the internal diffusion coefficient Di for FE-II decreased progressively with alcohol addition. This pointed to an imbibition of alcohols by both gels. The order of the Di values for the various anions is not changed by the addition of alcohol, which probably shows that Di is governed by the complexing ability of the anions with the surface.

This work gives a critical review of the behaviour of hydrous oxides as ion exchangers. The various physicochemical properties of this important class of adsorbents and the effect of preparation procedures are summarized. Data on the ion exchange kinetics and thermodynamics are presented and discussed. Selectivity is also discussed in terms of the different parameters. Both aqueous and mixed media are considered. In addition, the different sorption mechanisms, some relevant sorption models, and potential applications in chemical separations are given. Moreover, this review attempts to point to the experimental and theoretical confusions now alarmingly met in the literature.

The simple Langmuir isotherm equation for adsorption of a single adsorbate on a single-site surface is still frequently applied to ion-exchange reactions. When a single straight line is not obtained by this equation, a multi-site surface is assumed and additional terms of the equation are used for analysis. Ion exchange should be considered as a competitive adsorption of two sorbates. Apart from the fact that the selectivity coefficient or separation factor must be constant, other conditions of application of the equation are frequently overlooked. This paper is intended to stress these conditions and to point to the uncertainties in the quantities and conclusions derived from the Langmuir plots of ion-exchange data.

Sorption studies of SO 2− 4 , Cl − , Br − and N0 − 3 ions on ferric oxide gel have shown that these ions replace surface OH groups. The particle diffusion-controlled exchanges follow mainly the Helfferich first mechanism. This situation is different from that in weak-base resins in the similar case of ionization of weakly dissociated groups by reaction with H + co-ions. The interdiffusion coefficient D i decreases in the order SO 2− 4 S> Cl − > Br − > NO − 3 . It was deduced that, at least for SO 2− 4 , the complexing ability with FE(III) at the gel/water interface is weaker than that in solution.

In the purely hydroxide solutions of Li + , Na + and K + , the interdiffusion coefficient D i decreases always in the order propan-l ol>ethanol>methanol. This may show that on going from one alcohol to another, the decrease of ion size is the dominant factor. The role of ion size in determining D i seems to increase from Li + to K + . A considerable replacement of the more retarding OH - coions in the oxide by Cl - (by adding Cl - to the external alcoholic hydroxide solution) enhances alkali ion diffusivity. However, diffusivity may decrease at low Cl - concentration in solution due to limited replacement of OH - by Cl - and to retardation by both coions. This is more often with K + forming more ion pairs with Cl - in solution, which may result in a higher invasion of the oxide by Cl - . In the hydroxide/chloride solutions, the order of D i for Li + is the same as in the purely hydroxide solutions. W ith Na + , this order is mostly the reverse order (methanol>ethanol>propan-l-ol). This is due to the probably large differences from one alcohol to another in the Na + -Cl - coion interactions in the solid. The effect of these interactions outweighs the effect of an increasing role, relative to Li + , of ion size in determining D i . With K + , “irregular” orders are obtained, which are resul tants of increasing contradictory roles of alkali ion-Cl - interactions and ion size.

The Na + -working capacity of hydrous stannic oxide mostly increases on addition of methanol, ethanol or propan-1-o1. The largest increase is obtained with ethanol and the smallest one is obtained with propan-1-o1. Considering the previous results of Li + exchange, it can generally be said that the change of interdiffusion coefficient D i on addition of alcohol is the result of the effect of this addition on counterion-solvent, counterion-exchange site and alkali ion-coion interactions. The first factor contrary to other factors, enhances diffusion due to a dehydration (decrease of size) of diffusing counter ions. On going from one alcohol to the other in the purely hydroxide solutions, the decrease of the size of counter ions is the dominant factor, and D i decreases in the order: propan-1-o1>ethanol>methanol. As with the aqueous medium, D i increases on addition of NaCl to the alcoholic hydroxide solution due to a weaker interaction with Cl − coions, compared to OH − coions, inside the oxide pores. The order of D i in the alcohols is mostly reversed on adding Cl − , which is probably connected with Na + Cl − interactions. Differences in the diffusion behaviour of Li + and Na + are discussed.

A reversible H+/Cs+ exchange in ceric tungstate is displayed only at low loadings in both aqueous and methanolic solutions. The replacement of H+ by Cs+ probably causes structural changes inside the solid. At 30 and 60% methanol, the effect of alcohol addition on ion-solvent interactions, leading to a decrease of ΔG°, is counteracted by the effect on ion-ion interactions in the exchanger, leading to an increase of ΔG°. At 90% methanol, the effect of ion-ion interactions predominates and ΔG° becomes positive (i.e. selectivity reversal occurs). At 30% methanol, this effect leads to negative contributions to ΔH° and ΔS°. However, at 60 and 90% methanol, it leads to positive contributions to these quantities. These contributions are tentatively discussed in the light of the different interactions in the solid phase.

The working capacity of hydrous stannic oxide for Li+ increases in most cases on addition of alcohols. It decreases in the order ethanol > methanol > propan-1-ol and is determined by both ion—solvent and ion—ion interactions. The change in the interdiffusion coefficient Di on addition of alcohols depends on two seemingly opposite factors: the decrease in the size of the diffusing species and the increase in the strength of interaction of the counterions with the exchanger matrix. The former factor seems to dominate in 60% propan-1-ol, and Di surpasses its value in the aqueous medium. Di decreases always in the order propan-1-ol > ethanol > methanol, which may be due to the size of the diffusing species decreasing in the reverse order. As in the aqueous solution, the replacement of OH− co-ions by Cl− inside the oxide pores increases Di, and the increase is higher with higher Cl− concentration.

Equilibrium measurements of sorption of Co2+ from acidic chloride solutions on hydrous stannic oxide show that this sorption is accompanied by the release of an equivalent amount of H+. Besides, Co2+ and Cl− seem to be sorbed on separate sites on the oxide surface. The study of isotopic exchange kinetics in 0.05–0.1 M Co2+ solutions proves a chemical control mechanism for the reaction that can mainly be represented by two processes. The rate constant of the slow process, whose contribution becomes quite small in the oxide heated at 400°C, decreases with the increase of the capacity.

The kinetics of exchange of alkali ions with H + in hydrous stannic oxide was studied from hydroxide and hydroxide/chloride solutions. At ionic strength ⪖0.01 M, the rate is controlled by the interdiffusion of the alkali ion and H + inside the hydrous oxide. Co-ions seem to be sorbed to a limited extent. There effect on the interdiffusion coefficients seems to be related to the strength of alkali ion-OH − interaction near the mouths of the pores. The replacement of OH − by Cl − inside the oxide leads to the enhancement of the exchange rate for Li + and Na + , and particularly for Li + . For K + , practically no change is observed.

The surface characteristics of stannic oxide (α-stannic acid) heated at different temperatures change only little after storage for about two years over saturated ammonium chloride. At the same heating temperature, the diffusivity of Na+ in the oxide decreases with the increase of capacity due to a stronger electrostatic retardation by the exchange sites. The effect of the heating temperature of the oxide on this diffusivity is connected mainly with the change in the porous texture.

The kinetics of self-diffusion and trace component diffusion of Cs + in hydrous stannic oxide heated at 50 and 400°C was studied. The diffusion coefficients decrease with the increase of the exchange capacity due to the increase of the strength of electrostatic interaction with the exchange sites. They are generally higher at 400°C due to the change of the microporous texture to a mesoporous one. This is more important than the contrary effect of the decrease of site acidity at 400°C. The self-diffusion and trace component diffusion coefficients are almost equal, probably due to the presence of almost the same amount of free water in the oxide which seems to retain the alkali ions in a substantially dehydrated form.

NO3−/halide exchanges on two hydrous cerial samples were studied. It was found that NO−3 is always preferred to Br− and I− but not to Cl−, where the selectivity differs from one sample to the other and depends on both capacity and heating temperature. The two samples prefer the halide ion with a smaller crystallographic radius. The increase of site density (capacity) at any heating temperature leads to the increase of selectivity (or affinity) to NO3− relative to the halides and to the halide of the smaller radius. At high capacity, this is supported by the relatively small surface area of the samples, which renders the smaller ions closer to the exchange sites. Porosity and pore size seem to to be of minor importance in respect to the effect of heating temperature on selectivity. Basicity of the sites seems to be more important in the case of NO3−/halide exchanges than in the case of halide/halide exchanges. In the latter exchanges, (surface) site density is more important. Surface structure, different in the two samples, also has a considerable effect on the selectivity behavior of these exchanges.

For three prepared hydrous ceria samples, it was found that the surface properties can be affected by the preparation procedure details. A substantial crystallinity improvement occurs at 400°C. Infrared and capacity data show that OH condensation starts at 100°C and that H 2 O is involved in anion exchange. Surface structure, and not surface area, is the factor determining the capacity. The acidity and basicity decrease with heating temperature.

Two hydrous stannic oxides, prepared by metal oxidation with nitric acid and by acidification of sodium stannate, are microcrystalline, displaying certain differences in their infrared patterns. They are microporous at 50–200°C, but mesoporous, with a different pore size distribution, at 400°C. At 50°C, they have rather similar cationic capacities and p K values but the change of these with heating temperature is markedly different. Besides, their anionic capacity is, at all heating temperatures, largely different. Surface structure, and not surface area, is the main factor determining the capacity. It is seemingly also the reason behind the differences displayed by the two samples.

The effect of the physicochemical properties of four ferric oxide gels on their ion exchange behaviour was studied. The gels are apparently bifunctional cation exchangers in which both OH groups and sorbed H 2 O act as exchange sites. A study of sodium tracer diffusion has shown that ionic mobility in these gels increases with the increase of pore size and decrease of capacity. Selectivity measurements indicated that an increase in site density leads mostly to an increase of affinity or selectivity for the ion of smaller crystallographic radius. An increase in porosity and pore size leads to a decrease of selectivity. Besides, differences in crystallinity or surface structure seem to be an important factor in determining selectivity.

Adsorption of radiocesium and radiocobalt from aqueous solutions on a new managanese oxide was studied. The adsorption of both cesium and cobalt follows the Freundlich adsorption isotherm. The described oxide can be used for the preconcentration and separation of traces of cobalt and cesium. Tracer cobalt can be gained from highly concentrated saline solutions, a fact which has not been reported before for any other ion exchanger.

Self-diffusion of Na+ and Cs+ in hydrous ceria heated at 50, 200 and 400°C was studied. In all cases, the mobility decreases with increasing heating temperature and ion exchange capacity. The self-diffusion coefficient of Cs+ in ceria heated at 200 and 400°C is lower than that of Na+ at relatively high capacities. In the other cases, this coefficient is about the same for the two ions. Tentative explanations are given for these results.

The thermodynamics of alkali cation exchange on hydrous tin oxide was studied. The data obtained were compared with those for hydrous zirconia and ceria. It was found that ions of smaller crystallographic radius are more preferred by tin oxide than by zirconia. This result, unexpected from the viewpoint of oxide acidity, seems to be related to the smaller pore size and lower water content of tin oxide. The alkali ions present in fine pores in tin oxide seem to interact more strongly with the exchanger than in the case of the much less acidic ceria. The large increase in the entropy of alkali-ion-form tin oxide on replacing a smaller alkali ion by a larger one seems to be related mainly to the lower degree of freedom of the former ion inside the solid. When hydrous tin oxide is saturated with Na + or Cs + and heated at 400 and 800°C, some of these ions are released and some remain fixed. The remaining exchange sites are highly selective to the counterion present during heating; this selectivity is higher for the sample heated at the higher temperature.

Iron oxide gels purified as far as possible from sorbed cations and anions were prepared and heated at different temperatures. Their main structural features were determined by X-ray, infrared and thermal analysis. Furthermore, their porous texture was evaluated from nitrogen adsorption isotherms and their cation exchange capacities at different pH values were determined. The results obtained show that apparently bifunctional exchangers with varying porous textures, acidities and capacities are obtained. This would make it possible to study the effect of their physicochemical properties on their ion exchange behaviour. It is also shown that the capacity is not closely correlated with surface area and that water molecules adsorbed on the oxide surfaces act as cation exchange sites.

Alkali cation exchange on three hydrous ceria samples heated at different temperatures was studied. The decrease of pore size seems to result in a general increase of the magnitude of selectivity, provided that this decrease does not involve the formation of relatively very fores (inaccessible to nitrogen). Exchange in these pores is probably behind the high selectivity to sodium over the other alkali metal ions. A particular selectivity behaviour due to a profound decrease of pore size is probably due to a formation of cavities of proper dimensions for a certain ion, ion exclusion, or a basic dehydration (involving probably primary hydration shells) of counter ions.