Structure of synthetic monoclinic Na-rich birnessite and hexagonal

American Mineralogist, Volume 82, pages 946-961,'Ig97

Structure of synthetic monoclinic Na-rich birnessite and hexagonal birnessite:I. Results from X-ray diffraction and selected-area electron diffraction

Vrcron A. Dnrrsrt EwBN Srr.vnsrnnr2,x ANRToT-I I. GonsHKovr3 .a.No Ar,lJN MaNcneu2'Geological Institute of the Russian Academy of Sciences, 7 Pyzhevsky Street, 109017 Moscow, Russia2Environmental Geochemistry Group, LGIT-IRIGM, University of Grenoble and CNRS, 38041 Grenoble Cedex 9, France3lnstitute of Ore Mineralogy of the Russian Academy of Science, 35 Staromonetny Street, l}g0l7 , Moscow, Russia

Ansrnacr

Synthetic Na-rich birnessite (NaBD and its low pH form, hexagonal birnessite (HBi),were studied by X-ray and selected-area electron diffraction (XRD, SAED). SAED patternswere also obtained for synthetic Sr-exchanged birnessite (SrBi) microcrystals in which Srwas substituted for Na. XRD confirmed the one-layer monoclinic structure of NaBi andthe one-layer hexagonal structure of HBi with subcell parameters a : 5.1]-2 A, u : z.gqsA , c : 7 .34 A , F : 103 .3 ' and a :2 .848 L , , : 7 . t g A , y :120 . , r espec t i ve l y . I naddition to super-reflection networks, SAED patterns for NaBi and SrBi

"oniaitt satllite

reflections.On the basis of these experimental obervations, sffuctural models for NaBi and HBi are

proposed. NaBi consists of almost vacancy-free Mn octahedral layers. The departure fromthe hexagonal symmetry of layers is caused by the Jahn-Teller distortion associated withthe substitution of Mn3* for Mna*. The supercell A : 3a parameter arises from the ordereddistribution of Mn3*-rich rows parallel to t0101 and separated from each other along [100]by two Mna* rows. The superstructure in the b direction of NaBi type II (B : 3b) comesfrom the ordered distribution of Na cations in the interlayer space.-The maximum valueof the layer negative charge is equal to 0.333 v.u. per Mn atom and is obtained whenMn3t-rich rows are free of Mna*. The idealized structural formula proposed for NaBi typeII is Naor.r(M4irrM4lrrMq;*)Or. NaBi type I has a lower amount o1 Mn.* and its ideilcomposition would vary from Nao,ur(MnfltrrMqjur)O, to Naorr(M4jrMdjr)Or. Satellitesin SAED patterns of NaBi crystals result from the ordered distribution of Mnoi and Mn2*pairs in Mn3*-rich rows with a periodicity of 6b.

The structure of HBi consists of hexagonal octahedral layers containing predominantlyMn4* with variable amounts of Mn3* and layer vacancies. The distribution o1 layer vacan-cies is inherited from the former Mn3* distribution in NaBi. Interlayer Mn cations arelocated above or below vacant layer sites. The driving force of the NaBi to HBi ffansfor-mation is probably the destabilization of Mn3*-rich rows at low pH.

INrnooucrroN

Birnessite belongs to the phyllomanganate family. Thefust occurrence of this mineral was reported by Jones and

kov 1981; Cornell and Giovanoli 1988) and in marineand ocean manganese nodules and micronodules (Burnsand Burns 1976; Chukhrov and Gorshkov 1981; Chu-khrov et al. 1978,1985, 1989; Glover 1977: Drits et al.

* Present address: CSIRO Division of Minerals, Box 312,Clayton South, 3169, Australia

0003--004x/97l09 I 0-0946$05.00 946

1985). Birnessite is also a major component of someMn-rich ore deposits (Chukhrov et al. 1981,1989; Usuiand Mita 1995). Birnessite possesses unique surfacecharge (Healy et al. 1966; Murray 1974), cation exchange(Balistrieri and Murray 1982; Le Goff et al. 1996), andredox (Stone et al. 1994) properties, which make it highlyreactive with respect to sorption phenomena (Tir et al.1994; Paterson et al. 1994). Furthermore. birnessite issynthesized easily under laboratory conditions (Giovanoliet al. 1970a, l97obl. Bricker 1965; Strobel et al. lggT:Cornell and Giovanoli 1988) and, consequently, often hasbeen used as a model manganese oxide in environmentalchemical studies (Stone and Morgan 1984; Stone 1987;Xyla et al. 1992; Manceau and Charlet 1992; Bidoglio etal. 1993).

Despite the fact that the main structural and chemicalproperties of natural (Chukhrov et al. 1985, 1989; Man-

DRITS ET AL.: BIRNESSITE STRUCTURE 947

ceau et al. 1992) and synthetic (Post and Veblen 1990;Kuma et al. 1994) birnessite has been determined, manydetails concerning its structure and crystal chemistry re-main poorly understood, specifically diffraction features,crystal-chemical formula, and physico-chemical proper-ties for the different individual birnessite species, and thestructural mechanism of their formation and transforma-tion. The solution to the first two problems predeterminesthe success in the solution of the others.

The present paper and the companion paper (Silvesteret al. 199'7) deal with the structure of synthetic Na-richbirnessite and its modifications that occw under acidicalteration. This paper is concerned with structural infor-mation obtained from X-ray diffraction (XRD) and se-lected-area electron diffraction (SAED). The second pa-per presents the results of extended X-ray absorption finestructure (EXAFS) analysis and chemical studies. Allthese new results will be discussed in the light of datapreviously reported in the literature to obtain a more com-prehensive understanding of birnessite structure.

PnBvrous sTRUCTURAL sruDIES

One of the main difficulties in the structural study ofbirnessite is that it exhibits several natural and syntheticvarieties. Furthermore, it occurs in an exffemely dispersedstate and generally has a low degree of structural perfec-tion. For these reasons, until recently, even the unit-cellparameters of the various described birnessite modifica-tions were not determined unambiguously. Giovanoli etal. (1970a) and Burns and Burns (1977) proposed that thebirnessite structure is analogous to that of chalcophanite(ZnMn,0,.3H,O; Wadsley 1955; Post and Appleman1988). In each chalcophanite layer one out of seven Mna*octahedral sites is vacant. Vacant sites form a superstruc--ture with the layer unit-cell parameters: a = b : 7.54 i\and 7: 120". Interlayer Zn cations are located above andbelow vacant Mn sites. Based on a close anion packingnotation this structure can be expressed asAbCb'A'c'BcAc'B'a'CaBa'C'b'AbC, where A, B, and Crepresent sites for O atoms; A', B', and C' sites for HrOmolecules; a, b, and c layer Mn sites; and a', b', and c'interlayer sites.

Burns and Burns (1977) speculated that interlayer al-kali and alkaline earth cations occupy the same positionsin the birnessite structure as ZrP* cations in chalcopha-nite. Giovanoli and Arrhenius (1988) proposed that in bir-nessite layers one out of six Mn sites is vacant and thatMn2* and Mn3* cations occupy interlayer sites above andbelow layer vacancies. A vacancy model for phylloman-ganates was also developed by Strobel et al. (1987). Allthese models are based on intuitive considerations de-duced from the determination of unit-cell parameterswithout any quantitative analysis of the diffractionpattefns.

Hexagonal one-layer birnessite

The first detailed structural study of natural birnessitewas performed by Chukhrov et al. (1985) using XRD and

SAED. This pure birnessite sample came from manga-nese micronodules dredged from the Pacific Ocean floor.This sample had a hexagonal unit cell with a = 2.838

and c : 7.10 A. Using the same symbolic notation as forchalcophanite, the layer sequence for the model structuresuggested by Chukhrov et al. (1985) is:

AbCb," b,AbC

This interpretation was supported by X-ray absorptionspectroscopy (Manceau et al. 1992). The structural anal-ogy with the chalcophanite structure was then establishedon a quantitative basis, both minerals having similar layerand interlayer structures, but with birnessite exhibiting a

one-layer hexagonal cell and a random distribution of va-

cancies within the layer. Those XRD patterns of syntheticand natural birnessites with a hexagonal unit cell de-scribed by Giovanoli et al. (1976) and Glover (197'7) Nesimilar to that of Chukhrov et al. (1985).

Birnessite with monoclinic subcell

One occurrence of natural one-layer monoclinic bir-nessite has been reported by Gorshkov et al. (1992). Thestructure of synthetic monoclinic birnessite samples con-taining exchangeable Na*, Mg2*, and K* cations in theirinterlayer was determined for the first time by Post andVeblen (1990) using the Rietveld technique. The resultsobtained by these authors confirmed the similarity of syn-thetic birnessite and chalcophanite structures. These bir-nessite samples have a monoclinic subcell, the shape andsize of which varies with the nature of the interlayer cat-ions. The idealized one-layer monoclinic structure of

Na-rich birnessite (NaBi) may be transformed into a

three-layer rhombohedral structure, the layer sequence ofwhich may be compared with that of chalcophanite (Man-

ceau et al.1992):

AbC CaB BcA AbC NaBi (monoclinic)

AbC BcA CaB AbC chalcoPhanite

For the sake of simplicity only the stack of Mn octa-hedral layers is represented here. The minerals differ bytheir layer stacking and, consequently, do not possess the

same interlayer structure. In agreement with the pure lay-ered structure found by Post and Veblen (1990)' noMn-O-Mn corner linkages were detected by EXAFS inthe synthetic monoclinic NaBi sample (Manceau et al.1992).

Origin of birnessite superstructures

Giovanoli et al. (l97}a) were the first to observe the

existence of superreflections in the SAED pattern of syn-thetic NaBi. They assumed that these superstructureswere caused by an ordering of Mn vacancies as for chal-cophanite. Similar conclusions were drawn by Chukhrovet al. (1978) from the SAED and energy-dispersive anal-ysis (EDA) study of Ca-rich oceanic micronodules.

From the analysis of SAED patterns, Post and Veblen(1990) concluded that it was most likely that superreflec-

948 DRITS ET AL.: BIRNESSITE STRUCTURE

tions resulted from a regular distribution of interlayer cat-ions rather than of vacancies. This was supported by thefact that Na-rich and Mg-exchanged birnessite had dif-ferent superstructures but the same layer charge. Manceauet al. (1992) proposed that the two dimensional super-periodicity observed in the SAED patterns for NaBi par-ticles resulted from a periodic distribution of Na* atomsbecause of their electrostatic interactions. Accordingly,the exchange of Na* by Mgr* should lead to a new super-periodicity because of the decrease of the number of in-terlayer cations. This effect was actually observed by postand Veblen (1990). However, according to these authors,Mg2+ atoms in Mg-exchanged birnessite, hereafter re-ferred to as MgBi, are located almost above or belowoctahedral sites. Such an arrangement of interlayer Mg2+cations is possible only if the underlying sites are vacant(Manceau et al. 1992). This observation must be consid-ered as a strong argument in favor of layer vacancies inthe structure of monoclinic birnessite. Thus, there is acontradiction in the data presented by post and Veblen(1990). On the one hand, SAED data and the constancyof the interlayer charge for Na and Mg birnessite supportthe idea that the superstructure results from the ordereddistribution of interlayer species. On the other hand, thestructural refinement of MgBi indicates the presence oflayer vacancies.

Kuma et al. (1994) studied a large collection of syn-thetic birnessite and buserite samples exchanged bymonovalent (Na, K, Li, and Cs) and divalent (Ca, Mg,Sr, Be, and Ba) cations. Each species was studied by pow-der XRD and SAED. They found, in agreement with postand Veblen (1990), that a, c, and B subcell parametersdepend on the type of interlayer cation. They assumedthat layer vacancies are associated with interlayer cations,which were found to be linearly arranged along the b axis.It was concluded that these vacancy-rich rows were sep-arated by two complete rows of Mn octahedra. They em-phasized that, in this respect, the structures of 7 A-bir-nessite and l0 A-buserite are notably different fromchalcophanite.

The vacancy problem is closely related to the problemof the valence of Mn cations and to the determination ofa correct crystal chemical formula for each birnessitemodification. For example, the same chemical analysis ofa birnessite sample can be interpreted either asNaouMnflMn;:O4 or as NaouMnl;rno,rOo, where nstands for a lattice vacancy. In general, there might bedifferent proportions of Mna*, Mn3*, and Mn2* cations,and there exists the possibility of replacing some O atomsby OH groups as well as some layer vacancies. For thesereasons, no reliable technique exists to calculate the crys-tal-chemical formula of birnessite. The determination ofthe origin of superstructures observed in different birnes-site species appears to be a prerequisite for determiningtheir structural chemistry.

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0Degrees 2e (CuKo1)

Frcunr 1. (a) X-ray diffraction pattern of natural Ca-richbirnessite with a turbostratic structure. (b) X-ray diffraction pat-tern of HBi (pH 3) showing the diagnostic reflections 102 and103. (c) X-ray diffraction pattern of NaBi (pH 9) showing thediagnostic reflection pairs 200,110 and 310,020 that result fromthe departure of the hexagonal symmetry of the Mn layers.

Characteristic diffraction features of birnessiteNatural and synthetic birnessite samples may have tur-

bostratic, one-layer hexagonal, and one-layer monoclinicstructures.

The XRD patterns for birnessite with a turbostraticstructure display 00/ reflections and only two two-dimen-sional diffraction bands havins lfi) and 110 indices withd,- - 2.42 A and d,,o - l.4i A, respectively (Fig. la).XRD patterns for one-layer hexagonal birnessite contain,in addition to 00/ and hkO reflections, diasnostic lO2 and103 peaks with d,o,: 2.03 A and d,o. : t\t A G'ig. tU).These reflections should have similar intensities if inter-layer cations are located above or below the layer octa-hedral sites (Chukhrov et al. 1985).

The XRD patterns for one-layer birnessite with mono-clinic subcells show 200, 110, and 310, 020 reflectionpairs with d values equal to 2.52 -+

_0.0I, 2.49 + 0.01,1.445 4- 0.005, and 1.423 -r 0.005 A, respectively (Fig.

N

o

o


1c). The presence of reflection pairs is induced by astrong distortion of the hexagonal symmetry of layersleading to an alb ratio different from \4. As a result ofthis distortion drro ) doro, whereas for an hexagonal sub-cell with the same b, c, and B valugs as in the monoclinicsubcell, the condition of b : al\/3 leads to do.o > drro.

ExpBnruonrAl METHoD

To determine the origin of superstructures, birnessitesamples saturated with Na and Sr were studied. Na-richbuserite (NaBu) was prepared following the procedure ofGiovanoli et al. (1970a). The solid phase was washed bycentrifugation until the supernatant pH was approximate-ly 9-10 (more than six times). The solid phase was ini-tially characterized by XRD in a wet paste form, yieldingthe typical 10 A layer spacing indicative of the NaBuphase. NaBi was obtained directly from the NaBu sus-pension by filtering and drying the solid at 40 "C. Thelow pH forms of birnessite were prepared at constant pHover a pH range of 2-5 from NaBu and then filtered anddried. Giovanoli et al. (1970b) and Silvester et al. (1997)show that the low pH forms of this mineral are Na* freeand in the following discussion will be referred to by thegeneral name HBi.

Sr-exchanged birnessite samples, hereafter referred toas SrBi, were prepared by shaking a NaBu suspension atl0 g/dm3 in 1 moVdm' Sr(NO.), solution for 24 h. Theexchanged solid was then filtered and dried as for NaBi.

Powder XRD patterns were obtained using CuKa ra-diation with a Siemens D5000 powder diffractometerequipped with a Si(Li) solid-state detector. Intensitieswere measured at an interval of 0.05' 20 and a countingtime of 3 s per step. SAED pattems were recorded usinga JEM 100c microscope equipped with a Kevex spec-trometer and operated at 100 kV. Samples from diluteNaBu and Sr-exchanged buserite suspensions were sus-pended onto carbon grids, without previous drying orheating of the sample. The samples were subsequentlymounted on a tilting sample holder. SAED patterns wereinterpreted as described by Drits (1987).

ExpnmlrnurAl RESULTS

X-ray diffraction

The XRD pattern of our NaBi sample (Fig. lc) is al-most identical to that reported by Post and Veblen (1990).The measured d values correspond to a one-layer mono-clinic subcell having parameters a : 5.112, b : 2.849, c: 1 .34 A, and F : 103.3" . XRD patterns of HBi samples(Fig. lb) were all indexed with a one--layer hexagonal unitcell with a : b : 2.848, c : 7.I9 A, and y : 120'. The101 reflections are broadened because of the presence ofstacking faults in microcrystals.

Selected-area election diffraction

NaBi. Figures 2a and 2b show the two main types ofSAED patterns observed for NaBi rnicrocrystals (birnes-site type I and II, respectively). Both patterns contain aset of strong hko reflections distributed according to a

Frcunr 2. Selected-area electron diffraction patterns forNaBi (a and b) and HBi (c) microcrystals. Upper right insert ofSAED pattern (a) shows the triplet of satellites.

AY3

... fl? : : o ' . 1 a v o

o . . o . + A d

o . . o . . o

. 4 . . o .

o . . a . . o

. 4 . . o . .

o . . a . . a


Frcunn 3. Idealized distribution of diffraction maxima ob-served in the SAED pattern of Figure 2a. Large and mediumsolid circles correspond to subcell and supercell reflectrons, re-spectively. Small black points correspond to satellites. A* andBx are the supercell parameters.

pseudo-hexagonal symmetry and corresponding to thelayer subcell determined by XRD. According to theSAED observations, the proportion of birnessite type tris much higher.

NaBi type I. The distribution of reflections in theSAED pattern of NaBi type I (Fig. 3) may be describedby a ce"ntered layer supercell with parameters A : 3a =15.52 A, B : b : 2.85 A, and y = 90". Each of thesubcell and supercell reflections is surrounded by six sat-ellites with one pair along [01]* at + b*/6, and two otherpairs located along [11]* and [11]x, respectively. The dis-tance between a central hlfr refl,ection and any one ofthese four satellites is

[(A*)' + (Bx)z1uz16 : ll6d,,o.

Thus, only a cell with A{ = 4*16and B$ = 8*16 mayaccount for the positions of all diffraction maxima, in-cluding the satellites; this cell will be referred to as theS cell. The intensity of satellites depends on the value ofk (see Fig. 2a). In particular the satellites around /200reflections have equal intensities. Satellites around reflec-tions with k + 0have different intensities: Inner satelliteswith k - Yu ne more intense than outer satellites with k* '/u. The difference between the intensities of inner andouter satellites increases with increasing /c values (Fig.2a).

NaBi type II. Another type of SAED pattern observedfor NaBi microcrystals is shown schematically in Figure4 (cf. Fig. 2b). Two sorts of satellites may be distin-guished. The distribution of superreflections correspondsto a base-cente-red supercell with A : 3a : 15.52 A, B: 3b : 8.55 A, and 7 : 90'. The satellites of the firsttype (open circles) are near some ftk reflections at B*12= -r b*16 along [01]*. These satellites have the coordi-

Frcunr 4. Idealized distribution of diffraction maxima ob-served in the SAED pattern of Figure 2b. Large and mediumsolid circles correspond to subcell and supercell reflections, re-spectively. Open circles and small black ellipsoids elongatedalong [11]* and [11]* are satellites. A* and B* are the supercellparameters. In all patterns, the A* axis is oriented horizontally.

nates h,k + 1/2. As above, the satellite intensity dependson the value of ft.

The distribution of the second type of satellite resem-bles an incomplete six-pointed star (Figs. 2b and 4). Thesatellites are elongated along different directions, whichare rotated by 60'. The distance between these satellitesand their nearest subcell or supercell reflection is1/zl[(A*),+ (Bx;zluz Satellites elongated along [11]x are placed inthe middle of two nearest main reflections located along[11]*. There are no satellites along the [10]* direction.Thus an S cell with parameters A! : A*12 = a*16 andB! : B*12 : b*16 is necessary to describe the positionsof all diffraction maxima, including satellites.

HBi. Figure 2c shows an SAED pattern typical forHBi. The reflections are distributed according to the ex-pected hexagonal symmetry with layer hexagonal cell pa-rameters of a : 2.84 A or^base-centered cell parametersof a -- 4.927, b :2.84 A. and 7 = 90'. Such SAEDpatterns have been observed previously for natural andsynthetic one-layer hexagonal birnessite (Giovanoli et al.l970b; Chukhrov etaL.1978,1989). However, the SAEDpatterns shown in Figure 2c contain extremely weak butnoticeable additional reflections distributed according tothe base-centered supercell with A : 3a : 14.78 A, B :2.845 A, and 7 : Qgo.

SrBi type I. Three different SAED patterns were iden-tified for SrBi microcrystals (Fig. 5). Figure 6 reproducesschematically the position and shape of diffraction max-ima observed in SAED patterns of the first type (Fig. 5a).A supercell can be identified with the parameters Ax :

a

Ft +

n0ro

t i t r t ' t , ' . t o o

l l i r

\ . . . r / , ' \ . . . .

/I e ' *2r ' ' . '9 , i - : . o

a a a a o a a a a

tB'p.lr l

o..o. .o

\ /. . . . F i t r . .

o . . a' . . JslJ:' . 'a . .o, 'l-J

\oa o a a A a o a

O o . O . . O

DRITS ET AL: BIRNESSITE STRUCTURE

Frcunn 5. Main types of SAED pattems observed for SrBi microcrystals. The various microcrystals differ from each other bytheir super-cells. (a) SrBi type I, (b) SrBi type II (c) and (d) SrBi type III.

951

a*/3, B* : b*, and y* : 90'. One can note that each hkreflection is surrounded by a pair of diffraction maximastrongly elongated in the [31]* direction. Each hk reflec-tion is also suffounded in the direction parallel to I I I . 1] *,

almost perpendicular to [31]x, by a pair of these elon-gated maxima. The positions of the main spots and elon-gated maxima can be described by a periodic two-dimen-sional network. The cell of this network has theparameters A! : y,(IlA*+B*) and B$ : th(-3A*+B*)

oriented along the supercell directions [11.1]* and [31]*,respectively.

One can note (Fig. 6) the regular distribution of thediffuse (elongated) maxima relative to the main reflec-tions: One pair of these maxima are placed along theB$ axis at distances of +2B$ and the other pair along theAf axis at distances of *2Af . Such mutual arrangementof main and diffuse reflections leads to specific values ofthe /e, and ,t. indices. In particular h" + k" : 7n (n : 0,1 ,2 , . . . ) f o r t he ma in re f l ec t i ons and h ,+ k " :2 ,5 ,9 ,13. . . for the diffuse maxima. The nodes correspondingto h. + k : 1, 3, 4, 6,8. . . . have zero intensity.

SrBi type II and III. The second type of SAED pat-

952 DRITS ET AL: BIRNESSITE STRUCTURE

Frcunn 6. Idealized distribution of diffraction maxima ob-served in the SAED pattern of Figure 5a. Large and mediumsolid circles correspond to subcell and supercell reflections, re-spectively. Black ellipsoids elongated along [31]* are satellites.A*, B* and Af, B$ are the supercell and S* cell parameters,respectively.

terns (Fig. 5b) corresponding to SrBi microcrystals con-tains a reflection network that can be described by a rect-angular primitive cell with A!, : a*/3 and Bf : b*/4.Note that among ft00 reflections, those with ft = 2n + Iare absent. The SAED patterns shown in Figures 5c andJ are a superposition of three reflection networks, corre-sponding to at least two separate structural varieties. Thefirst reflection network (small black squares) is rectan-gular and is identical to that shown in Figure 5b. Thesecond diffraction pattern, which can be extracted fromFigure 7, is illustrated schematically in Figure 8. The po-sitions of these points can be described by a two-dimen-sional network with the cell parameters Af : l"(l3A* +B*) and Bt : %e3A* + B*). Here A* and B* are theconventional super-cell parameters. Relative to this su-percell A$ and B$ are oriented along [13.1]x and [3t]xdirections, respectively (Fig. 8). As in the previous caseeach reflection of the subcell and supercell is surroundedby four diffuse maxima elongated in the 13t1x direction.Two are located along the [13.1]* direction at -r2A$ fromany Leflection, whereas the other two are located alongthe [31]* direction at +28$ from the same reflection.

Because the h" + k values for all diffraction maximaare even, the corresponding S cell in real space is char-acteized by a base-centered cell. However, the values ofh,+ k" depend on the type of reflection (Fig. 8).

The remaining reflections of Figure 7 (third reflectionnetwork) may be generated by a 180' rotation of the sec-ond network around the [10]* axis. Thus the schematicSAED pattern shown in Figure 7 is the superposition of

Frcunn 7. Idealized distribution of diffraction maxima ob-served in the SAED pattern of Figure 5c. Large and mediumsolid circles correspond to subcell and supercell reflections, re-spectively. The small squares correspond to reflections observedin the SAED pattem shown in Figure 5b. The other reflectionsare connected by a mirror plane and correspond to a secondstructural component, which is twinned. S* cells are shown foreach twin.

diffraction patterns from crystallites having a rectangularcell and two other crystallites having twinned S cells. Themain difference between the SAED patterns for SrBi mi-crocrystals in Figures 5c and 5d (SrBi type III) is that thediffuse streaks of Figure 5c are transformed into normalsharp and intense reflections in Figure 5d.

INrnnpnnr.q.rroN

Structure of the interlayer

NaBi. We assume that the superreflections in theSAED patterns originate from an ordered distribution ofinterlayer cations and HrO molecules. This arrangementmust reflect the total layer charge and ordering of nega-tive charges within layers. The validity of these assump-tions is discussed below. In NaBi. surface O atoms and

Frcunn 8. Distribution of diffraction maxima correspondingto one of the twinned structures observed in Fisure 7

B:b


3b

l o(Y1

ll

t<

t o(f,

tlt<

abFIcunr 9. Supercells for NaBi types I and ll wrth A : 3a

and B - b (a), and A : 3a and B : 3b (b). In the first cell,solid circles and open squares correspond to interlayer sites withdifferent Na occupancies. In the second cell, solid and open cir-cles correspond to Na and HrO positions, respectively.

sites occupied by interlayer Na and HrO molecules arevery close in projection inthe a-b plane (Post and Veblen1990).

NaBi type I. The interlayer supercell with A : 3a :

15.52 Aand B = b : 2.85 A is represented in Figure 9a.Among the six available sites for Na and HrO within theinterlayer supercell the two sites with coordinates 0,0 and7/2,/z mast have a higher scattering power because the cellis base-centered. According to the chemical analysis (Sil-vester et al. 1991) each octahedral site has 0.30 Na atomsassociated with it. This means that the supercell interlayercontains 1.8 Na atoms. If Na atoms were at supercellnodes 0,0 and /2,1/z with 0.9 probability, then Na-Na pairswould be aligned along the b axis, which is not acceptablebecause Na must alternate with HrO molecules. As a con-sequence, it is necessary to assume that all six supercellsites are occupied by either Na or HrO, but the probabilityof Na occupancy at 0,0 and 1/2,/z must be higher.

NaBi type II. Another distribution of Na and H,O mol-ecules, which avoids neighboring Na cations, is presentin NaBi type II. The supercell (A : 3a = 15.51 A, B :

3b : 8.55 A) contains 18 sites available for Na and H,O(Fig. 9b). It is easy to see that the 5.4 (18 x 0.3) atomsof Na can now be readily distributed in such a way thateach Na is surrounded by six HrO molecules. Thus thissupercell contains 6 sites for Na and 12 sites for HrOmolecules. In the case of a homogeneous distribution ofNa atoms, each of the six Na sites should be occupied

with a probability of 0.9 (0.9 x 6 : 5.4). However, be-cause of the base-centered distribution of interlayer Na,it seems reasonable to assume that the 0,0 and L/2,r/2 po-sitions have a higher occupancy probability. If Na atomsoccupy all available sites with 1007o probability, the max-imum possible content of Na atoms per layer octahedronis 0.33.

HBi. According to previous XRD data (Chukhrov etal. 1985) birnessite layers contain many vacancies. Thevery weak superreflections observed in the SAED pat-terns can be interpreted as resulting from a random oralmost random distribution of layer vacancies. But thisinterpretation may be misleading. Indeed, if interlayer Mncations are placed just above or below vacancies, then inprojection inthe a-b plane vacant sites will have the samescattering power as filled layer sites. In this case the dis-tribution of hl<O reflections would not be sensitive to thepossible order-disorder of vacancy distribution. If inter-layer Mn cations are placed above and below vacanciesthe scattering power of vacant sites would be enhancedand the absence of hl{ superreflections would indicate arandom distribution of vacancies. Silvester et al. (1991)

show that interlayer Mn cations are located either aboveor below the layer vacancies, which is consistent with theweakness of superreflections for HBi. These superreflec-tions arise from the regular distribution of interlayerMn2+/3+ and the associated interlayer HrO molecules, eachcation being octahedrally coordinated to three layer Oatoms and to three interlayer HrO molecules. The super-reflections in the SAED patterns reflect the tendency to-ward an ordered distribution of vacancies and the asso-ciated interlayer Mn2+/3+ and HrO molecules, over thenodes ofthe base-centered cell withA :3a:3t/3b andB : b : 2 . 8 4 5 4 .

SrBi type I. The supercell observed in the SAED partern of Figure 5a with A : 15.59 and B : 2.83 A isanalogous to that determined for NaBi type I. For charge-balance reasons we assume that 0.15 Sr atom per octa-hedral site exists. The interlayer super-cell must then con-tain 0.9 cations. Because of the high scattering power ofSr, the base-centered cell, and the low Sr interlayer con-tent, there is no doubt that Sr occupies 0,0 and /2,/z sitesrather than the other four supercell positions. As forNaBi, Sr atoms cannot occupy two successive sites along

t011.SrBi type II and III. Rectangular networks of reflec-

tions observed for SrBi can be described by a supercellwithA = 3a : 15.ff iA andB : 4b : 1132 AGrg. 10).There are 24 avallable sites and the cell contains 3.6 Sratoms. The cell would therefore contain four sites with907o probabllity of occupancy. A possible model for thedistribution of interlayer Sr is represented in Figure 10.Note that the periodicity of the projection of Sr and H,Oalong the A axis is A/2. For this reason ft00 reflectionshave h : 2n (Fig.5b and 5c). Unit cells for the structuralcomponents responsible for the other reflections observedin Figures 5c and 5d are considered in the next section.

$ -


III

dIIIIIIII

a

>ltlqq)+

-+ *B=4b-

FIcuns 10. A model for the interlayer distribution of Sr(large solid circles) and HrO molecules (small open circles) inSrBi microcrystals with supercell parameters A : 3a and B :4b.

The origin of satellites for Na-rich birnessiteWilson (1964) showed that if each hkl node of the re-

ciprocal lattice is surrounded by two satellites at +,\ fromthe node along the b* axis and the outer pair is weakerthan the inner pair, then the crystal contains a modulationalong the b axis with a periodicity ,\. In addition, thestructure amplitude will be higher for higher do,o values.

NaBi type I. The theoretical considerations developedby Wilson (1964) are used to interpret qualitatively someSAED features observed in Figure 2a. In Figure 11, Naand HrO are located in such a way that do,o values changeperiodically fro^m d, : do,o * 6 to d. : doro - 6 with I: 6b : 11 .I A. In the model it is considered that therelative scattering power of atoms located in planes par-allel to (010) decreases with decreasing interplanar dis-tances from 1.0 to 0.7 with a 0.05 step. Using the atomicpositions of Figure 11, structure factors of superreflec-tions and satellites were calculated by assuming that therelative positions of these planes, or the y/8" coordinatesfor Na and HrO molecules along the B" axis for half theperiod, are equal to, 0.0, 0.087, O.ll2, O.ZS7,0.34, O.4ZI,and 0 .50 (A " :3a : 15 .51 A ; B " : 6b : I 7 . t 41 . ac -cordingly, d values between two nearest planes vary be-

l .mT

l <

FIcunn L1. Idealized structural model for NaBi interlayersof the first type. Regular white and dotted triangles correspondto the O atom layer. The solid circles represent sites enrichedwith Na. The vertical thick lines with variable distance corre-spond to (010) planes, which contain Na sites. Oblique lines withvariable distances correspond to (110) planes.

tween 1.488 and 1.351 A, whereas in the strictly periodicstructure this interplanar distance should be equal to1.425 A (bl2). As noted in the previous section, the xcoordinates for Na and HrO correspond to their ideal val-ues. Using this model significant diffraction intensitieswere obtained only for nodes corresponding to the mainreflections and for satellites located at distances of -+b*16.

Additionally, this model led to greater intensity for innersatellites relative to outer satellites, as observed experi-mentally (Figs. 2a and 3). This model also accounts forthe presence of satellite pairs along [11]* and tlTlx *-rections near each ft,treflection at distances ofr*/6, whererx : [(A* ), + 1B*1,)',' (Figs. 2a and 3). As seen in Figure11 the (110) and (110) planes also contain Na and H,Omolecules and their interplanar distances are also modu-lated along their normal. This periodic variation resultsfrom the fact that the displacement of Na and HrO alongthe b axis can be decomposed into two components par-allel and perpendicular to (110, 110) planes. As can beseen in Figure 1l the wavelength of the variation of thed,,o.,, values is equal to tr,,o.,,. 6drro.r*.

NaBi type II. The same interpretation applies to sat-ellites of Figures 2b and 4. The directions of displace-ments for Na and H,O within the S cell (A" : A = 3a,B" : I : 28 : 6b) are shown in Figure 12. Calculationswith the relative scattering power of planes decreasing(from 1.0 to 0.7) with decreasing d values explains qual-itatively both the origin of satellites at distances of k -+1/z from /rk reflections and the different intensities of innerand outer satellites.

The second type of satellites are a direct consequenceof the specific distribution of Na and H,O within the su-

lH2Ol


a

O o ' € or O €

o' O,

O o o

o' a,

. o + o O

. O o

O o < o O

o

€ O o '

.O .o o' O'

+ o r o a o 6

O l o € ( a

@ O €

O ' o € a O

o O ' o

O ' o r c . O

o o f . o €

O ' @ € { O

Frcunn 12. Idealized structural model for NaBi interlayersof the second type. Large and small circles represent Na siteswith different degrees of occupancy Small open circles corre-spond to HrO sites. Arrows attached to circles indicate the pe-riodic displacement of interlayer species along the B axis. In theupper left portion of the diagram two supercells with A : 3a andB : 3b are shown. Unit cells I and II in the form of a rhombhave the same sizes and shapes but differ slightly in the arrange-ment of Na and HrO sites. I and ,\' are wavelengths of variationof interplanar distances for (010) and (1 10), respectively

percell with A = 3a and B : 3b and of their periodicdisplacements along the b axis. As shown in Figure 12,planes parallel to (110,1T0) pass through Na and H,Osites. The distance between these planes varies periodi-cally along the [13] and [13] directions because of theperiodic displacement of Na and HrO along the b axis.The periodicity of wavelengths (I') is equal to 2drro,1o,and in the reciprocal lattice each reflection is surroundedby a pair of satellites located along [11]* and [11]* withrelative coordinates h -r t/2 and k + Vz. Due to the syn-chronized variation of the interplanar distances and oftheir scattering powers, an asyrnmetrical distribution ofintensity for inner and outer satellites is obtained and isobserved experimentally (Fig. 2b).

Frcunn 13. A model of the mutual arrangement of Sr (largesolid circles) and HrO molecules (small solid circles) in the in-terlayers of SrBi microcrystals that are characterized by theSAED pattern of Figure 5a The probability of Sr occupancy isabout 0.4-0.45 Arrows show the directions of Sr displacementsfrom their ideal positions. A,B and A",8" are the parameters forthe supercell and S cell, respectively. ,\, and ,\, are the wave-lengths describing the periodicity of the d',0.

As noted previously, some of these satellites are elon-gated along the [1l]x or [1T]* directions (Fie. a). To un-derstand the origin of this feature two adjacent rhomb-shaped unit cells (I and II) were chosen to have sidesparallel to [11] and [11] and edge lengths of (Az + Bzltz(Fig. 12). These cells (I and II) have the same size butdiffer slightly in their atomic positions. Because of thispeculiar atomic zurangement, planes parallel to (110) andlocated in two neighboring type I and II unit cells do notscatter strictly in phase because Na and HrO located inthe same plane are shifted in opposite directions (FiS. 12).The perturbation of the wave phases leads to the elon-gation of satellites in the [1l]x direction. The same effectoccurs for planes parallel to (110) leading to the forma-tion of satellites elongated along [11]*. Note that the ab-sence of satellites along the [10]x direction is explainedby the fact that the (h00) planes are always parallel andhave a uniform spacing.

The origin of satellites for Sr-exchanged birnessite

SrBi type I. Figure 13 shows the relationship be-tween the supercell with A : 3a, B = b, and 7 : 90'and the S cell with A, : lz(A + 38) and B' : lz(llB- A). As can be seen in Figure 13 the S-cell containsseven sites occupied by Sr with a probability equal to0.45-0.40 so that each Sr is surrounded by six nearestHrO molecules. As for NaBi microcrystals, satellites inthe SAED patterns of SrBi are associated with the pe-riodic displacement of Sr cations around their ideal po-sitions. The simplest model that would explain the ob-served effects is one in which Sr cations are shiftedperiodically along the normal to (310) planes. Such adisplacement leads to a periodic variation of d'0. As

- d

a l

II

-. -r-ie; i.,

a lt \ l

>Jt loD I

t \l \t \II) 'y-J| .,/

iIIt .

)a l' J ll , ll " lt r ' l (.r

" l-' a , - 4 .

1 \ - ' l ' .t ' v - (

l \ .r lIt

I r '

- r a .

{-l q

f -

l a

-+

I

II

a 'I

a

J, l

^ '\ lI

Ia /

t 'II

II

I


lT.r sj-r -1t lt o lt l

a

. la t a

t l

Frcunr 14. A model of the mutual arrangement of Sr (largesolid circles) and H,O molecules (small solid circle) in the in-terlayers of SrBi microcrystals that are characterized by theSAED pattern shown in Figure 5c. The probability of Sr occu-pancy is about 0.4-0.45. The arrows show the directions of Srcation displacement in conespondence with the periodic varia-tion of d.,0. A,B and A",8" are the parameters of the supercell andS cell, respectively.

seen in Figure 13 the S cell contains seven 1310) planespassing through Sr cations. If the values of dlro varywith I' : l7d1,,ll2 then each main reflection is sur-rounded by two satellites at distances of 2 B{ :2/[74,0] along the [31]x direction, as observed in theSAED pattern. This model also accounts for the zerointensi ty of nodes wi th f t . + fs : *1, *3, - r4, . . . aswell as the presence of satellites at distances of 2A{from hk reflections along [11.1]*. It also provides anexplanation of the origin of diffuse and elongated sat-

Frcuns 15. Mutual arrangements of Sr and H,O in the twooblique cells of twins. Large solid circles correspond to sitesoccupied by Sr in the twin components with A, and B, parame-ters. Large open circles correspond to Sr sites in the other twincomponent with A,* and B,* parameters. Small open circles cor-respond to HrO molecules

ellites (Figs. 5a and 6). As mentioned previously, theS cell contains an odd number of (310) planes passingthrough Sr sites. For this reason it is impossible to havean oscillation of d,,o with a period exactly equal toUdj,oll2. In the simplest case two periods may takeplace with ),,,: [4d1,ofl7 and }',: L3d.,o]l ' | (Fig. 13). Arandom alternation of these two periods with an aver-age periodicity of [7d1,o]/2 results in elongated diffusesatellites at distances of +2Bt from hk reflections. Acomplete treatment of this structural model would un-doubtedly give continuous streaks instead of the sharpmaxima observed experimentally.

SrBi type II and III. Figure 14 shows the S cell withA, : '/z(A + 38) and B, : Vz(l3B - A) corresponding tothe diffraction data shown in Figure 8. This cell containseight sites for Sr and eight (310) planes passing throughSr cations. Again, we can assume that Sr cations are shift-ed in the direction normal to (310) planes in such a waythat dlro values vary with the wavelength )t : [8d1ro]12because the satellites in Figure 8 are placed along the[31]x direction at distances of !28!. Taking into accountthat the S cell is base centered and contains an even num-ber of (310) planes containing Sr, a model for the dis-placement of Sr within the S cell, with I : 4dtro can beproposed (Fig. 1a). This model explains the presence ofsatellite pairs at distances of +2Af from hk reflectionsalong [3.1]*. The diffuse character of the satellites andtheir elongation in the [31]* direction are induced by thevariable values of Sr displacement throughout the crystal.Consequently, an irregular fluctuation of ,\ occurs in thecrystal and again elongated streaks are observed, insteadof sharp reflections. As mentioned previously, the reflec-tions with ft, + k : -r4, +12 are very weak but sharpbecause Sr cations shifted from their ideal sites (Fig. 14)give zero contribution to the intensity for thesereflections.

Finally, consider SAED patterns for the third type ofSrBi microcrystal, which contains only sharp reflections(Fig. 5d). Diffuse and elongated maxima of Figure 5c aretransformed into normal point reflections in Figure 5d. Itis likely that the probability of the Sr site occupancy isnow the same for all supercells of each twin. Figure 15shows the relationship between the supercell with A : 3a: 15.59 A and B : b : 2.83 A and the two twinnedoblique supercells with A : %(d, + b2)t/2 : 8.87 A, A :4b : 1I.32 A, and y: 118.7".

The corresponding reciprocal unit cell [AS : Yr\ge*+ B*) and Bt : y"(B* - 3A*)l agrees with the reflectionpositions observed experimentally (Fig. 5d). There are 12available sites and the supercell contains 1.8 Sr cations.The cell would therefore contain two sites with a siteoccupancy probability of 9OVo. A possible model for thedistribution of Sr is represented in Figure 15. In eachtwin, Sr is surrounded by six HrO molecules and the pe-riodicity of the projection of Sr along the A axis is Al2.This position of Sr atoms is almost certainly determinedby the ordered distribution of Mn4* and Mn3* atoms inadjacent birnessite layers as discussed below.

t?

I

I

I

IrIl '

III s r

, f>+

- 1 - fl l

t r t l

J I

t t lr t l

r t t ll . l . l . l


DrscussronOrigin of the layer charge

The analysis of satellites for Na-rich and Sr-exchangedbirnessite microcrystals indicates a dynamic fluctuationof the positions of scattering atoms. This displacementcan be achieved very easily in the interlayer space be-cause of the great mobility of the interlayer species inresponse to layer charge distribution. In addition, the va-riety of diffraction effects observed for NaBi and SrBi,all having very similar layer structure and cation com-positions, can hardly be explained by different layer va-cancy distributions. Thus we can infer that the mainsource of negative layer charge is the presence of Mn3*cations, rather than vacancies. The presence of a consid-erable amount of Mn3* in the layer of NaBi is supportedby other considerations. One explanation for the distor-tion from the hexagonal symmetry of the layer, demon-strated by the deviation of the alb ratio from V3, is thepresence of Mn3* cations and the consequent Jahn-Tellerdistortion. One could assume that the long bonds of thedistorted octahedra are oriented along the a axis of themonoclinic subcell, leading to a relative lengthening ofthe a parameter in comparison with b. In the layer struc-ture of crednerite (T<ipfer et al. 1995) the layers containonly Mn3* and the unit cell is characterizedby a: 5.578,b : 2.881 A; alb : 1.936. A comparison of these valueswith those for NaBi (a : 5.115 and b : 2.841 A,; alb :

1.818) shows that whereas the b parameter has a lowsensitivity to the presence of Mn3*, the c parameter hasa very high sensitivity. If we assume that a : b\,5 --

4.93 A for a pure Mna* layer and that there is a linearrelation between the Mn3*/Mna* ratio and the elongationof the a parameter, then the fraction of Mn3* in NaBilayers would be 0.38. This very crude calculation is con-sistent with the measured exchange capacity (Silvester etal. 1997) and seems a reasonable explanation of the layerdistortion.

However, as discussed below, the replacement of onetype of interlayer cation by another may be accompaniedby the appearance of layer vacancies. For example, Mgcations in MgBi occupy sites above or below layer cationsites (Post and Veblen 1990). If these layer sites wereoccupied by Mn cations then a strong electrostatic repul-sion would exist. These layer sites must therefore be va-cant to provide an energetically favorable O triangle forMg adsorption. Furthermore, it is difficult to believe thatthe decreise of the a parameter from NaBu (5.23 A) toNaBi (5.16 A) and from Mg-exchanged buserite (5.173A1 to vtgni @.94 L) described by Kuma et al. (1994) isdue only to the interlayer structure modification. A sim-ilar decrease of the a parameter has been described forthe Mg-for-Na exchange in birnessite both by Post andVeblen (1990; from 5.175 to 5.05 A) and by Kuma et al.(1994; from 5.16 to 4.94 A\. The a parameter of Be-ex-changed birnessite (4.915 A) is such that the alb ratioequals the t/5 ratio corresponding to ideal hexagonalsyrnmetry. Such strong variations of the a parameter are

Frcuns 16. Ordered arrangement of Mn3* and Mna* cationsamong available octahedral sites. A supercell with A : 3a andB : 3b and an ordered distribution ofNa (solid circles) and HrO(open circles) is shown.

more likely due to changes in the composition of layercations, in particular the Mn3*/Mno* ratio, a high a valueindicating a high proportion of Mn3*. In"agreement withthis assumption, the dehydration of 10 A NaBu and thesubstitution of Mg or Be for Na would be accompaniedeither by the oxidation of Mn3* or the disproportionationof Mn3* to Mn2* and Mno* and the concomitant formationof layer vacancies (Silvester et al. 1997).

Cation ordering in Mn layers

An interesting question is how Mn3* is distributed inthe layers. Because of the Jahn-Teller distortion, a randomdistribution of Mn3* and Mn'* among available sites com-bined with a random azimuthal orientation of Mn3* oc-tahedra would lead to unfavorable steric strains. The sim-plest way to avoid layer strains is the distribution of Mn3*in rows parallel to b and the orientation of the longMn3*-O bonds along the a axis and, as a consequence, toassume that Mn3* cations tend to order. Then, each Mn3*would share two edges parallel to (010) with nearest Mnr*octahedra and foui edges parallel to (310) and (3T0)planes with Mno* octahedra (Fig. 16). In addition, suchan arrangement of heterovalent Mn should provide a sim-ilar degree of undersaturation for layer O atoms. As canbe seen in Figure 16, O atoms coordinated by twoMn3*and one Mna* will form two short Mn3*-O bondsand one long Mna*-O bond. Oxygen atoms coordinated

l o'cn

r{


by two Mna* and one Mn3+ will form two short Mno*-Obonds and one very long Mn3*-O bond. The differentdimensions of Mn3* and Mn4* rows along the A axis arethe main reason for the deviation of NaBi and SrBi layersfrom the hexagonal symmetry. Figure 16 shows that Mn3*and Mna* octahedral rows alternate regularly along the Aaxis in the sequence Mn3+Mn1+Mn4+Mn3+ leading, to-gether with the ordered distribution of interlayer Na andH,O, to a supercell with A : 3a and B : 3b. The max-imum proportion of Mn3* is then 0.33. Additional supportfor this model comes from the Na:Mn ratio for NaBiequal to 0.33, which is consistent with a l:1 Na:Mn3*ratio. This structural model allows us to explain the mainexperimental diffraction features. First, the origin of thesupercell parameter A : 3a for NaBi and SrBi is evident.Second, without the assumption of Mn3* ordering it isdifficult to explain why Na and Sr cations, which differin valence, size, and hydrolysis properties, occupy thesame crystallographic positions within the unit cell withA : 3a and .8 : b or B : 3b. In the absence of Mn3*rows the distribution of interlayer cations described pre-viously for these microcrystal types would probably notbe the lowest energy configuration. The high concentra-tion of Na and Sr cations along Mn3" rows can be justifiedonly if they are forced in this arrangement to provide alocal charge compensation (e.g., Fig. 10). Finally, if theformation of layer vacancies during the NaBi to HBitransformation results from the migration of Mn3* fromlayer to interlayer sites thus forming layer vacancies(Mni"i", -+ n,,* + Mn3,i-,,".), then the distribution of layervacancies should reflect the former distribution of Mn3*.As a consequence, layer vacancies should also be orderedaccording to the supercell with A : 3a and B : b. Asmentioned previously, the superposition along c of inter-layer Mn and vacant layer sites strongly weakens the in-tensity of hlil superreflections. However, interlayer HrOmolecules coordinated to interlayer Mn are distributed inthe same way as vacant sites and enhance this intensityslightly. This regular distribution of interlayer species ac-cording to the predicted unit cell (Fig. 2c) confirms thatlayer vacancy distribution reflects the distribution of Mni*in the original NaBi structure.

Structure of the Na-rich birnessite interlayerIn both types of NaBi superstructures, adjacent layers

are shifted by -0.327a and the super-cells have a com-mon periodicity of A = 3a. However, they differ by theirI periodicity. In this section the interlayer structure ofboth types of NaBi is described in detail. The total layercharge and the mutual arrangement of Mna+A{n3+ cationscan then be derived.

NaBi type II: Supercell with A = 3a, B = 3b, and y= 90". The cell shown in Figure 17 contains six Na atomseach of which surrounded by six H,O molecules. Thecation content is Nao..Mnfi.trM{1r. Due to the ordereddistribution of Mna* and Mn3*, successive rows of O at-oms are coordinated to two Mn3* and one Mna", to twoMna* and one Mn3*, and to three Mno*. These different

l o

f+

FrcunB 17. Mutual arrangement of Mn3* and Mna* in twoadjacent layers within the supercell with A - 3a and B : 3b inthe idealized case where c cos1 / a : -0.333 instead of theobserved value of -0.327. Large solid circles represent Na sites.The corner of the triangles correspond to overlapping O atomsof the adjacent layers. Na and HrO are distributed along b ac-cording to the sequence Na-H,O-HrO-Na. . . .

O atoms differ by their degree of undersaturation of pos-itive charge. Accordingly, Na cations will possess differ-ent bonding strength with nearest O atoms. Figure 17shows that two-thirds of Na cations are located betweentwo saturated O atoms of adjacent layers whereas the oth-er third of Na cations are associated with undersaturatedO atoms.

NaBi type I: Supercell with A = 3a, B = D, and 1= 90o. Let us assume (l) that all Na atoms associatedwith Mn3* are placed only in rows separated from eachother by distances of Al2 along the A axis and are dis-tributed according to the sequence Na-HrO-Na-HrO. . .and (2) that the rows are shifted at random relative toeach other by t% (Fig. l8). Such an arrangement of in-terlayer species leads to B : b as observed experimen-tally. The maximum amount of Na and Mn3* per cell isthen equal to one atom and the cationic composition cor-responds to Nao,urMnfi-Mniju,. In this structural model,Mn3*-rich rows now contain only 5OVo of Mn3". As canbe seen in Figure 18, any additional Na in rows adjacentto Mn3'-rich rows results in the formation of Na-Na pairsat distances of 2.85 A. To increase the amount of Na andMn3* within the same supercell from 0.167 up to 0.20-0.25 atoms per octahedron, Na and HrO would have to

3E'


l o

il

l<

Frcunn 18. Model for the ordered distribution of Na andHrO along b axis, with the sequence Na-HrO-Na-HrO-Na. . Therows containing Na and HrO are separated from each other byAl2 along the A axis and are displaced at random by +b/,. Na:Mnrn ratio is l:6.

altemate along the rows in a disordered way (e.g.,Na-HrO-Na-HrO-HrO-Na-HrO. . .). In layer domains witha negative charge greater than 0.167 v.u. per octahedron,Na associated with Mn3*-rich rows should exhibitNa-HrO-HrO-Na subsequences. In this case additional Namay be placed between Mn3*-rich rows to provide thelayer charge comp^ensation and to avoid Na-Na pairs atdistances of 2.85 A.

S cell. In spite of the different superstructures, bothNaBi types have a common periodicity along b for thestructural modulations leading to the appearance of sat-ellites. Several models may be proposed to explain thisstructural feature of NaBi. For instance, one may assumethat within the Mn3* rows there is a regular substitutionof Mna* for Mn3" according to the sequence Mn4+-5Mn3+-Mn4+-5Mn3+. Such an ordered distribution of Mn3t andMnlt results in an S cell with As : 3a and B, : 6. Thisordered sequence of heterovalent Mn cations should leadto a periodic distribution of interlayer cations along theb axis and therefore to a modulation of d. It is easy tocalculate that the S cell contains 26 Mno* and l0 Mn3*,which corresponds to an Mn4* content of 0.123 per oc-tahedral site. This value is close to the experimental valueof 0.734 (Silvester et al. 1997). Another way to create aperiodic charge distribution along Mn3* rows results fromMn3* disproportionation. Let us assume that with the pe-riodicity I : 6b two Mn3* cations transformed in oneMna* and one Mn2" (Fig. 19). Within the supercell therewould be 26 Mn'", 8 Mnr*, and 2 Mn2* within the layer.According to this extreme case the cationic compositionof NaBi calculated per octahedron can be writtenNfo..,(M$ rrM$jr.M4irr). Again, different values ofnegative charge distributed periodically in Mn3*-richrows should lead to a periodic variation of interlayer cat-ion contents and of d values. The same effect should be

Frcunn 19. Periodic distribution of Mna* and Mn'?* alongMn3*-rich rows in the supercell with A : 3a and B : I : 6b.The periodic distribution of Na (solid circles) and H'O (opencircles) with A : 3a and B : 3b supercell parameters is alsoshown.

observed if Mn'z* migrates in the interlayer with simul-taneous formation of vacant layer sites. This structuralformula obtained for NaBi is very close to that derivedindependently from solution chemistry (Silvester et al.1991).

The present structural model for monoclinic birnessitepresents some similarities with that of Kuma et al. (1994).Both are based on a linear distribution of negative charg-es. However, Kuma et al. (1994) assumed that the nega-tive layer charge results only from layer vacancies, andthey did not consider the possibility of heterovalent Mncations. In contrast, the present model assumes that thesource of negative charge is due mostly to layer Mn3*,the density of layer vacancies being low, if any. Further-more, it offers the advantage of accounting for the largevalue of the a parameter and its decrease upon the for-mation of layer vacancies through either cation exchange,direct oxidation of Mn3" and Mnz*, or NaBu to NaBitransformation by means of the Mn3* disproportionation.In addition, this model allows the derivation of all ex-perimental electron diffraction patterns (Figs. 2 and 5) byintroducing only slight displacements of interlayer cationsfrom their ideal positions described by the supercell withA : 3 a a n d B : b .

Structural heterogeneity of birnessite

Whatever the interlayer cation, synthetic birnessitesamples usually exhibit two or three different types ofSAED patterns. Several reasons can be suggested for thisdiversity of diffraction patterns, including the density ofthe layer charge, the nature and distribution of heterova-lent cations in layers, and the nature of interlayer cations.

NaBi. We assumed that microcrystals having the unit


cell A : 3a and B : 3b have a higher layer charge thanmicrocrystals with A : 3a and B = b. Two opposingforces control the distribution of interlayer cations: (1)their tendency to distribute homogeneously to minimizetheir electrostatic repulsion and (2) the requirement ofproviding local charge compensation. For microcrystalswith a relatively low charge, the latter situation seems todominate (NaBi type I), whereas ar high layer charge in-terlayer cations tend to distribute homogeneously (NaBitype II). However, it is noticeable that in both NaBi va-rieties the origin of satellites is the same: They appear asthe result of the periodic variation of interplanar spacingalong [01] with the same wavelength of I : 6b. Thisresult supports the idea that the Mn3+-Mn4+ distributionalong Mn3*-rich rows is independent of the layer charge.

SrBi. The diversity of SAED panerns is greater withSrBi than with NaBi. This observation indicates thatstructural heterogeneities of layers caused by lateral vari-ations of charge (i.e., distribution of heterovalent cations)are more easily achieved with Sr because of its large ionicradius, smaller content, and divalent character. In conrastto Na, Sr cations are shifted along the normal to (310)planes (Figs. 13 and 14). The displacement in this direc-tion most likely reduces the electrostatic repulsion be-tween divalent cations. The varying periodicity of dlropossibly reflects variations in the layer charge as well assome variation in Mnr" and Mna* ordering. In contrast toNaB_i, the regular structural fluctuation along the normalto (310) does not lead to a fluctuation of do,o despite thefact that Sr cations are placed on the planes parallel to(010). This effect comes from the fact rhat the projectionon the b axis of all individual displacements of Sr atoms,from their ideal position with a given y coordinate, isequal to zero when averaged over a 4A period.

It is worth mentioning that twins of SrBi microcrystalsexhibit a strict periodicity. It is likely that the most stablestructure, with a complete local compensation of layercharge, occurs when the position of Sr cations in twonearest interlayers are connected by a glide plane (Fig.l5). The last variety of SrBi is characrerized by a prim-itive unit cell with A : 3a and B : 4b. As can be seenin Figure 10, Sr cations are localized in the same levelalong the A axis as in all other structural varieties. Thepeculiar distribution of Sr most probably reflects a spe-cific ratio and periodicity of Mn3+ and Mno* along Mn3*-rich rows.

AcxNowr,nocMENTS

The authors thank Bruno Lanson for his helpful scientific insights andLaurent Eybert-Berard for his valuable assistance in prepring the figures.V.D acknowledges financial suppoft from CNRS. V.D. and A G acknowl-edge financial support from the Russian Science Foundation (grant 95-051409) In the preparation of this manuscript in final form, the authorswere kindly assisted by Bruno Lanson

RnpBnnNcns crrnDBalistrieri, L S and Munay, J.W. (1982) The surface chemistry of DMnO,.

Geochimica et Cosmochimica Acta, 46, 104l-1152Bidoglio, G., Gibson, PN , O'Gorman, M , and Roberts, K J (1993) X-rav

absorption spectroscopy investigation of surface redox transformationsof thallium and chromium on colloidal mineral oxides. Geochimica etCosmochimica Acta, 51, 2389-2394

Bricker, O (1965) Some stability relations in system Mn-O,-H,0 at 25'Cand one atmosphere total pressure. American Mineralogist, 50,1296-1354

Burns, R.G and Burns, V.M (1976) Mineralogy of fenomanganese nod-ules In G P Glasby, Ed, Marine manganese deposits, p. 185-248. El-sevier, Amsterdam.

-(1977) The mineralogy and crystal chemistry of deep-sea man-ganese nodules, a polymetallic resource of the twenty-first centuryPhilosophical Transactions of the Royal Society of London (A), 286,283-301.

Chukhrov, FV and Gorshkov, A I. (1981) Iron and manganese oxide min-erals in soils. Transactions of the Royal Society of Edinburgh,72, 195200

Chukhrov, FV, Corschkov, A.I., Rudnitskaya, E S , and Sivtsov, A.V.(1978) Bimessite characterization. Investiya Akademie Nauk, SSSR,Seriya Geologicheskaya, 9, 67 76

Chuktuov, FV., Sakharov, B A, Gorshkov, A I., Drits, V.A., and Dikov,YP (1985) Crystal shxcture of birnessite from the Pacific Ocean. In-ternational Geology Review, 27, 1082-1088 (translated from InvestiyaAkademie Nauk, SSSR, Seriya Geologicheskaya, 8, 66-73).

Chukhrov, EV, Gorshkov, A.I, Berezovskaya, V.V., and Sivtsov, A V.(1987) New data about mineralogy of Kertch ore deposits InvestiyaAkademie Nauk, SSSR, Seriya Geologicheskaya,4, 60-77.

Chukhrov, RV, Gorshkov, AI., and Drits, V.A (1989) Supergenic man-ganese hydrous oxides, 208 p. Nauka, Moscow

Cornell, R M and Giovanoli, R (1988) Transformation of hausmanniteinto birnessite in alkaline media Clays and Clay Minerals, 36, 249-257

Drits, VA. (1987) Electron diffraction md high resolution electron mi-croscopy of mineral structures, 304 p. Springer-Verlag, Berlin.

Drits, VA , Petrova, VV, md Gorshkov, A.I (1985) Manganese mineralsof Fe-Mn nodules from the sediments of the central part of PacificOcean and their post-sedimentation transformation Lithology and RawMaterials. 3. 17-39.

Giovanoli, R and Arrhenius, G. (1988) Structural chemisky of marinemanganese and iron minerals and synthetic model compounds In PHalbach, G. Friedrich, and U von Stackelberg, Eds, The manganesenodule belt of the Pacific Ocem: Geological environmental, nodule for-mation, and mining aspects, p 20-37 Verlag, Stuttgart

Giovanoli, R, Stahh, E, and Feitknecht, W. (1970a) Uber Oxidhydroxidedes vierwertigen Mangans mit Schichtengitter 1. Mitteilung: Natrium-mangan(Il,Ill)manganat(IV) Helvetica Chimica Acta, 53, 454-464

-(1970b) Uber Oxidhydroxide des vierwertigen Mangans mit Schich-tengitter. 2 Mitteilung: Mangan(Ill)-mangana(IV). Helvetica ChimicaActa.53. 454-464

Giovmoli, R., Feitknecht, W, Maurer, R., and Hani, H (1976) HomogeneKeimbildung und Keimwachstum von 1MnO,. Chimia, 30, 268-269

Glover, E.D (1977) Characterization of a marine bimessite AmericanMineralogist, 62, 27 8-285.

Gorshkov, A.I , Drits, VA, Putilita, VS., Pokrovskaya, E.V., and Sivtsov,AV. (1992) Natural and synthetic bimessites Lithology and Raw Ma-terials. 6. 67-81.

Healy, T,W., Herring, AP, and Fuerstenau, DW. (1966) The effect ofcrystal structure on the surface properties of a series of mangmesedioxides. Journal of Colloid and Interface Science. 21. 435-444.

Jones, L.H.P and Milne, A.A. (1956) Birnessite, a new manganese oxidemineral from Aberdeenshire, Scotland Mineralogical Magazine, 31,283-288

Kuma, K., Usui, A, Paplawsky, W., Gedulin, B , and Arhenius, G. (1994)Crystal structures of synthetic 7 A and 10 A manganates subsrituted bymono- and divalent cations Mineralogical Magazine, 58, 425-447

Le Goff, P, Baffier, N., Bach, S , md Pereira-Ramos, J P (1996) Synthesis,ion exchange md electrochemical properties of lamellar phyllomanga-nates of the bimessite group. Material Research Bulletin, 31,63 75

Manceau, A and Charlet, L (1992) X-ray absorption spectroscopic studyof the sorption of Cr(II! at the oxide-water interface. Journal of Colloidand Interface Science, 148, 425-442

DRITS ET AL: BIRNESSITE STRUCTURE 961

Manceau, A., Gorshov, AI, and Drits, VA (1992) Structural chemistryof Mn, Fe, Co and Ni in manganese hydrous oxides: Part II Informationfrom EXAFS spectroscopy and electron and X-ray diffraction Ameri-can Mineralogist, 71, 1144-1 15'7

Munay, J.W. (1974) The surface chemistry of hydrous manganese dioxide.Joumal of Colloid and Interface Science. 46, 357-371

Paterson, E , Swaffield, R., and Clark, L (1994) The influence of structureon Ba and K uptake by a synthetic phyllomanganate Clay Minerals,t / a t < a a a

Post, J.E and Appleman, D.E (1988) Chalcophanite, ZnMn.O,3H,O:New crystal structure determinations. American Mineralogist, 73,1401-1404.

Post, JE and Veblen, DR (1990) Crystal structure determinations ofsynthetic sodium, magnesium, and potassium birnessite using TEM andthe Rietveid method American Mineralogist,

'75, 4'77-489.

Silvester, E , Manceau, A , and Drits, V.A ( 1997) The structure of mono-clinic Na-rich birnessite and hexagonal birnessite: II Results fromChemical Studies and EXAFS Spectroscopy American Mineralogist,82,962-978

Stone, A.T (1987) Reductive dissolution of manganese (III,IV) oxides bysubstituted phenols. Environmental Science and Technology, 21,979-988.

Stone, A T. and Morgan, J.J. (1984) Reductive dissolution of manganese(III) and manganese (IV) oxides by organics Environmental Scienceand Technology, 18, 6l'7 -624.

Stone, A T., Godtfredsen, K L., and Deng, B (1994) Sources and reactivityof reductant encountered in aquatic environments In G. Bidoglio and

W. Stumm, Eds., Chemistry of aquatic systems: Local and global per-

spectives, p 331-374. EURO courses senesStrobel, P, Charenton, J C, and Lenglet, M. (1987) Stuctural chemistry

of phyllomanganates: Experimental evidence and structural models. Re-

vue de Chimie Min6rale, 24, 199-220.Taylor, RM, McKenzie, RM, and Norrish, K (1964) The mineralogy

and chemistry of manganese in some Australian soils Australian Jour-

nal of Soil Research, 2,235-248.Tdpfea J, Trari, M , Gravereau, P, Chaminade, J P, and Donmerc, J.P

(1995) Crystal growth and re-investigation of the crystal structure of

crednerite, CuMnO,. Zeitschrift fiir Kristallographie, 210, 1 84- 187

Tu. S, Racz, G.J., and Goh, TB (1994) Transformation of synthetic bir-

nessite as affected by pH and manganese concentration Clays and Clay

Minerals, 42,321-330.Usui, A and Mita, N. (1995) Geochemistry and mineralogy of a modern

buserite deposit from a hot spring in Hokkaido, Japan. Clays and Clay

Minerals, 43, 116 127.Wadsley, A.D. (1955) The crystal structure of chalcophanite,ZnMnO,3

H,O Acta Crystallographica,S, 165 l'72.Wilson, AJ.C (1964) X-ray optics: The diffraction of X-rays by finite

and imperfect crystals, 147 p. Wiley, New York.Xyla, A G, Sulzberger, B., Luther, G.W., Hering, J G, Van Capellen, P,

and Stumm, W. (1992) Reductive dissolution of manganese(Ill'IV)(hydr)oxides by oxalate: The effect of pH and light Langmuir, 8, 95-

103

Maruscrrpt RECENED Jurw 6, 1996M,q.uuscrIpt ACCEnED Mn 20, 1997

Documents

Structure of synthetic monoclinic Na-rich birnessite and hexagonal