4.ion exchange in solid state

8/2/2019 4.Ion Exchange in Solid State

1/24

Chern. Rev. 1988. 88 , 125-148 125

Role of Ion Exchange in Solid-state Chemistry

ABRAHAM CLEARFIELD

Departmnt of chemlsby, Texas ABM Unlverrny. College Station, Texas 77843

Received May 18. 1987 (Revised Manuscript Received October 1, 1987)

Confenfs

I. IntroductionA. BackgroundB. A Brief History

11. Hydrous OxidesA. Particle Hydrates

1. Structural and Ion ExchangeDescription

2. Conductivity Behavior of ParticleHydrates

3 . Catalytic Applications

B. Framework Hydrates1. Antimony Oxides2. Niobium and Tantalum Oxides3 . Proton Conductivity of Hydrous Oxides4. Hydrous Oxides with Tunnel Structure:

MnO,111. Ion Exchange in Mixed Oxides

A. Ternary Oxides1. AB0,-Type Structures2. Spinels3 . Metal Titanates

B. Quaternary Oxides1. Mixed-Metal Titanates2. Complex Niobates and Tantalates wlth

Tunnel or Perovskite StructuresC. Conclusions

Related CompoundsA. a-Type M(1V) Layered Phosphates

IV. Group I V (Groups 4 and 14) Phosphates and

1. Structure and Ion Exchange Behavior2. Phase Changes of ollayered M(IV)

3. Gas-Solid and Fused-Salt Exchange

B. Group V (Groups 5 and 15 ) Phosphates

A. The NASICON SystemB. P-Aiuminas

Phosphates

Reactions

V. Fast Ion Conductors and Ion Exchange

1. P-Alumina: Structure and Conductivity2. 8"-Alumina: Structure and Conductivity3 . Ion Exchange in P - and @"-Alumina

1. Proton Conduction Mechanisms2. Layered Phosphates3 . Protonic fil@"-Aiumina4. Protonated NASICONs5 . Miscellaneous Conductors

VII . Some Theoretical Considerations

C. Roto n Conductors

VI. Anion Exchangers

A. Thermodynamics of Io n ExchangeE. Kinetic FactorsC. Conclusions

12 512 512612 712 712 7

128

12 8

12 812 812 912 913 0

13113113113113 2132132133

134134

134134135

136

136137137138

139139140140141141142142143143143144145

138

, --

Abraham Clearfield was born in Philadelphia and receivedand M.A. in Chemistry from Temple University. His W I D . ,was earned at Rutgers University and conferred in 195served as a chemist with the U S . Army Quartermaster Ccwas employed from 1955 to 1963 by the National Lead CIndustries). H e pined the aculty of Ohio University in Athamin 1963 and became a full professor in 1968. He serve

\.eedL1

as a

ionairogram director lor Thermodynamics and Structure at the hScience Foundation during 1974-1975 and as Chaiman of the S o l iState Subdivision of the American Chemical Society in 1986. Hepined the faculty of Texas ABMUniversity in 1976. He is currentlyAssociate Dean for Research 01 the College of Science and Di-rectw of the Texas AaM Univerrilv MaterQls Scienc~ R ~ l r a m Ha. -~ ~ ~ ~ ~~ . .~., _.has published over 160 papers in the areas of crystallography.solid-state chemistry. and inorganic ion exchange materials. andhas won awards for excellence in teaching and research.

I . Infroducflon

A. Background

Until recently, i o n exchange behavior was th oug ht to

be confined to a very l im i t ed number of inorganiccompound types. Amo ng these the best known are th eclays and zeolites, both natural and synthetic. However,as work progressed in th is area, i t has come t o be rec-ognized th at io n exchange properties are possessed bymany different classes of compounds.'-3 A pa r t i a llisting of such compou nd types is presented in Tab le1. A discussion encompassing a l l these classes of com-pounds would require the preparation of a series ofbooks. Therefore, our i n t en t i o n i s t o limit the scopehere to those aspects of norganic i on exchange whichare relevant t o solid-state chemistry. T he topics cov-ered wi ll inc lude synthesis of new exchangers and ex-amination of th eir properties, th e use of i on exchange

0009-2665/88/0788-0125$06.50/0 0 1988 American Chemlcal Society


2/24

126 Chemical Reviews, 1988, Vol. 88, No. 1 Clearfield

TABLE 1. Principal Classes of Inorganic Ion Exchangers

exchange capacity,type example mequiv/g

1. smectite clays

2. zeolites3. substituted aluminum

phosphates4. hydrous oxides

5. group IV (groups 4 and 14")

6. other phosphates7. condensed ph osphates8. heteropolyacid sa lts9. ferrocyanides10. titanate s11. apatites

12. anion exchangers

13. miscellaneous type s14. ast ion conductors

phosphates

montmorillonitesMz/n"+[A14-xMgxlSis)Ozo(OH),Na,(A1Oz),(SiOz),~zHzOsilicoaluminophosphates, metal-substituted aluminophosphates(M,"+Al~-,Oz)(Poz)(oH),i,(a) SiOz.xHzO, Zr02 .xHz 0(b) (H30)2Sb206,olyantimonic acidZr (HP04)2 .H20 , n (HP04)2 .H20

uranium phosphates, vanadium phosphates, antimony phosphates

M,XY1zO1o.xHiO (M= H, Na+, NH4+; X = P, As, Ge, Si, B; Y = Mo , W )M,,,"+Fe(CN), (M = Ag+, Zn2+,Cu2+,Zr4+;n = ion charge)Na2Tin02n+ln = 2-10)Cal(t,H,(P04)6(0H)2-.

hydrotalciteM ~ & ~ Z ( O H ) ~ B C O ~ . ~ H Z ~alkaline-earth sulfates, polyvanadates, sulfides&alu mina , Nal+,A11101,+,pNASICON, Nal+,Zr2Si,P3~,0,2

N e 0 3

0.5-1.5

3-7depends upon

value of x1-22-54-8

-80.2-1.51.1-6.12-9cation and

anion exchangeanion exchange

2-4

1.5-32-7

"In this paper th e periodic group notation in parentheses is in accord with recent actions by IUPAC an d ACS nomenclature committees.A and B notatio n is elimina ted because of wide confusion. Grou ps IA and IIA become groups1 and 2. Th e d-transition elem ents comprisegroups 3 through 12, and t he p-block elements comprise groups 13-18. (Note that the former Roman n umber designation is preserved in the

last digit of the new numbering: e.g., I11- and 13.)processes to synthesize new compounds, proton and fastion transport, electrochemical devices based on ionicconductors, and some theoretical considerations. As faras possible we shall try to describe in structural termsthe underlying basis for the observed behaviors andrelated structure to synthetic processes. In addition tothe limitations listed above, we shall also not discusszeolites and clay chemistry as these subjects are ade-quately treated in the li te ra t~ re . ~- ~

B. A Brief Hlstory

Ion exchange was originally discovered to take placein soils by ThompsonlO and by Way and Roy." Th eactive compounds were later identified by Lemberg12and by Wiegner13 o be clays, glauconites, zeolites, andhumic acids. It was not until the beginning of the 20thcentury that this knowledge was applied to the devel-opment of water softeners. Th e first commerciallyavailable ion exchangers were amorphous alumino-silicate gels.14J5 These gels later become known aspermutites but their instability toward acid solutionsand variability of behavior led chemists to seek alter-natives. This search eventually led to the synthesis oforganic ion exchange resins.I6 These resins soon dom-inated the field because of their uniformity, chemicalstability, and the ability to control resin properties bysynthetic procedures. This situation existed until thepost World War I1 era. Th e advent of nuclear tech-nology initiated a search for ion exchange materials thatwould remain stable above 150 "C and in high radiationfields. Early attention was focused on hydrous oxidessince it was known that they "sorbedn or coprecipitatedmany ions. It was soon discovered that hydrous oxidescombined with anions such as phosphates, vanadates,molybdates, and antimonates produced superior ionexchanger^.'^-^^ Much attention was paid to the syn-

thesis and properties of zirconium phosphate, whichactually had been discovered earlier21 but remainedhidden as a classified document. These materials were

amorphous in nature and no t resistant to alkaline hy-drolysis. In fact, extensive hydrolysis occurred even inhot water. Therefore, they were never extensivelyutilized. The exception was the use of amorphouszirconium phosphate as a sorbant in portable artificialkidney devices.22

A new direction was applied to the field of inorganicion exchangers when Clearfield and S t y n e ~ ~ ~ emon-strated that zirconium phosphate could be crystallized.The availability of crystals allowed the structure of thispolymorph of zirconium phosphate to be deter-

and with this knowledge, the observed ionexchanger behavior could be explained in structuralterms.l

Paralleling the progress in the field of nonsilicate ionexchangers, rapid developments were taking place inzeolite chemistry. The brilliant research of Barrer andhis co-workers2* inally led to the synthesis of t he fau-jesite family of zeolites by Union Carbide ~ c i e n t i s t s . ~ ~However, the rapid growth of zeolite technology notonly stemmed from their use as sorbants and ion ex-changers, but also as catalysts, particularly in petroleumcracking.30 Zeolite chemistry in all its aspects hascontinued to expand, and many new and startling dis-coveries have been made in recent year^.^^^^^Zeolitesand molecular sieves exhibit such complex behavior thatcontinued research will undoubtedly lead to new dis-coveries and technology. An interesting facet of theirtechnology pertinent to our discussion is the recent useof zeolite-A as a water softener in certain detergents.The tonnage now consumed is larger than that of all theorganic resins combined.

While the original intent of the studies in hydrousoxides and salts of polyvalent acids was the explorationof their ion exchange properties for separations andremoval of ions from nuclear waste streams, our in-creased knowledge of these behaviors needs to be ap-plied in order to properly describe the solid-statechemistry of many of these compounds. We shall beginby describing how these concepts apply to the hydrous


3/24

Ion Exchange in Solid-state Chemistry

1 l l I I I I I I I l 1 8 ~ ~-0

0

mI\

.-

aa

Chemical Reviews, 1988, Vol. 88, No. 1 127

P H

Figure 1. Ion exchan ge capacity versuspH for hydrous tin andzirconium oxides. (From ref 18, with permission.)

oxides and how new compounds may be synthesizedand characterized on this basis.

I I . Hydrous OxidesThe hydrous oxides may be divided into two main

types: those like ZrOzand S n0 2 or which ion exchangeoccurs only on the surface, and others such as thehydrous oxides of antimony and manganese, whichcontain cavities or tunnels in which exchangeable ionsreside. The former compounds have been termedparticle hydrates while the latter are referred to asframework hydrates.33

A. Particle Hydrates1. Structural and Ion Exchange Description

All of the group I11 (groups 3 and 13; see footnote ain Table 1 or more on this group notation) and groupIV (groups 4 and 14) metals produce hydrous oxidesthat belong to this class. They are usually prepared bythe addition of base to a soluble salt of the cations.Polyvalent cations hydrolyze in water, and as hydroxylgroups accumulate on the cation through hydrolysis orbase addition, they polymerize by 01a ti on .~~ hen theolation process produces sufficiently large particles ofreduced charge, precipitation occurs. Rapid precipita-tion in the cold leads to amorphous or poorly crystallineproducts, whereas aging, boiling, or hydrothermal pro-cedures yield crystalline hydrous The X-raydiffraction patte rns of th e crystalline hydrous oxidesresemble the same crystalline oxides formed by high-temperature reactions. For example, Ti02 orms ana-tase or rutile while hydrous ZrOz can be prepared inboth monoclinic and cubic forms.35 Unit cell dimen-sions determined for these crystalline hydrous oxidesare not too different from those of the calcined oxides.Thus, the structure of the interiors of the hydrous ox-ides must closely resemble th at of the calcined oxideswith all the hydroxyls and water molecules on thesurface.

The surface groups are responsible for the observedion exchange behavior. These hydrous oxides canfunction as both cation and anion exchangers. Thisamphiprotic behavior arises as follows:

M-OH + X- e M X + OH- (1)M-OH + A+ i=! MOA + H+ (2 )

Process 1 takes place more readily in acid solutionwhereas reaction 2 is more prevalent in alkaline solu-

A B

Figure 2. Schematic repres entation of hydrous oxide particles.(A) Zr02.nH20 with hydroxyl ion in the intercrystalline waterneutralizing the positively charged pa rticle surface. (B) S n 2 0 .nHzO with H30+ n the intercrystalline water neutralizing thenegatively charge d particle surface. (Fro m ref33, with permission.)

tion. In Figure 1 are shown ion uptake curves as afunction of pH for hydrous t in and zirconium oxides.18We note that tin oxide, being more acidic than ZrOz,exchanges more cations a t the same pH than does ZrOPThe opposite is true for anion exchange. These phe-nomena are general. For example, hydrous silica orsilicic acid is so acidic tha t i t exhibits few, or no, anionexchange properties. The point a t which the cation andanion curves cross is called the zero point of charge(ZPC) nd represents a region in which the charge onthe surface is zero. The pH at which this occurs for anygiven oxide is not constant but depends upon thepreparative conditions, crystallinity, and degree of ag-

A simple but more quantitative structural model ofhydrous oxides has recently been proposed.33 The de-scription of hydrous zirconia and how it forms by ola-tion followed by oxolation is essentially that presentedby Clearfield.36-3s-40 The hydrous oxide interior isconsidered to be a dense, anhydrous oxide. Within theZr02 particle, each Zr4+ ion is surrounded by eight ox-ygen atoms (fluorite structure) while each oxygen is inturn coordinated by four metal atoms. The situationon the surface is quite different. Here the oxygen atomsare all associated with protons, either as hydroxylgroups or as water molecules, because the process ofoxolation and particle condensation is incomplete.Terminal oxygens are thought to belong to watermolecules and bridging oxygen to hydroxyls. Thi sproton association makes the clusters positivelycharged. The assumption is made tha t all the metalatoms at the surface retain the full oxygen coordinationof the interior, which results in an O/Zr ratio, R , greaterthan 2, since the surface oxygens are not fully coordi-nated to metal atoms. The authors33 calculate R to be2.6 for a cubic-fluorite particle of 25 A and 2.15 for a100-Aparticle. Individual particles are closely associ-ated to form the hydrous oxide gel in which watermolecules fill the voids between particles. This inter-particle solution, s, s either acidic (SA)or basic (&),depending upon the degree of protonation a t the par-ticle surface. A schematic representation of this modelfor both Zr02 and S n0 2 s given in Figure 2.

According to England e t al.,33 hydrous t in oxideproduces a weakly acidic solution when equilibratedwith water. This indicates that the surface is negativelycharged by dissociation of hydroxyl groups to form 02 -as shown in eq 3. The liberated protons then reside

(3)

ing.3637

M-OH + H2 0 e MO- + H 3 0 +


4/24

128 Chemical Reviews, 1988, Vol. 88, o. 1 Clearfield

1000/T i ' K - ' l

Figure 3. log ( U T ) ersu4 reciprocal temperature for SnO z.nH zOand ZrOz.nHzO. (From ref 33, with permission.)

in the intraparticle water. The situation described hereis analogous to that present in organic cation exchangerswhere the water contained in the swelled resin beadscontains signficant amounts of cation an d/or hydro-nium ions. Any amount of cations in solution, eventrace amounts, would exchange with the intraparticlehydronium ion to produce a measurable acidity in theexternal solution. Tin oxide should then be able toexchange cations in acid solution and reference toFigure 1 shows that this is true. In contrast to tin oxide,hydrous zirconia has a more highly protonated surfaceand yields basic solutions by donation of protons fromthe interparticle water to the surface, and this is re-flected in its ion exchange behavior also (Figure 1) . Itis well to point out here tha t precipitation of hydrouszirconia in acid solutions allows an appreciable amountof anion to be incorporated into the gel. Washing withbase or refluxing results in displacement of the anionsby hydroxyl groups while refluxing also results in par-ticle growth.35

2. Conductivity Behavior of Particle Hydrates

Th e foregoing picture of hydrous oxides would in-dicate that they should exhibit an appreciable protonconductivity. Conductivities have been measured forboth tin and zirconium hydrous oxides of known watercontent by the complex admittance method.33 Theresults are shown as a function of temperature in Figure3. Two mechanisms of proton motion in aqueousmedia have been described$l a drift of hydrated protons

in the solution and a Grotthus mechanism. In particlehydrates with high water content the latter mechanismwould be expected to dominate. Increasing the acidityor basicity and water content, as well as the surfaceareas, would be expected to increase the conductivity.Although conductivities for several additional hydrousoxides have recently been determined,42 quantitativecorrelation with theory has not yet been achieved. Afurther discussion of the conductivity mechanism isgiven in section II.B.3.

3. Catal'ic Applications

Hydrous oxides such as those of zirconium and tita-nium are utilized as catalysts or catalyst s ~ p p o r t s . ~ ~ , ~ ~

For many such uses high surface areas are required. Itis clear from the foregoing description of the oxides thathigh surface areas require small particle size. A cubicparticle 25 A on a side would have a surface area ofabout 600 m2/g. Thus , by rapid precipitation in thecold with good stirring, one can prepare particles o f thissize. To ensure the absence of impurities, the oxidesneed to be washed with ammonia in order to removeany exchanged ions, dried under vacuum, and calcinedat low temperature. Heating to temperatures muchabove 400 "C results in agglomeration of particles withloss of surface area. An excellent exposition of theseprocesses for hydrous zirconia has been given byRijnten.45

B. Framework Hydrates

We shall illustrate the nature of these oxides by de-scribing two of them, antimonic acid or antimony pen-toxide and manganese dioxide.

1. Antimony Oxides

Hydrous antimony oxide, Sb2 05 aH 20 , an be pre-pared in amorphous and crystalline forms. Theamorphous product is unstable, crystallizing on agingor on treatment with acid at elevated temperature^.^^It is now well established that three crystalline phasesexist: a pyrochlore phase, a layered phase with an il-menite-like s tructure, and a cubic phase.47 The mostcommon form of hydrous antimony oxide was referredto in the early literature as polyantimonic acid becauseof it s high proton content (ion exchange capacity of 5.1m e q u i ~ / g ) . ~ * l ~ ~owder pattern X-ray diffraction datawere analyzed to show that this compound actually had8 pyrochlore structure33 and a composition representedby the formula (H30)2Sb206-H20. his formula yieldsa theoretical ion exchange capacity of 5.08 m e q ~ i v / g ~ ~and fits the thermogravimetric curve closely.49 Al-though the hydronium ions and water molecules werenot located, they are thought to reside in the cavitieswith the oxygen atoms centered on the 16d and 8bpositions of the ideal cubic space The rigidpyrochlore framework imposes size limitations on cationselectivity. Abe51 observed the following selectivityorder: Na+ > K+ > Rb+ > Cs+ > Li'. In order forthese hydrated ions to exchange into the pyrochlorecavities, they need to be partially dehydrated. Theselectivity order represents a compromise betweencation size and hydration energy. For lithium ion thehydration energy term must dominate, preventing highuptake.

The ilmenite-type (layered) antimony oxide wasprepared by acid treatment of KSb03, which itself hasan ilmenite structure.47 n the potassium compound thestructure is a true three-dimensional one, but uponreplacement of the K+ ions with hydrogen (hydronium)ions, the structure is broken down into layers with theexchangeable hydrogen ions situated beteen the layers.However, the composition is very similar to that of thepyrochlore compound.

Antimonic acid with a cubic structure was preparedby acid treatment of cubic KSb03. The latter com-pound was in turn synthesized by heating K2H2Sb2-0 7 . 4 H 2 0 in a sealed gold tube at 500-700 "C under3-kbar pressure for 10 h. Its formula is (H30)+HSb2O6or HSb03.xH20. NMR spectra for all three polymorphs


5/24

Ion Exchange in Solid-State Chemistry Chemical Revlews, 1988, Vol. 88, No. 1 129

T. *CTABLE 2. Summary of Conduct ivity and NM R Data forHMO3 x H 2 0 Compounds a t H igh Humidi t f

4 3 0 "C), E h ) , EJNMR),M structure n1 m-' kcal/mol kcal/mol

sb cubic KS b03 type 8.7 X lo-' 9.1 f 0.1 2.3 f 0. 4Sb layer 2.2 X 4. 3 f 0.1 2.3 f 0.4Sb pyrochlore 3.2 X 4.6 f .1 2.3 f .4Nb pyrochlore 1.5 X lo4 5.0 f .5 5.3 f 0.8Ta pyrochlore 1.4 X lo-' 6.0 f .1 6.0 f 0.8

From ref 47. with Dermission.

exhibit a three-peak pattern indicative of hydroniumion.47 This does not preclude the presence of freeprotons, so the choice between the two formulas for thecubic polymorph is still not clear.

2. Niobium and Tantalum Oxides

Hydrous niobium and tantalum oxides, as preparedby hydrolysis of the pentahalides or addition of baseto aqueous acid solutions, are amorphous and behaveas hough they are particle hydrates.52 However, theircation exchange capacities, 3.3 mequiv/g for niobiumoxide and 2 mequiv/g for tantalum oxide, are excep-

tionally high for particle hydrates. Cations are ex-changed under fairly acid conditions, indicating a highlevel of hydronium ions in the interparticle water. Thiitype of behavior would be expected of M(V) hydratesbecause the high charge on the metal results in strongbonds to oxygen and strong repulsion of the proton.Anions are also taken up in highly acid solution by thesehydrates but they can be eluted with NH3.

A pyrochlore form of niobium and tantalum oxidescan also be prepared by treating the thallium(1) niobateor tantalate with a ~ i d . ~ ~ * ~ ~ * ~he general formula is(H30+)HM(V)206. NMR spectra indicate the presenceof both hydronium ions and free protons in accord withthis formula.47

A related reaction to the one described above involvedthe preparation of H Nb 03 and HTaO, by acid treat-ment of the corresponding lithium compounds at95-100 0C.55 The transformation is a topotactic oneinvolving a trans ition from the rhombohedral lithiumniobate structure to a cubic perovskite one. LiNb03and LiTaO, are ferroelectric and nonlinear opticalmaterials of considerable industrial and theoretical im-portance. Since crystals of these compounds are oftencleaned in boiling acid, it is likely that surface exchangewill take place. Subsequent heating of the crystal willremove the protons as water, leaving a lithium-deficientsurface layer of composition M205(Li20)1,. Partiallyexchanged LiNbO, is of particular interest since the

proton exchange produces a large change in index ofrefraction of the crystal^.^ Therefore, a more detailedstudy of the structural changes in the systemLil,H,Nb03 was undertake^^!^^^ A number of distinctphases were discovered in this system along with severalphase transitions. An order-disorder mechanism wasinvoked to account for these observations.

3. Proton Conductivity of Hydrous Oxides

Proton conductivity data for the framework hydratesare summarized in Figure 4 nd Table 2.47 The con-ductivity of the protonated species is at least 2 ordersof magnitude higher than that of the corresponding,cation species. This effect has been attribu ted ta dif-

HSbO,. aHzO!CUBIC)

Figure 4. Proton conductivities for HM 03- nH 20 (M Sb, Nb,Ta) compared to other good proton conductors. Values for 1 MHCl are also shown for comparison. (From ref 47, with per-mission.)

ferent mechanisms of conduction of the two ~p ec ie s. 4~The high conductivities of t he protonated frameworkhydrates are thought to result from a Grotthus mech-a n i ~ m . ~ ~ ~ ~n this conduction process hydronium ionsmust reorient to the proper position for proton transferto occur. Reference to Table 2 shows th at for the an-timony framework hydrates the activation energy asmeasured by NMR is lower than that obtained from thedo pe of the conductivity curves. According to Chow-dhry et al.,47 the N M R rocess measures hydronium ionreorientation while the conduction process measuresproton hopping as well as the reorientation. The highactivation energy observed for the protonated cubicpolymorph is the same as th at for K+ ion conductionin KSb 03 (see Figure 8, ref 47). In the potassiumcompound conduction must involve translation of theions. Therefore, in the cubic protonated hydrate theconduction process likely involves translation also (eq4). In the case of the niobium and tantalum hydrates,

E,(a) = E,(NMR) + E,(hopping) + EJtransl) (4)half the protons must be bonded to lattice oxygens.The conduction process may involve hopping of protonsfrom hydronium ion to the lattice, in turn facilitatinglattice-to-water hopping. In such a process both theNMR- and conduction-determined E, values are thesame and therefore must be controlled by the hoppingmechanism.

that the NMR E , valuesfor the pyrochlore hydrates scale inversely as the elec-tronegativity values of the metal atoms. The M-O bondis more covalent in the Sb compound th an in Nb andTa since it has the higher electronegativity. Hence theBr~s ted cidity of the Sb compounds should be higherand the proton more delocalized, accounting for itslower E, value. Particle hydrates also conduct by aGrotthus mechanism when sufficient water is presentin the pores.33

Chowdhry et al. point


6/24

i a o Chemical Reviews. 1988, voi. 88 . NO. I Clearfield

TABLE 3. Data for Mn(IV) Oxides with Tunne l S t ruc tu resmineral name approx formula structural features unit cell dimensions

octahedr a forming 1 x 1 unnelsoctahedr a forming 2 x 1 tunnels

a = 4.39 A, = 2.87 A (tetragonal)a = 4.39 A. b = 9.21 A. c = 2.87 A

pyrolusiter a m s e te

8-MnOlMnOz

nsutitehollandite (a-Mn O,)cryptomelaneeoronad temanjiroitepsilomelane

(romanechite)todorokite

t -Mn02

y-MnO, or y-MnO.BaMn8OI8%H20ideal)KMn.O...xH.O." 1PbMn,016.xH,0NaMn80,cxH,0Ba2Mn,010.xH,0

(K+ ubstitution for Ba*+)(Na,Ca,Mn)Mn,OrxH,O

Mn01.9-Mn01,g8rideally Mn,OaOH

-(orthorhombic)

intergrowths of pyrolusite and ramsdellite a = 9.65 A, = 4.43 A2 X 2 tunnels, lower valent Mn present (I = 9.96 A, = 2.86 Amemhers of hollandite family a = 9.84 A, = 2.86 A

3 X 2 unnels parallel to b axis a = 9.56 A, = 2.88 A, = 13.85 A,8 = 92.5O

3 X 3 tunnels, m onoclinic cell

like y-MnO,, fibrous form

a = 9.15 A, = 2.85 A, = 9.59 A,

a = 4.85 A, = 8.90 A, = 2.80 A8 C i 900

(c = 2.80 X 3 = 8.40 A in the supercell)

NMR s tudies of two particle hydrates, SnO,.nH,Oand TiOz.nHzO, have been carried out in order to gainfurther insight into the conductivity mechanism of thisclass of Three types of protons wererecognized those associated with hydroxyl groups,water, and hydronium ions. Two sets of spin relaxationconstants and correlation times were observed. Thus,the particle hydrate model (section ILA) was extended

to include protons present in micropores of less than100 A (region B) and those present in macropores oflarger diameter (>loo0 A). Surface hydroxyl groups areassociated with both types of pores, which allows ex-change of protons between the different pores to occur.Th e pores are formed by the joining of oxide particlesto form agglomerates containing cracks and fissures.The solution in the macropores is thought to be almostliquid-like, while that in the micropores will be moreconstrained and thus exhibit a longer correlation time.Hydrous tin oxide exhibits two activation energies inits conductivity behavior (Figure 3), with values of 9.3kcal/mol below 288 "C and 5.6 kcal/mol above thistemperature. A t the lower temperature th e conduc-

tivity measures the activation energy for motion ofprotons in the macropores. At higher temperature,motion through the micropores becomes rate deter-m i n i . Hydrous titania (and zirconia) exhibits a singleactivation energy of 6.9 kcal/mol, indicating transpor tin the macropores over the entire temperature range.As water is lost, other transport mechanisms must be-come operative.

4. Hydrous Oxides with Tunnel Structure: MnO,

Th e cation exchange behavior of manganese dioxideis important in connection with measurements of th epotentials in electrochemical cells and batteries.Electrical potential depends upon th e pH of the sur-

rounding solution, which in turn is influenced by therelease or uptake of protons by the MnO, electrode.The cation exchange properties of manganese dioxidewere first reported by GhoshGz nd later expanded uponby Kozawa.= Initially it wm though t th at the ion ex-change reaction took place only on the surface and onthis basis Brenet et al.64365 ormulated manganese di-oxides, precipitated from aqueous systems, as oxo hy-droxides, eq 5. In this formula n is the degree of ox-

MnO,(OH),.mH,O (5)idation (close to 2) and z is a s m a l l fraction. Later workhas shown tha t macroamounts of ions could be ex-changed.".E' Tsuji and Abe6' prepared their exchanger

ho l l and l t e g r o w4 " 0 2

Te P r e S m t O t l Ye " S l l t l k

r-nm2

w r o m l t e rmsaelltte

Figure 5. Various forms of MnO, with tunnel structures. (From:Burns, R. G.; Bums , M. B . Manganese Dioxide Symposmm,Tokyo, 1981; Val 2, Chapter 6).

by addition of Mn(I1) to a KMnO, solution. X-raydiffraction patterns showed that the product had acryptomelane structure in which hydronium ions werepresumably located in the tunnels.

Figure 5 depicts schematically the various forms of

MnO,. A more complete listing is given in Table 3. Itshould be noted that hollandites and other membersof the hollandite family are not true manganese dioxidesbut contain mono- or divalent cations in the tunnels.On reatment with acids, the cations are exchanged forhydronium ions. Thus, theoretically a completely newfamily of ion exchangers is possible. The exchangerdescribed by the Russian workers" was prepared elec-trolytically from MnSO,. This exchanger exhibited ionexchange capacities of the order of 4 mequiv/g. Shenand Clearfield6* also used an electrolytic method toprepare hydrated manganese dioxides. They found thatthe exchange capacity was a maximum for the productsprepared at temperatures of less than 25 "C and de-


7/24

Io n Exchange in Solid-state Chemistry

creased with increasing temperature of preparation.The manganese dioxide prepared at 95 "C exhibitedonly surface exchange. For preparations obtained below25 "C the ideal formula is (Mn01.75)20H-~H20. ll ofthe protons could be exchanged by Li', but lesseramounts were exchanged as the exchanging cation sizewas increased. Th e exchanger is thought to have ay-Mn02-like structure (Figure 5), but its amorphousnature sufficiently disorders the tunnels to allow somelarge ions to be exchanged.

When the Li+-exchanged form of the y- Mn02 washeated to 500 "C, it was converted to the spinel Li-Mn204.68 The reaction could be formulated as

(Mn01,75)20Li - iMn204 + 1 / 4 0 2 (6)Treatment of the spinel with dilute acid converted itto HMn204 with retention of the spinel structure. Thehigher the temperature to which the lithium-exchangedmanganese dioxide was heated, the less Li+ was releasedon acid treatment. In the spinel form, HMn2 04, heexchanger is highly specific for Li'.

The sodium ion exchanged phase of the electrolyti-cally prepared hydrous Mn02 formed an as yet un-

identified phase on heating to 500 "C. However, incontact with boiling water this phase was converted tobirnessite with uptake of water. Birnessite is a layeredform of Mn02, usually termed 6-Mn02, which containssodium ions.69y70 The formula proposed by Shen andClearfield,G8 Na4Mn14027-9H20, nd that of the pro-tonated form obtained by them from the sodium phase,H4Mn&8*7H20, differ from the corresponding formulasgiven earlier by Gio~a no li .@-~ ~ any sodium, potassium,or other cation phases of manganese are known, somewith tunnel structure and others with layered struc-

some of which exhibit ion exchange behav-ior.67*7679 t may thus be possible to prepare protonforms of many of these compounds and exchange dif-

ferent ions into them. Subsequent heat treatment maythen produce new as yet undiscovered phases. Thesame may be true in the case of other framework-typehydrous oxides.

Chemical Reviews, 1988, Vol. 88 , No. 1 131

I I I . Ion Exchange in Mixed Oxides

Th e ease of ion exchange in classes of compoundssuch as zeolites, hydrous oxides, and acid phosphates(section IV) may be understood in terms of the presenceof interconnected cages, channels, or layers of sufficientdimensions to allow ion transport. Fast ion conductorssuch as P-alumina (section V.B) and NASICON (sectionV.A) also undergo ion exchange as well as fast iontransport. They have rigid frameworks but the mobilecations ,occupy only a part of the sites allowing transportinto empty sites. A preliminary consideration ofthermodynamic and kinetic factors led England et al.37to conclude that ion exchange may be a widespreadphenomenon in inorganic solids, well below their sin-tering temperatures, and that high concentrations ofvacancies may not be necessary. These points will bediscussed in what follows.

A. Ternary Oxides

1. ABOrType Structures

In /3-NaA102 the Na+ and A13+ ions occupy a lternatetetrahedral sites in a hexagonal close-packed wurtzite

TABLE 4. Ion Exchange &actions of Layereda-NaCr02-Type Structuresn

subs trate structu re reagent Droduct structu rea-LiMOz (M = a -NaCr02 Pd + PdClz PdMOz delafossitea-LiFeOz a-N aC r02 CuCl(1) CuFeOz delafossitea -L iM02 (M = a-N aC r02 AgN03(1) AgMOz delafossite

a From ref 37 , with permission.

Co, Cr, Rh)

Co, Cr, Rh)

structure.80 Geissner et a1.81~82 reated ,f3-NaA102 inmelts containing Ag+, Cu+, and T1+ and were able tototally replace the Na+ by utilizing a large excess ofcations in the melt. Similar reactions were obtainedwith aluminates which possess a cristobalite structure.

In the a-NaCr02 structure both the Cr3+ and Na+ areoctahedrally coordinated in alternate layers. Lithiumion exchange was carried out in excess molten LiN03for 24 h at 300 "C. Close to 100% of the Na+ wasexchanged with a corresponding formation of LiCr0237Aqueous exchange of H+ for Na+ was also carried outat room temperature with a 25% HNO, solution. A92% exchange was observed with formation of HCr-

O2.0.92H20. Heating to 400 "C yielded HCr02. Table4 summarizes work by Shannon et aLE3 n which theyconverted lithium salts of several metals with the a-NaCr0 2 structure to delafossite-type phases.

2. Spinels

Joubert and DUri84is5 have reported an ion exchangereaction for the spinel (Lil.5Zn,.5)tet(Ge4.5Lil.5)oct0 oproduce (Zn3)tet(Ge4.5~1.5)oct012.he reaction was car-ried out in molten ZnS04 at 500 "C for 3 weeks. Asimilar exchange reaction was carried outm on Li4Ti5014,in which the ion arrangement is (Li3)tet(LiTi5),t012. Atotal of 68% of the Li' was replaced by Zn2+ n 4 daysat 500 "C. In all the spinel reactions, vacancies arecreated only in the octaheral holes. Exchange does notproceed to the point where tetrahedral vacancies arecreated.

We have already discussed the case of LiMn204 nsection II.B.4. The ready exchange for H+ occurs be-cause of incomplete crystallization of the compound at500 "C which provides defects th at facilitate exchange.It should be mentioned that Huntere6 claimed to haveremoved all the Li+ from LiMn204 y acid treatmentto yield X-Mn02. He explained this transformation onthe basis of a surface reaction to remove Li+ followedby a disproportionation of Mn3+ o form soluble Mn(I1)and insoluble Mn(1V). A possible explanation for thedifference in behavior reported by Hunter and thesimple ion exchange reactiod8 lies in his use of strongacid solutions whereas the exchange reaction was car-ried out in 0.1 M acid. Another case in point is that ofFe304 or magnetite. It has an inverse spinel structure(Fem),(FeDFem),Ok Natural magnetite has been usedto remove radioactive cations from nuclear wastestreams.87 More recently, magnetite was prepared insitu and found to remove 5-50 times as many ions asdid ferric hydroxide.% Presumably the sequestered ionsreplace iron ions in both the tetrahedral and octahedralholes but no evidence for this was presented. However,it might be expected from what has been already de-scribed in this report that poorly crystallized spinelsmay readily exchange ions with aqueous electrolytes


8/24

132 Chemical Reviews, 1988, Vol. 88 , No. 1 Clearfield

a b

C

+ f

Figure 6. Idealized structu re of t i tana tesand t i taniobates witha layer structure: (a) A,Ti,,M,O,; (b ) T1,T i409; (c) N a2 Ti 30 7;(d ) K?Ti9bOI4; e ) CsTi2Nb07; 0 KTiNbOS. (From ref 114, withpermission.)

and well-crystallized spinels would behave similarlywith molten salts. Such reactions may prove useful indoping ferrites.

3. Metal Titanates

Alkali metal ti tanates have been synthesized whichconform to eit he r of the following two series:M2Tin02n+l nd Na4Tin02n+2. n the former series,members with n = 1-9 have been preparedsw7 and inthe latter series, compounds with n odd (n = 1, 3,5 ) *loo K2Ti205, NazTi3O7, Na2Ti409, nd Na4Ti5012have layered structures of the type shown in Figure6b,c.lo1 On the other hand, Na2Ti6Ol3 nd K2Ti6Ol3have a rigid structure in which three octahedra shareedges and are joined through corners to similar chainsof three octahedra to form tunnels (Figure 7).94 Ionexchange behavior has been demonstrated in K2Ti205,Na2Ti307, nd M2Ti409. In fact alkaline earth tetra-titanate s can be prepared from T1Ti409 by ion ex-change.92

Treatment of the tri- and tetratitanates with HC1converted them to H2Ti 307 nd (H30+)HTi4Og, e-spectively.lol Dehydration of the latte r compoundyielded H2Ti8Ol7, which was assigned a tunnel struc-ture.lol Further dehydration resulted in the formationof Ti02(B), a new form of titanium d i o ~ i d e . ~ ~ , ' ~ ~ twould be interesting to determine whether additionalnovel oxides would result from partial and completedehydrat ion of protonated oxides under mild condi-tions. Reversibility of exchange was found to be dif-ficult as even the use of 1 M base only partially ex-

Figure 7. Na2TieO la n projection. Th e corners join adjacen tsheets to form a framework enclosing tunnels. Tw o sodium ions(circles)are present in a unit tun nel which contains three pseu-docubic positions. (Fro m ref94 , with permission.)

changed out the protons. Apparently the shrinkage ofthe cell volume on protonation makes reversibility ofexchange difficult. However, subsequent work pro-duced half, three-quarter, and fully exchanged Na+ andK + hases from the acid forms.lo3-lo5 A one-quarterexchanged phase was also obtained in the K+ ystem.Structural explanations for these exchange data havebeen presented. The tri- and tetratitanates in the acidform were also found to form amine intercalates by ionexchange of alkylammonium salts for H+.lO6 Onlymodest conductivities were exhibited by the tri- andtetratitanates, but a marked improvement was observedwhen a small amount of Nb5+ somorphously replacedTi4+. This increase was attributed to a slight expansionof the lattice.lo7

A new sodium titanatelo8 was prepared by the actionof NaOH on anatase. It has been shown to approximatethe formula Na4Ti9020.nH20109 ut t o differ from Cs4-Ti4020110 and Ba4Ti902@ The latter compound is re-

ported to have a hollandite structurelll whereas thehydrated sodium nonatitanate has a layered struc-ture.lW Complete exchange of H + for Na+ is readilyaccomplished. The nonatitanate shows promise as anexchanger for the removal of 90Sr2+ rom nuclear wastestreams.l12 Uptake of Sr2+ by Na4Ti9020 ielded Na2-(SrOH)2Ti9020 ecause of the high pH produced byhydrolysis of the sodium phase whereas H4Ti9OZ0 asconverted to SrzTi9020.113 Heating these exchangedproducts resulted in the following reactions:

(7)r2TigOz0- SrTi03 + 7TiO260 0 sC

870 "CNa2[Sr(OH)12Ti9O20 -

Na2Ti6Ol3 + unidentified Sr phase + H20 (8)6 . Quaternary Oxides

1. Mixed-Metal Titanates

Several mixed niobium and tantalum titanates havebeen synthesized and shown to possess ion exchangeproperties. These include ATjNb05,1157116sTi2Nb0,,117A3+Ti5Nb014,118 Ti3Nb09,119 nd a nonstoichiometricseries K1-xTil-xNbl+r05,115,116 elated to the firs t listedcompound. The structures of these compounds arelayered ones built of octahedra sharing edges andforming a double plane as shown in Figure 6a. The


9/24

Ion Exchange in Solid-state Chemistry Chemical Reviews, 1988, Vol. 88, o. 1 133

Figure 8. Topotac t ic format ion of the emp ty tunnel s t ructureTi2Nb 2O9 y condensation of two [TiN bO J layers from the layerox ide HTiNb09 . (From ref 114, with permission.)

other structures are derived from this one by shearingthe layers in the direction perpendicular to its plane,every third octahedron for CsTi2Nb07 nd K3Ti5NbO14and every second octahedron for K1-xTil-xNbl+x05.114In some cases the layers are parallel but in others, suchas ATiNb05, the layers are related by a glide plane.'17In each case the A+ ions lie between the planes andserve to bind them together by electrostatic attractions.

Protonated forms of the titanoniobates can be pre-pared by treatment of their alkali metal forms with acid.ATiNb05 oxides are easily exchanged, yielding th eunhydrated protonic oxide HTiNb03114,115 n contrast,CsTi2Nb07 equires treatment with concentrated acidand forms a hydrate, HTi2Nb07*2H20. t dehydratesat 100 "C. A3Ti5NbOI4 ompounds are much less re-active and cannot be completely exchanged by acids.The highest protonic content produced by direct reac-tion is I(0.5H2.5Ti5Nb014. owever, by exchanging Na+for K+ in this compound one obtains H1.5Na1.5Ti5Nb014and HNa2Ti5NbO14, which in acid solutions are con-verted to H3Ti5NbO14-H20. emoval of the water de-stroys the compound.

An interesting product can be prepared by splittingout water from HTiN b0 3 The anhydrous compoundTi2Nb2Og s formed because the layers come togetherin the manner shown in Figure 8. The resultant mixedoxide contains tunnels and because of its open structurecan intercalate lithium and sodium to formAxTi2Nb20g-type ompounds.l14

Because of their acidic character, the protonic oxidesreadily intercalate amines, and an extensive literaturehas developed on this Lal and HowelZ3have measured the conductivity of protons in HTi2-NbO7*2H20. Th e water molecules in this compoundform a double layer between the oxide sheets, whichwere expected to facilitate proton transport. Conduc-tivities (17-21 "C) were (2 f 1) X lo4 Q -l cm'l aspressed pastes and (9 f 3) X to (8 f 4) X lo4 Q -lcm-' as pressed powders.

A series of compounds has been prepared over theyears by isomorphous replacement of Ti4+ n Ti02 byM2+ and M3+cations. The general composition of thesecompounds is AbTi2-xMnx04 M(I1) = Mg, Zn, Ni, andCu and A is an alkali metal) and AXTizzMmxO4 M(III)= Fe(II I), Mn(III)].124 Th e structures consist of bothTi 06 and M0 6 octahedra arranged in two differentways. When 2x is less than 0.5 the structure is of thehollandite type with the A+ (K+,Rb+, T1+, Cs+) ions inthe A t higher values of 2x (0.6-0.8) thecompounds have a layered structure in which the Ti-Mg octahedra share edges and corners in a corrugatedf a ~ h i 0 n . I ~ ~ngland et al.I3O determined the ion ex-change behavior of one of the latter type compounds

Figure 9. Schematic representat ion of the s t ructu re of KC a2-Nb3OI0. Th e shaded squares represent NbO s octahedra in pro-jection, and th e circles within the layers are the calcium atoms.T h e c axis is perpendicular to the layers. (From ref135, withpermission.)

with composition Cs0.7Ti1.65Mg0.3504. ince there are1-2x, or in this case, 0.3 mol of vacant Cs+ sites, ion

exchange should be facilitated. Li+, Na', and K+ couldbe exchanged for Cs+ from fused-salt reactions in the300-400 "C temperature range in 1 2 h. Aqueous ex-change required up to 4 days for complete reaction.Concentrated or even strong acid solutions dissolve thetitanate, but the protonated form could be obtained byuse of dilute acid solutions.

2. Complex Niobates and Tantalates with Tunnel orPer0 vskite Structures

Recently, Dion e t al.131 synthesized a new family oflayered oxides with the general formula ACa2Nb3010 A= K, Rb, Cs, T1).131-133 he layers consist of perovskiteslabs separated by A+ ions (Figure 9). Because suchcompounds have a lower layer charge density thanstrontium titanates of the type Sr2[Srn-1Tin03n+l],134 onexchange was observed to occur under relatively mildconditions whereas the strontium titanates did not reactunder similar conditions. Following this discovery,Jacobson e t al.135were able to prepare a series of suchlayered perovskites with the general formula K-[Ca2Na,-3Nb103n+l] 3 5 n I ). The layer thicknesswas found to increase by about 3.9 A for each increasein n, eaching a value of 27 A for n = 7. Protonic andhydronium ion phases of each series member could beprepared by exchange of the K+ ion phases in 6 M HC1at 60 "C for 16 h.135J36 Alkylamines could then be in-tercalated into the solid acids, forming interesting bi-layer s t r u ~ t u r e s . l ~ ~ J ~ ~

Members of the layered perovskite family with n =2 were prepared138 by substituting La3+ or Ca2+. Theresultant compounds, ALaNbz07 A = Li, Na, K, Rb,Cs, NH,), onsist of double perovskite slabs with A+ ionlayers between them. The solid acid HLaNb20, wasobtained by acid treatment followed by dehydration.It exhibited Bransted-type acidity and intercalatedamines.

Two interestin series of niobates with proposedformulas of Al2M$s&Yw30W nd A10M29.2078 have beens y n t h e s i ~ e d . ' ~ ~ J ~ ~he former series has members withM(V) = Nb and Ta and M(V1) = W and Mo while thelatter contains M(V) as either Nb or Ta. The structure


10/24

134 Chemical Reviews, 1988, Vol. 88, No. 1 Clearfield

T

I

~

I

CI

I

Figure 10. Schematic representation of the stacking of groupsin A12M33090. From ref 139, with permission.)

[ i oI

Figure 11. Represen ta t ion of the s tacking a long thec axis ofthe hexagonal cell of the com poundAloMzs,z07s (From ref 140,with permission.)

of the A12 M33 0w- t~e ompounds is built up of M7O18

pyrochlore-like units141 nterleaved withM4O12

units(Figure 10) 139.142 Similarly, the A 1 J ~ f B . 2 0 7 8 ompoundsare built up of intergrowths of &018-M03-M+015 unitsas shown in Figure l l . 1 4 0 Phases which could not beprepared by the direct reaction of oxide mixtures, suchas the hydronium and silver ion phases, were obtainedby ion exchange. An interesting solid-solid exchangereaction in which K+ replaced TI+ according to eq 9 was

T112M30M3090 KC1 - 12M30M3090 TlClt (9 )possible due to th e volatility of TlC1. This reaction issimilar to those carried out by Clearfield et al. withzirconium phosphate143 nd H-zeolites.14 These latterreactions depended upon the volatility of HC1.

TA B L E 5. Some Important Acid Salts of Tetravalent Metalswith Layered Structureof the a-Type

inter- ion exchangelayer capacity

compd formula dist,A mequiv/gtitanium phosphate Ti(H P04 )z-H z0 7.56zirconium phosphate Zr(H P04)z .Hz0 7.56

germanium(1V) phosphate Ge(HP04)z-Hz0 7.6tin(1V) phosphate Sn(H P04)2 .Hz0 7.76

t itanium arsenate T~ (H A SO ~) ~. H~ O.77zirconium arsenate Zr (HA s04) z~H z0 7.78tin(1V) arsenate Sn(HA s04)Z.H 20 7.8

hafnium phosph ate Hf(HP0 4)yHzO 7.56

lead(1V) phosphate Pb(HPOd)z-H,O 7.8

7.766.644.177.086.084.79

5.785.144.80

C. Conclusions

In the foregoing section we have seen th at ion ex-change can occur in a wide variety of oxide structuretypes, including framework or tunnel structures, layeredcompounds, and even lattice structures with close-packed oxygens. Th at facile ion exchange may occurin structures which are easily expandable (layeredstructure or hydrate) and those with tunnels or grossnonstoichiometry is easy to comprehend. To the verylarge number of such compounds described above must

be added those that will be described below such as thegroup IV (groups 4 and 14) and related phosphates,/3-alumina, and silicophosphates. It is then clear thatthere is an abundance of inorganic compounds whichmeet the requirements for facile ion exchange. How-ever, the reader should also keep in mind that manycompounds not normally considered to be ion exchan-gers, such as the spinels and the AB0 2 class of com-pounds, may be induced to exchange under forcingconditions.

I V , Group I V (Groups 4 and 14 ) Phosphatesand Related Compounds

A. a-Type M(I V ) Layered Phosphates1. Structure and Io n Exchange Behavior

This group of compounds consists of phosphates,arsenates, and mixed phosphate-arsenates of bothgroup IVA (group 14) and group IVB (group 4) ele-ments. A partial list of the a-type compounds is givenin Table 5. A number of comprehensive reviews onvarious aspects of the properties and behavior of thesecompounds have been published recent1y,12*148 ncludingmaterials ion catalysis,147 ndmembranes.149 Therefore, we shall confine ourselves tothose aspects of this subject which have relevance tosolid-state chemistry and to newly synthesized com-

pounds and their properties. By way of introductionwe remind our readers that members of this class ofcompounds are obtained as amorphous gels when pre-pared by rapid precipitation in the cold, but refluxingin solubilizing media slowly transforms them into~ r y s t a l s . ~ ~ * ~ ~he slow transformation, in the case ofzirconium phosphate, allows the operator to build in thelevel of crystallinity desired for a particular purpose.The crystals have a layered structure in which the metalatoms lie nearly in a plane and are bridged by phos-phate groups as shown in Figure 12. Thus threephosphate oxygens bond t o metal atoms so that theremaining phosphate oxygen, which is bonded to aproton, points into the interlayer space. The layers are


11/24

Io n Exchange in Solid-state Chemistry

60 0


NoZr, (PO, )

Zr P, 0,

Figure 12. Idealized structu re of a-zirconium phos phate showingone of the zeolite cavities created by th e arrang emen t of the layers.(From Alberti , G. In Study Week on Membranes; Passino, R. ,Ed.; Pontificiae Academie S c ie n t" Scripta Varia: Rome,1 9 7 6p 269 (with permission)).

f C t+ Na+ (Aq) e 7 . 6 i

7

.)7 . 6 i , 1;**i

IZr(HP04)2*H20

ZrNaH(POu)2 ' 5H20

Figure 13. Schematic representat ionof layer expansion by ionexchang e according to Alberti.14' Region between da shed linesis solid solution region. (After ref148, with permission.)

staggered, forming a network resembling a hexagonallyshaped cavity (heavy outline in Figure 12) with thewater molecule in the center.

In this a-layer the metal atoms are six-coordinate andthe phosphates tetrahedral, forming T-O-T-type layersas in smectite clays. The major difference is that thephosphate groups are in inverted positions relative tothe silicate groups in clays. As a result the negativelayer charge is concentrated on the proton-bearing ox-ygens rather than spread over the many silicate oxygensin the clay interlayer boundary. Although the layersare held together only by van der Waals forces, theinterlayer distance is small and the bottlenecks providejust enough space for an unhydrated K+ to diffuse intothe interlamellar spacing. Thus , the crystals exhibitsieving behavior in acid solutions, and ions larger thanK+ re exchanged at an appreciable rate only in alkalinesolution. Under these conditions sufficient protons onthe layers are neutralized to allow swelling of the layersto admit larger ions.26$27

Exchange takes place in the crystals by diffusion ofions from the outer surface inward with advancingphase boundary. The layers expand to accomodate notonly the ions but waters of hydration. Until the ex-change process is complete, each particle or crystallitecontains two phases, the original proton phase in theinner region and the cationic phase in the outer regionsof the crystallite. A l b e ~ t i ' ~ ~ as developed a pictorialdescription of this phenomenon. This is illustrated inFigure 13 . The wedge-shaped region connecting theinner proton region and the outer cation region containsboth counterdiffusing ions. Th at the two phases coexistin the same particles was shown by heating the partiallyexchanged crystallites to 110-120 "C. A t a level of 25%exchange of Na+ in zirconium phosphate, the phasespresent are ZrNaH(P04)2.5H20 and Zr(HP04)2.H20.After heating, the sodium ion redistributes itself to formtwo new phases:150 ZrN%.2Hl.a(P04)2 nd ZrN%.aHl.2-

NoZr, (PO,),

H

IO

Figure 14 . Phases formed by sodium-exchanged a-zirconiumphosph ate. (From ref 145, with permission.)

TABLE 6. Identification of the Phases in Figure 1 4 O

designationABCDEFGHNa-INa-I1C-ZrP7-ZrP

compos of phaseZrNaH(P04)2.5H20ZrNaH(P04)2 .H20ZrNaH(P04)zZr(NaP04)2.3H20Zr (NaP04)2 .H20Zr (NaP04)zZr (Na P04 )2Zr(NaP04)z

Zr (HPO,),Zr (HP04)2

ZrN%.2H1.8(P04)2ZrN%.8H1.2(P04)2

interlayerspacing,A

11.87.97.39.88.48.47.67.66.867.27.417.15

a From ref 145, with permission.

in the correct proportions determined by thesodium ion content. Th e same type of redistributionoccurs with amine intercalates which are not fully in-tercalated. Regions with more amine in contact witha region containing less or no amine, as determined bythe interlayer spacings of the phases, redistribute theamine to form a single phase on heating.15' Counter-

diffusion of the ions is not necessary as some regionsnear the crystal edges will always be free of the ingoingions. Thus, we picture the ingoing ions as displacingthose ahead of them into adjacent cavities in a con-certed movement, with an equal number of displacedions pushing those near the edge out into the solution.

2. Phase Changes of a-Layered M(IV) Phosphates

Not only are new phases obtained on exchange butthe hydrated ion containing phases also dehydrate instages when heated. We have already mentioned re-distribution of ions on heating the partially exchangedsodium ion phase in the previous section. Figure 14 andTable 6 list all the phases formed in the zirconium


12/24

136 Chemical Reviews, 1988, Vol. 88, No. 1

phosphate-sodium ion system.145 Figure 14 is not anequilibrium diagram as the tempera ture ranges of sta-bility for each phase were determined by TGA andX-ray patterns taken at room temperature. A morecomplete explanation of the phase diagram is givene1~ewhere . l~~ owever, we note here t ha t the half-ex-changed phase A, when heated to 500 "C, forms Na-Zr2(P04),. This triphosphate is not a layered compoundbut has a three-dimensional framework structure andis the end member of the NASICON family of fast ionconductors (see Section V.A).15*

Even more complicated behavior is exhibited by thefully exchanged phase D, Zr(NaP04)2.H20, s evidencedby the reaction scheme in eq 10. Phases F, G, and H

D-E-F-G-H-NaZr2(P04)3 I (10)

are anhydrous and retain a layered structure. However,phase I was recently s h ~ ~ n ~ ~ ~ - ~ ~ ~o be a solid solutionof composition Na1+4xZr2-x(P04)3 ith 0.95 1 1 .88.The relative proportions of the triphosphate and phaseI depend upon the tempera ture of heating and time.153

Diagrams similar to those of Figure 14 have beenpublished for Li', Mn2+, Zn2+, and Cu2+.157-159 hesediagrams are especially interest ing in the light of thediscovery of fast ion conduction in compounds of thetype Nal+xZrzSixP3-x012.160J61n the range 1.8 I I2.3 such compositions exhibit conductivities of the orderof 0.2 Q- l cm-l a t 300 0C.160 One end member of theseries is NaZr2(P04)3, hich can be prepared by heatingZrHNa(P04)2 o 500 0C.27 The reaction in the case ofLi+ is more complicated. Heating ZI - ( L ~ P O ~ ) ~ o 600 "Cyields a triphosphate-like phase which has not yet beenclearly characterized. However, further heating to1000-1100 "C converts it t o LiZ r2(P 04)3 lus lithiumphosphates. Refluxing in acid then dissolves the lith-ium phosphates and converts the lithium salt to H-Zrz(PO4),. Triphosphate phases are also obtained withdivalent cations in a similar manner.159J62 Many of thephases formed with divalent ions have not been wellcharacterized, and no similar exploration of phasesobtained with other group IV (groups 4 and 14) phos-phates has yet been carried out. Study of such systemsmay provide a rich field for investigation of new ionexchangers, fast ion conductors, and new solid solutionphases with interesting magnetic properties.

3. Gas-Solid and Fused-Salt Exchange Reactions

The a-layered compounds are able to undergo ionexchange with anhydrous ~ a 1 t s . l ~ ~ or example, whenNaCl is mixed with anhydrous a-ZrP and heated, HC1is evolved and sodium-exchanged phases are obtained.The phases formed are Na-I, Na-11, C, and F or G(Table 6), respectively, as the sodium ion content of thesolid increases. The reactions are reversible and en-dothermic since hydrogen ion is selectivity favored bythe exchanger. Similar reactions have been carried outwith zeolites144 nd dry organic resins in the H form.

The rates of the reverse reactions of Zr(NaPO,):, andHC1 have been measured.163 Phase F (Table 6) was thestarting phase and the sequence of reactions is as fol-lows:

Zr(NaPO,):, (F)

lo o "C 200 "C 260 "C 800 "C 1100 o c

(1) (2 ) (3 ) (4 )C - a-I1 - a-I -

!?--Zr(HP04)2 11)

Clearfield

The reactions were diffusion controlled and could befit by an equation for a spherical particle in a well-stirred fluid of limited volume. Diffusion coefficientsare of the order of cm2/s in the temperature range150-200 "C with activation energies of 10.3, 12.7, and15.9 kcal/mol for reactions 1, 2, and 3, respectively.Extrapolation of the diffusion coefficient values to 25"C yields coefficients of the order of cm2/s. Thus,the gas-solid reactions are considerably slower than ionexchange reactions in solution, which typically havediffusion coefficients of the order of cm2/s.164

Ion exchange in fused salts must be carried out withsalt forms of the exchanger because release of protonswould be followed by decomposition of the acid formed,or at high enough tempera tures, decomposition of theexchanger. The ingoing ion usually forms a solid solu-tion with a low solubility limit. When this limit isexceeded, a phase transformation occurs accompaniedby a vertical line in the composition isotherm since twoimmiscible phases are now present. The isotherms arecharacterized by hysteresis loops when the ingoing ionexpands the interlayer distance ap p r e ~ i a b l y . l ~ ~ J ~ ~ nthe reverse reaction the smaller ion is able to accom-modate itself in the space left vacant by the outgoingion and no change in layer spacing occurs. In somecases the original phases (or the phase obtained fromone complete exchange cycle) have a large enough in-terlayer spacing to accommodate all ions without ap-preciable interlayer spacing change. In such cases theexchange is reversible.

Proton conduction properties of t he layered phos-phates are presented in section V.C.2.a.

We have described in this section some aspects of thechemistry of the a-layered forms of metal phosphates.There are in fact several other layered types of groupIV (groups 4 and 14) phosphates so that this class ofcompound is exceedingly large and varied.l~~~

B. Group V (Groups 5 and 15) Phosphates

Recently, Piffard e t al.167 repared KSb(P04)2 y thesolid-state reaction of a mixture of K2CO3, NH4H2P04,and Sb205. This compound has a layered structure inwhich the layer is very similar to tha t of a-zirconiumphosphate. Extraction of the K+ in strong acid yieldedHSb(P04)2.xH20,168 hich behaves as a strong acid andion exchanger in much the same manner as zirconiumphosphate, but with half the exchange capacity. Atantalum analogue TaH(PO& was prepared earlier bythe reaction of a Taz05 with H3 P0 4 at elevated tem-p e r a t u r e ~ . ~ ~ ~ince tantalum compounds readily hy-drolyze in basic solution, the alkali metal salts were

prepared by a solid-solid reaction170200-400 "C

TaH(P04)2 + MCl- Ta(PO& + HC1(12)where M = Li+, Na+, K, Rb+, and Cs+. This reactionis very similar to those carried out with zirconiumphosphate.143 However, the Russian workers also usedhydroxides in place of chlorides to facilitate the reac-tion.

Niobium phosphate, formed by dissolving Nb metalin H F followed by treatment with phosphoric acid, hasa composition of NbO(P04).3H20. It is a layered com-pound and able to intercala te amines.17' In concen-


13/24

Figure 15. Schematic representation of the layer structure inH3SbSP20,,. (From re f 173, with permission.)

trated phosphoric acid, the compound NbO(P04)-3H3P04.5H,0 is obtainedlT2 n which H3P 04 s inter-calated between NbO(P04) ayers. Exchange of phos-phoric acid for other intercalates has been demon-strated.

A novel layered phosphatoantimonic acid of compc-sition H3Sb3P,0i4.xH,0 has also been prepared fromthe corresponding potassium salt by acid treatment.lT3It has a layered structure consisting of SbO, octahedraand PO4 etrahedra as shown in Figure 15. As in zir-conium phosphate the phosphates are of the mono-hydrogen type, ut there are only two phosphate groupsand three hydrogens per formula I t is postulated thatthe th ird hydrogen atom is trapped on P-O-Sb or Sb-

0-Sb framework oxygens. In spite of the difference inthe hydrogen ion bonding, all the alkali metals exceptLi exhibit a single end point when H,Sb3P,0i4 is ti-trated with the respective alkali metal hydroxides. OnlyLiOH forms both a two-thirds exchanged phase and afully exchanged phase.

V. Fast Ion Conductors and Ion Exchange

Many of the ion exchangers d d n the previoussections are ionic conductors, but very few exhibit fastion conduction. Contrasting behavior is shown by fastion conductors such as @hmina and NASICON, whichexchange ions bu t slowly from aqueous media and re-

quire forcing conditions for rapid exchange. As we shallsee these conductors contain partially occupied, inter-connected channels or cavities. In th e case of iontransport, barriers exist to the movement of ions fromone site to the next. The ion may be considered to besituated in a potential well and require energy, E,, tosurmount the barrierJT4 This barri er may take the formof narrow passageways from one cavity to another ormovement from a lattice site to an inters titial site orcreated vacancy. Conduction requires single-ionmovements, often in a concerted manner. In most or-ganic resins counterdiffusion of ions can take place asthe cavities in which ions reside can be quite large.'T5Solid electrolytes such as @-alumina or NASICON do

Chemical Revlews. IS88, Vol. 88. No. 1 137

TA B L E 7. Time ( t ) To Achieve F ( t ) / F ( - ) = 0.95 forSpherical Par t ic les of Radius r = 100 pm'

D, mz 3 10- lo4 1CP lo-'*t 2.5 s 4.2 min 6.9 h 69 h 29 days 290 days

e-ples

Na+ in m olten NaCl (1300 K)Lit in 1 M LiCl(aq) (300 K)Na+/Li+ in Amberlite IR1 (300 K)Na + in Na ,%alumina (300 K)Na+ in N a &alumina (600 K)NH ,+/NE t,+ in phenolsulfonic acid (300Na+ in NaCI ( intrins ic region, 900 K)Na' in NaCI (ex trin sic region , 650 K )Na+ /Tlf in analcite (373 K )

a From ref 37. with oermission.

D, mz s-'

1.3 X loJ3.0 X lob4.0 X1.8 x 1065.0 X [email protected] x 10'01.0 x 10-124.5 x 10-13

2.1 x 10-3

"kJ/mol

30~

161621

1737462

not swell and ion transport mus t occur by a concertedor inters titialcy mechanism such as already suggestedfor zirconium phosphate. Thus, counterdiifusion of ionscannot occur in these materials and E. values may behigh for exchange at moderate temperatures. Thebarriers to exchange are overcome by using forcingconditions, such as exchange from fused salts, or underpressure, or heating in concentrated aqueous solutionswere the exchange rate may then become quite appre-ciable. We have already seen that th e exchange ofprotons for cations in many inorganic compounds re-quires strong acids at, or near, boiling temperatures.

Diffusion coefficients for ion conduction involve onlysingle-ion coefficients, bu t those for ion exchange rep-resent mixed or interdiffusion coefficients. England etaIJ7 have calculated the time required to achieve 95%complete ion exchange for a range of diffusion coeffi-cients. The results, including examples, are collectedin Table 7. The implications of these da ta are clear;ion exchange is a necessary, but not sufficient conditionfor fast ion conduction.37 However, even if the diffusioncoefficients are small at room temperature they can heincreased by several orders of magnitude a t sufficientlyelevated temperatures, thus producing extensive ionexchange even below the sintering temperature. How-ever, the converse may not be true. Many compoundswhich exhibit facile exchange may still be poor con-ductors a t elevated temperatures because the barr iersto conduction remain. For example, many layeredcompounds which exchange readily a t room tempera-ture are poor ion conductors because the interlayerdistance may decrease with increasing temperature,resulting in constricted passageways for the ions. Theseprinciples will be illustrated in what follows.

A. The NASICON System

NASICON (Na super ion conductor) is the acronymfor the solid solution series of compositionNal+zZr,Si,P,O1z. The structures of the phosphateend member ( x = 0 ) and th e silicate end member ( x =3) are A portion of the phosphate structure.is shown schematically in Figure 16. I t is rhombohe-dral, R3c, and consists of Zr2(P04)3- nions (shadedportion) in which the zirconium atoms are octahedrallycoordinated by oxygen atoms from six different phos-pha te groups. The anions are then linked together inthree dimensions by the remaining phosphate oxygens.Thii arrangement produces cavities (labeled A in Figure16) in which the sodium ions reside.


14/24

138 Chemical Reviews, 1988, Voi. 88. No. 1 Clearfield

c m

Figure. 16. Projection onto the ac plane of the NaZr,(PO,),structure showing cavity built up by linking of PO4 and ZrO,groups. S haded portion represents the Zr 2(P0& ion. (From ref160. with oermission.)

. .. . /I / - /

Figure. 7. Sche matic representation of the condu ction planein p-alumina showing the various ion sites. The oxygen atompositions are represented by crosses, the m id-oxygen (mO) sitesby filled circles, Beevers-Ross (BR ) sites hy unfilled circles, andthe anti Beevem-Ross (A BR) sites hy squares. Heavy lines denotethe unit cell and dotted lines the 3 X 3 supercell.

Na,ZrZSi,POlz in which th e bottlenecks are muchFigure 16 illustrates only half the unit cell. The other

half is identical with the one shown but is shifted by1/2c. This creates three more cavities between the twohalves which are empty. As the larger silicate groupsreplace phosphate, the additional sodium ions requiredto balance the charge fill up the cavities. At the sametime the bottlenecks between cavities enlarge"' so thatE. is reduced and the conductivity increases. Compc-sitions for which 1.8 < x < 2.3 are monoclinic and ex-hibit the highest conductivities.

The NASICON system has been examined mainly forits conduction properties, but H o r ~ g ' ~ J ' ~ ound thatboth Ag+ and Li+ could he' exchanged for Na+ in Na,-

ZrzSizPO1z n molten salts. Direct synthesis of Li+forms has not been achieved except for the end member( x = 0). The lithium ion phase, Li3Zr2Si2POI2, btainedby exchange, exhibited conductivities 4 orders of mag-nitude lower than the corresponding sodium ion phase.This was attributed to th e small size of Li+. It wassugge~ted"~ hat the high charge-bradius value for thision results in strong bonding to the framework oxygens.In contrast, sodium ions are located near the center ofth e cavities and thus are more weakly bonded. Thisidea gains some credence from the observation thatLiTi,(PO,), is a much better Li+ conductor than isLiZr(P04)3.179 he unit cell dimensions for the titaniumcompounds are considerably smaller than those of th e

zirconium compound, indicative of smaller cavities witha greater chance of weaker, more uniform electrostaticbonding to oxygens forming the cavity. Both titaniumand zirconium compounds readily exchange protons forLi+ on treatment with acid," indicating th at the

larger.lS1 The protonated NASICONs exhibit reason-ahle conductivities at 300 "C (-10-3-10-4 n-' cm-').

Complete isotherms for Na+-K+ exchange in NASI-CONS with x = 0, 1, and 2 were determined'82 at 368& 4 "C by using mixed nitrates. In each case K+ waspreferred over Na+. According to Eisenman's theory'"(discussed in section VILA), the larger ion is preferredwhen the ion is coordinated to several oxygen atoms(weak field) rather than strongly bonded to one nega-tively charged oxygen atom (strong field). The activitycoefficients for the ions in the exchanged phases wereroughly equal to those found in p-alumina.

Compounds with the NASICON or triphosphate

framework structure constitute a very large class ofcompounds. Almost any four-valent cation can sub-sti tute for zirconium and partia l isomorphous dis-placement of M4+by W+, M2+, or M5+ s possible.'84J85In addition, compounds of the type A3M"',(P04), havebeen preparedls6 and s tudied in connection withstructure and conduction properties. The ion exchangebehavior of both groups of compounds has, however,been largely unexamined.

B. &Aluminas

1. (3-Alumina: Structure and Conductivity

pAlumina is the name applied "histo ridy" to a solid

solution series of general formula Na~+~All lOl,++. hename is actually a misnomer since initially the presenceof sodium ion in the compound was not recognized. Itsstructure consists of AlllOl, spinel blocks separated byplanes of sodium and oxygen atoms. These oxygen

bottlenecks are not small enough to prevent relativelyfree movement of the ions. Still, the room temperatureconductivity for LiTi2(POJ3 s ugJDc 7.9 X lodITL m-'. This is a case in point where ion exchange (ofsmall ions) is facile, but conductivity is poor. However,a t 300 "C, u = 5 X R-I cm-I . In contrast, Na*cannot be exchanged with H* n acid solution and theconductivity of the sodium phase is very low (uY6*c=

R-' cm-'). However, exchange does occur with

atoms bridge aluminum atoms in adjacent spinel hlocks.Th e space grou is hexagonal B 3 / m m cwith a = 5.68, and c = 22.5 1. The juxtaposition of Na' and planaroxygen atoms is shown in Figure 17, with the cationsites within the plane designated as Beevers-Ross (RR),antiReevers-Ross (ABR), and mid-oxygen uno) posi-tions. X-ray and neutron diffraction studies show thatthe occupancy of the several sites varies with the typeof ion, the magnitude of x, and temperature.lR- In the


15/24

Ion Exchange in Solid-State Chemistry

interest of brevity a simplified and generalized de-scription will be presented. In near-stoichioaetric @-alumina ( the stoichiometric composition is outside therange of homogeneity), the BR sites are filled.188J89 Thenext site to be oecupied is mO. Since the most commonvalues of x range from 0.15 to 0.3, roughly one-Sixth toone-third of the mid-oxygen sites are occupied. Thepresence of a cation near a n occupied BR site causesthe Na+ there to relax away from this site to an mO site.This so-called interstitial pair, at the mid-oxygen pos-itions, share a BR site, and this sharing lowers the po-tential barrier for ion motion from about 2 o 0.15 eV.lWIon motion is then thought to occur by an interstitialcymechanism in which one ion moves into a BR site whilethe adjacent ion in an mO site hops across to anothermO site near an occupied BR position. Thermal motionalso causes increased disorder, and at elevated tem-peratures a liquid-like smearing out of ion density is0 b ~ e r v e d . l ~ ~odium ion conduction within the planemay be as high as i2-l cm-l a t room temperaturein single crystals.

2. p"-Alum ina: Structure and Conductivity

In r -a lu mi na cations of lower charge such as Mg2+or Li+ replace some of the A13+in the spinel block. Thevalue of x is typically 0.67. T o accommodate this largeexcess of Na', the spinel blocks arrange themselvesdifferently. In P-alumina, the conduction planes aremirror planes with trigonal-prismatic coordination ofthe sodium ions in the BR sites. However, in r- a l u -mina, the spinel blocks are shifted relative to each othersuch that both the BR andABR sites are equivalentand tetrahedrally coordinated. The unit cell thus re-quires three spinel blocks with c N 33.75 A. Th e p'-arrangement of spinel blocks leaves more room in theconduction plane in the c direction so that ions can beslightly above or below the plane. Movement of ions

within the plane is thereby facilitated. Conductionoccurs by cation vacancy motion.Potential energy models predict that the activation

energy for a single sodum ion vacancy should be about0.02 eV,lgl but at room temperature the value observedis about 0.33 eV. This large energy value is attributedto ion-ion correlations. Evidence from diffuse X-ray~ c a t t e r i n g ' ~ ~ J ~hows an ordering of the vacancies witha superlattice built on a cell of dimensions 31/2a X 31/2a,as shown in dotted outline in Figure 17 . The orderedvacancies are trapped at these sites, requiring moreenergy to hop to a neighboring position than would berequired if there were no superlattice ordering. In-creased temperature causes a disordering of the su-

perlattice and a decrease in E , to near the calculatedvalue.

3. Ion Exchange in p- and p'-AluminaPerhaps the first reference to the ion exchange be-

havior of P-alumina is that of Toropov and Stukalova,lg3who prepared Ca2+-, Sr2+-, Ba2+-, and Rb+-exchangedphases from fused salts. Yao and Kummerlg4 carriedout an extensive examination of univalent and divalentexchange in Na-P-alumina. Exchange with univalentions was found to proceed rapidly a t about 300 "C infused-salt media. Complete isotherms for a number ofions were determined and these are shown in Figure 18.The shape of the isotherms depends upon the anion


N O N O , XNO,

LIOUID COMP

Figure 18. Equil ibr ia between P-alumina and var ious binaryni t ra te mel ts containing NaN 03 and another metal n i t ra te a t300-350 "C. From ref 194, with permission.)

used in the salt ba th as is shown by the following ar-guments:The ion exchange reaction can be represented by

RNa' + (MX), * RM+ + (NaX), (13)where R represents the P-alumina framework, and thebars indicate the solid exchanger phase. For this re-actionAGO =AGof(RM) - AGof(RNa) + AG"f(NaX)l - AGof(MX),

(14)

Since no anions enter the solid phase, only the last twofree energy terms change with change in the anion X.In some cases these changes are large and can makeeither Na+ or M+ the preferred ion. In chloride meltsK+ was preferred, but in iodide melts, Na+ was thepreferred ion.lS4 An extreme example is th at of Li',which is greatly preferred by the melt so that very littleis exchanged. However, if Ag+-p-alumina s substitutedfor the sodium phase in a chloride melt, total exchangeoccurs due to the low solubility of AgCl in the melt.

Self-diffusion coefficients (DJ or univalent ions weremeasured by using radiotracers. A t 350 "C the mag-nitudes of D i anged from cm2 s-l for Na' t o lo-'cm2 s-l for Rb+ with the order Na+ > Ag+ > Li+ N K+> Rb'. However, rates of ion exchange depend on in-terdiffusion Coefficients. For interdiffusion of alkalimetal ions in @-alumina t was found tha t, for K+ ex-change with Na-@-alumina, he interdiffusion coefficientincreases with K+ uptake, while if Na+ exchanges forK+ the opposite is true. These changes of interdiffusioncoefficients with composition are qualitatively in theright direction (see eq 33). However, exchange rates inP-alumina are also influenced by correlation effectslg5and the change in dimensions of the conduction planes.Both these factors require further examination.

Yao and Kummerlg4 also found that @-alumina wasa poor exchanger for divalent ions as they requiredtemperatures of 600 "C or greater to achieve partialexchange. In contrast, p-alumina was found to readilyexchange a variety of divalent ions.196 Fo r Sr2+ uptakethe diffusion coefficient determined from the ion ex-

- -


16/24

140 Chemical Reviews, 1988, Vol. 88 , No. 1 Clearfield

change rate was found to predict conductivities with afair degree of accuracy. Conductivities at 40 C are ofthe order of lo4 cm- while a t 300 C they are low20-1 cm-. Surprisingly, Pb2+ exhibited much higherconductivities approaching those of Na+-/3-alumina. Itwas also found that additions of small amounts of Ca2+to Na+-@-alumina owered the conductivity signifi-cantly, i.e., close to the conductivity of pure Ca2+-P-alumina.

Farrington and Dunn% attribute the differences inion exchange and conductivity behavior of divalent ionsin @-alumina nd @-alumina o the different conductionmechanisms. In &alumina the intertialcy mechanismrequires that both BR and mO sites be occupied tolower the barrier to diffusion. In divalent ion phasesthere are fewer cations than BR sites, so all the cationsare presumably trapped in deep potential wells at theBeevers-Ross sites. In /3-alumina this smaller numberof cations results in more vacancies. Since the BR andABR sites are equivalent in /3-alumina, approximately42% of the sites are filled and 58% are empty, allowingfor high conductivity by the vacancy mechanism. Morerecent structural worklW-lw has shown great variationin site occupancy. With Ba2+ all the ions are in BRsites, while with Pb2+, 90% of the ions are in the BRsites and 10% in mO sites, and with other ions (Sr2+,Cd2+, Mn2+),more than 50% of the ions are in mO sites.In addition, the bridging oxygen shows considerablevariation in its position between the spinel blocks, de-pending upon site occupancies of th e cations.

Several trivalent lanthanide @-aluminas have alsobeen preparedmlm1 as have mixed-valent forms.202 Thelanthanide phases all have a majority of the trivalentcations situated at mid-oxygen sites,202p203 ith onlyminor occupancy of the BR sites. Apparently, manyof the multivalent phases are characterized by orderingof the cations to produce superstructures.lS Eventhree-dimensional long-range order has been noted inCd2+-@-alumina nd lanthanide phases. These struc-tural s tudies indicate that a variety of different ionpositionings are possible for multivalent ions. Whilethese positionings have only a minor effect on con-ductivity, they exert a large influence on optical prop-erties.202 On the other hand, site occupancies of mix-ed-ion @-aluminas have a major effect on their con-d\!ctivity.

The extensive ion exchange chemistry of 0- nd j3-alumina allows a great variety of different cation formsto be prepared. This variety affords the opportunityto study the effect of ion environments on transport andother properties in a systematic fashion. The researcheffort needs to include ion exchange studies along withconductivity and structural information to correlatetheory with experimental data.204

C. Proton Conductors

In recent years there has been an intensification ofresearch aimed a t discovering new proton conductorsand the mechanisms of conduction in these compounds.This renewed activity has been driven by the potentialuse of such compounds in fuel cells, sensors, waterelectrolysis units, and other electrochemical devices.Research on proton conduction up to 1973 has beenreviewed by G l a ~ s e r . ~ ~ ~ ore recent collections of pa-pers may be found in the proceedings of international

TABLE . Values of Proton Conduc tivity in Some SolidElectrolytes

conductivity,materials R- cm- tem p, K

H ~ P M o ~ Z O U - ~ ~ H ~ O 1.8 X lo-* 298HSP W120 40-29H 20 1.7 X lo- 298

HUOzAs04.4HzO 6 X 310

3 X lo 4 2983 X loa 293

HUOzP04.4HzO 4x 10-3 290

HsUOz(IO&.4HzO 7 x 10-3 293

2 x 10-4 293ZrO2.2.3Hz0 3 x 10-5 293

5 x 10-3 298NH4+-@-alumina = 10-3 523

CGH1ZN12H2S04 =io- 298

H 30 C10Sbz08.4HzOSnOZ.3HzO

Hc-montmorillonite (1-4)X lo 4 290H-Al-mo ntmorillonite 104-10-2 290hydrated hydronium 8-alumina

NH4c/H(Hz0)x+-f -alumina =loa 298CeH,zNz*1.5HzS04 = l o 4 298

Adapted from: Huggins, R. A. In Materials fo r AdvancedBatteries, Proceedings of NATO Symposium on Materials fo rAdvanced Batteries, Aussois, France, 1979; Murphy, D. W., Ed.;Plen um Press: New York, 1980.

conferences. Huggins210 ummarized the situation

to 1979 and his list of proton conductors is given inTable 8. A major shortcoming of these compounds isthat they require water or ammonium ion species forfas t ion conduction. When this water or proton carrieris lost, the conductivity is decreased by several ordersof magnitude. This behavior points to a Grotthusmechanism, which was described in sections II.A.2 andII.B.3 in connection with proton conduction in hydrousoxides. A further brief discussion of proton conductionmechanisms is given below.

1. Proton Conduction Mechanisms

Proposed mechanisms for proton diffusion include,in addition to the Grotthus mechanism, complex iondiffusion (H30+, NH,) or vehicle mechanisms211 ndproton hopping or tunneling. The Grotthus mechanismrequires close proximity of water molecules which arefirmly held but able to rotate. Compounds with lesseramounts of water would be expected to conduct by thevehicle mechanism in which a nucleophilic group suchas H 2 0 or NH, acts as a carrier for the proton. Kreueret aL211 have pointed out that the activation energy forwater translation in hydrogen uranyl phosphate (HUP)and proton conduction in this compound are similar inmagnitude (0.8 and 0.64 eV, respectively) whereas theactivation energy for water rotation is 0.27 eV. Com-parison of activation energies led them to conclude thatthe diffusion mechanism is operative in HUP. Wewould then expect such a mechanism to be operativewith framework hydrates in which the acidity is highenough that the cavities are essentially occupied byH30+ or similar protonated molecules but little addi-tional water is present.

In both particle and framework hydrates, the loss ofall or most of the nucleophiles forces the proton to takepositions on the lattice oxygens. Proton transport thentakes place by hopping along lattice sites. This processrequires a constant breaking and making of bonds witha concomitant high activation energy. Proton con-duction is thus slow. Compounds with very highBransted acidity facilitate this mechanism. For acidsof the type H,MO,, the acid strength increases as one


17/24

Ion Exchange in Solld-State Chemistry Chemical Reviews, 1988, Vol. 88 , No. 1 14 1

TABLE 9. Specif ic Conductance of a -ZrP SamplesObtained by D ifferent M ethods of Prep arat ion andOrdered According to T heir Degree of Crysta l l in i ty

specificsample conductance,

no. prep method i cm-'1 precipitation at room temperature, 8.4 x 10-3

2 precipitation at room temperature, 3.5 x 10-3amorphousa

amorphous3 refluxing method (7:43), semicrystalline 6.6 X lo44 refluxing m ethod (10:100), crystalline 9..i X5 refluxing method (12:500), crystalline 3.7 x 10-56 slow precipitation from HF solution, 3.0 x 10-5

crystalline

a The specific conductance was measured by an isoconductancemethod at 25 O C . The values for samples 1 and 2 strongly dep endupon the conditions of preparation.212

proceeds up a group or to the right within a row of theperiodic table. Some examples are presented in thenext few sections.

2. Layered Phosphates

(a) Zirconium Phosphate. Almost all good protonconductors are good ion exchangers, which is in linewith thei r high Brransted acidity. Both zirconiumphosphate and HUP fit this description. In consideringthe proton conduction properties of a-zirconiumphosphate i t is necessary to recall that this compoundcan be prepared as a gel, as crystals, and in all inter-mediate stages between these extremes.' The specificconductance as a function of the degree of crystallinityis shown in Table 9. It is seen tha t the less crystallinethe sample, the higher the conductivity.212 Alberti e ta L 2 1 3 were able to show th at the surface conducts pro-tons lo00 imes faster than the interior. The activationenergy for surface conduction was 0.17 eV, whereas forconduction in the interior E , was found to be 0.65 eV.214In a-zirconium phosphate the water molecule is hy-drogen bonded by two of the phosphate protons andforms a hydrogen bond as donor to a third phosphateg r o ~ p . ~ ~ ~ ~ ~here is no evidence for the presence ofhydronium ions. Thus, conduction in the interior of thecrystal must occur by a modified vehicle mechanism asrotation of the water molecule is difficult. It is proposedthat the higher surface conductivity is due to the abilityof water located there to rotat e and participa te inGrotthus-type transport aided by water sorbed on the

As the crystallinity of the sample is de-creased, its surface area increases,215 accounting for theobserved increase in conductivity. In the case ofamorphous zirconium phosphate, swelling occurs in

water and hydronium ions may form,39 allowing for aGrotthus transport mechanism. Complete removal ofwater requires the proton t o hop, and accordingly theconductivity decreases to W 1 m-l at 200 "C,increasing to lo 4were able to delaminate zirconium phosphate by anintercalation-deintercalation eaction and prepare thinsheets from an aqueous suspension of the pellicles. Theconductivity parallel to the sheets was determined tobe approximately lo4 0-' cm-' at 300 "C and f2-lcm-' perpendicular to the sheets. This parallel con-ductivity is 2 orders of magnitude greater than ~ 3 0 0 0 Cfor a powder compact as reported by Jerus and Clear-field.216 Thus, by further manipulation of the pellicles

cm-l at 300 0C.216 Alberti et

it may be possible to increase the conductivity to nearthe maximum theoretical values obtainable for a pro-ton-hopping mechanism.

(b) Hydrogen Uranyl Phosphates. Hydrogen uranylphosphate, UOz(HPO4).4H20, as a layered structurein which uranyl ions are bridged by phosphate groups.The water molecules are arranged in squares with astaggered double layer of such squares in the interla-mellar space. HUP has been shown to exhibit excep-

17 OC. A brief summary of its structure, physicalproperties, and ion exchange behavior has been pub-lished recent1y.l The very high room tempera tureconductivity and intricate water structure make HUPan ideal subject for diffusional and conductivity studies.In addition, HUP can easily be fabricated into densedisks or films for use in electrochemical devices.218-220England et al.37 considered HUP to be a particle hy-drate with the water held loosely enough to favor aGrotthus mechanism. Evidence for water rotation wasobtained earlier by Childs e t a1.,221 who also suggesteda Grotthus-type mechanism. However, neutron dif-fraction222 nd NMR diffusivity indicateddifferent mechanisms for diffusion and conduction.These results led Kreuer et a1.21' to propose a vehiclemechanism for conduction. This question was furtherinvestigated by Tsai et by pulsed field gradientNMR. They concluded that diffusion and conductivitydo take place by the same mechanism but leave opento question the exact mechanism, whether Grotthus orvehicle.

A similar uncertainty exists concerning the mecha-nism of ion exchange behavior. The rate

4.ion exchange in solid state

Documents