American Mineralogist, Volume 77, pages 1144-1157, 1992

Structural chemistry of Mn, Fe, Co, and Ni in manganese hydrous oxides: Part II.Information from EXAFS spectroscopy and electron and X-ray diffraction

Ar,arN MeNcrau*Laboratoire de Min6ralogie-Cristallographie, Universitds Paris 6 et 7, CNRS UA09, Tour 16,4 place Jussieu,

75252Pais Cedex 05, France

ANaror-rr I. GonsmovInstitute of Ore Geology and Mineralogy (IGEM) of the USSR Academy of Science, 35 Staromonetny prospekt,

109017 Moscow, Russia

Vrcron A. DnrrsGeological Institute of the USSR Academy of Science, 7 Pyzhevsky prospekt, 1090 I 7 Moscow, Russia

Arsrn-lcr

The short- and long-range structural order ofvarious hydrous oxides were investigatedusing extended X-ray absorption fine structure (EXAFS), X-ray and electron diffraction(XRD and SAED) and in situ chemical analysis (EDS). Materials examined include syn-thetic sodium birnessite, natural magnesium birnessite, natural nickel copper birnessite,synthetic vernadite (6-MnO), a series of natural iron vernadite samples, natural manga-nese goethite, two natural cobalt nickel asbolane samples, and natural cobalt asbolane.The structure of birnessite is similar to that of chalcophanite, but there are no corner-sharing Mn4+-Mn4+ octahedra. Sodium and magnesium birnessite differ primarily in thestacking mode of their Mn octahedral sheets and their interlayer structure. Similaritiesand differences between these two minerals have been considered in terms of anion close-packed models. In contrast to birnessite, synthetic vernadite (6-Mnor) has edge- and cor-ner-sharing Mn4+ octahedra and a three-dimensional anionic framework; d-Mno, doesnot appear to be a c-disordered birnessite. In iron vernadite and manganese goethite, Fe3+and Mn4+ ions are segregated in coherent scattering domains. In both minerals, Mn atomsform phyllomanganate-like domains; in iron vernadite, Fe domains have a feroxyhiteJikelocal structure. Asbolane has been found to have a mixed-layer structure. Cobalt nickelasbolane has layers of Mnor, N(oH)r, and possibly coooH or co(oH), alternating reg-ularly along the c axis. In cobalt asbolane, Mn2+ tetrahedral layers are presumably regularlyinterstratified with layers of Mno, and (CoooH). This study provides new examples ofhybrid structures among low-temperature materials and confirms their heterogeneous na-ture on a very fine scale.

INrnonucrroN

Characteization of many hydrous manganese oxidesby structural methods is difficult and frequently incom-plete because of their poor crystallinity. Structural meth-ods can be divided into two groups: the diffraction tech-niques are sensitive to medium- and long-range order,and the spectroscopic techniques are sensitive to elec-tronic or local structure. Our present knowledge of hy-drous manganese oxide structure is based mostly on theformer. The tectomanganates have a relatively good crys-tallinity and have been well characterized. Significantstudies ofthe past I 5 years include Chukhrov et al. (l 978a,1979, 1981, 1985a, 1986), Turner and Buseck (1981),Post et al. (1982), Miura (1986), Vicar et al. (1986), andPost and Bish (1988). In contrast, most phyllomanga-

* Present address: Universit6 Joseph Fourier, LGIT, Bp 53X.38041 Grenoble Cedex, France.

nates have a lower degree of crystallinity, so few of theirstructures have been refined. Some studies of phylloman-ganates include birnessite and buserite: Giovanoli et al.(1970a,1970b), Chukhrov et al. (1985b, 1987a), Strobelet al. (1987), and Post and Veblen (1990); vernadite:Chukhrov et al. (1988); asbolane: Chukhrov et al. (1980,1987b), Drits (1987), Manceau et al. (1987); and mixed-layer buserite-asbolane: Chukhrov et al. (1983, 1987b).

The main difficulties in the study of poorly crystallizedhydrous manganese oxides are that (l) their X-ray dif-fraction (XRD) patterns usually contain only a few broad,weak reflections, and (2) several minerals (buserite, as-bolane, mixedJayer asbolane-buserite) give nearly thesame XRD patterns (Drits et al., 1985). Recently, how-ever, some progress has been made in simulating diffrac-tion effects of defect structure models (Drits and Tchou-bar, 1990; Chukhrov et al., 1989), and in applying theRietveld method for structure refinement of some of the

ooo3-004w 9 2 / | | | 2- | I 44$02. 00 r144

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II 1 1 4 5

better crystallized samples (Post and Veblen, 1990). Se-lected area electron diffraction (SAED) has also provenuseful for the structural analysis of hydrous manganeseoxides. Two characteristics that distinguish this methodfrom XRD are its higher sensitivity to long-range order(Drits, 1987) and the possibility of focusing the electronbeam on a small volume, thus allowing identification ofdiscrete phases in a polymineral matrix. For instance, thecombination of XRD and SAED completed with in situchemical analysis (energy dispersive spectroscopy, EDS)has permitted the determination of the structure of metal-containing asbolanes and mixedlayer buserite-asbolanes.However, even at the micrometer scale, diffraction meth-ods yield average structures at best. These are often notthe same as actual structure for at least two reasons: (l)natural manganese oxides often contain large amounts ofFe, and the similarity in scattering power of Mn and Feprevents their distinction by diffraction techniques; (2)minor and trace elements in manganese oxides cause lo-cal zones ofdiferent structure.

Several spectroscopic methods have been used to refinethe structural chemistry of manganese oxides. Some ofthese, such as extended X-ray absorption fine structure(EXAFS) and infrared spectroscopy (IR), are sensitive tolocal or near-local structure (Potter and Rossman,1979;Manceau et al., 1987; Manceau and Combes, 1988; Stouffand Bouldgue, 1988; Manceau, 1989; Manceau et al.,1989, 1990a). The main advantages of X-ray absorptionspectroscopy (XAS : EXAFS + XANES) are threefold:(l) almost all elements are spectroscopically active; (2)the structural information comes from more than thenearest coordination sphere of the X-ray absorber; and(3) the spectra from specific elements can usually be ob-tained separately. These aspects make XAS a valuabletool for determining a particular atom's local envrron-ment and for probing its electronic structure. The mainlimitation of XAS is its lack of spatial resolution in com-parison to SAED; that limits XAS use to single phases,i.e., where the X-ray absorber phase is unique.

This paper extends the work begun in the XANES studyof hydrous manganese oxides that was presented by Man-ceau et al. (1992). In this second part, we have combineddata from EXAFS, XANES, XRD, SAED, and EDS. Ouraims are (l) to clarify the structure of natural and syn-thetic birnessite, (2) to examine the possible structuralrelationship between vernadite and birnessite, (3) to de-termine the structural environments of Mn and Fe innatural iron vernadite and manganese goethite, and (4)to better define the structure ofasbolanes.

ExpnnrnnnNTAl DETAILs

Techniques

EXAFS spectroscopy. We used the EXAFS IV stationat the LURE synchrotron radiation laboratory (Orsay,France). The positron energy of the storage ring DCI wasl.85 GeV, and the current was between 200 and 250 mA.The incident beam was monochromatized with pairs of

o 2 4R (A)

6 0 2 4 6R (A)

Fig. I . MnK RDF of reference compounds, d-MnO, and bir-nessite. Solid line : room temperature (RT), dotted line : lowertemperature. (a) From lowest upward: buserite (RQ, 6-MnO.(RT, 4.5 K), todorokite, ramsdellite, and chalcophanite (RT, 30K). Notice the similarity of RDF of todorokite and 6-MnO'.Arrow indicates MnO" cornerJinked contributions. (b) Fromlowest upward: sample l[ {l2 (RT,77 K), sample Bl (Rl 77 K),synthetic birnessite (NaoMn,oO','9HrO) (RZ, 4.5 K). RDF havenot been corrected for phase shifts.

reflecting Si (31l) crystals. Harmonics due to high orderreflections were eliminated with borosilicate glass mirrors(Sainctavit et al., 1988). EXAFS data were reduced usingthe plane-wave approximation with a standard procedure(Teo, 1986). To maximize the detection limit of atomiccontributions (namely corner-sharing octahedra), (l) weobtained spectra at room temperature and at low tem-perature (77 K or a.5 K) (Fig. la), and (2) we attemptedto reduce spurious peaks in the MnK RDF (radial distri-bution function) caused by cutoffeffects during the Fou-rier transform. These Fourier transform artifacts wereminimized by using the Kaiser window with a r param-eter of 3. In such conditions, the detection limit for cor-ner-sharing Me atoms (Me : Mn, Zn . . .) is lower thanone Me per Mn atom (Manceau and Combes, 1988).

X-ray and electron diffraction and energy dispersiveanalysis. XRD patterns of powdered samples were re-corded with a DRON-UM or with a Siemens diffractom-eter using CuKa radiation and a graphite monochroma-tor. SAED patterns and EDS spectra were obtained usinga JEM-100C microscope equipped with a Kevex spec-trometer with a voltage of 100 kV and a nominal current.Samples were dispersed in aqueous solution with an ul-trasonic disperser. Thin colloid films covered with C wereused as support, and grids were mounted on a tilting sam-ple holder. Techniques for treating SAED patterns havebeen described by Drits (1987).

Materials

Reference minerals. A series of well-crystallized man-ganese oxides (Manceau and Combes, 1988), and otherMn-, Fe-, and Co-bearing minerals were analyzed to serve

al<Fu-

Rll

LL

lr

; ,r €!!rE99!r9

natu ra l

M 1 2

l 146 MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

TodorokiteRamsdellite

GoethiteAkaganeiteLepidocrociteFeroxyhiteFerrihydriteHematite

Heterogenite

TABLE 1. Name and ideal composition of references and samples

Refer-Mineral ldeal composition ence

Manganese oxidesPhyllomanganates

Buserite (2-D) MnO2Birnessite (2-D) NaoMn,oO,, 9H"OAsbolane (2-D) Mn(O,OH),(N|,Co)IO,OH), nH,OChalcophanite(2-D') ZnMn3O?.3H2OVernadite (2-D') 6-MnO,

Tectomanganates (3-D)

which has Zn atoms lying above and below vacant Mnoctahedra), have a few (Fig. 2;Partl).

Birnessite

The XRD pattern of our synthetic sample is quite sim-ilar to that ofsynthetic sodium birnessite studied by Postand Veblen (1990) (data not shown). The measured dvalues indicate a one-layer, monoclinic unit cell havingpa rame te rs a : 5 .174 A , b :2 .848 4 , , c :7 .336 4 ,0 - -

103.2.In contrast, our natural samples have a hexagonalu n i t c e l l w i t h a : b : 2 . 8 4 A , s : 7 . 1 0 A , 7 : 1 2 0 ' ,similar to the results of Chukhrov et al. (1985b).

SAED patterns of synthetic birnessite particles showthe same reflections as XRD patterns, but in additionthey show a weak set ofikO super reflections as describedby Giovanoli et al. (1970a), Chukhrov et al. (1989), andPost and Veblen (1990). Thus the unit-cell parametersdetermined from XRD do not correspond to the actualdimensional periodicity in the ab plane. Extra reflectionsindicating superperiodicity are missing on SAED patternsof natural birnessite particles.

The MnK RDF for the birnessite samples are plottedin Figure lb; the curve for natural sample Bl is identicalto that of the synthetic one. They show a single Mn peakcorresponding to a Mn4+ shell at 2.9 A. The other naturalbirnessite sample (Ml2) shows a small second Me con-tribution (Me : Mn, Co, Ni, or Cu) at about 3.4-3.5 Aat both analysis temperatures.

Vernadite

Diffraction features of vernadite (6-MnOr) and of theiron vernadite samples have been reported by Chukhrovet al. (1987c, 1988). The XRD patterns are nearly iden-tical; most show only two broad reflections, but somehave an additional, weak, shoulderlike reflection at d x2.2 A.We observed the same diffraction features in SAEDpatterns from most particles (Fig. 2a). However, a fewparticles have more Fe than Mn, and these display SAEDpatterns that have four ring reflections at 2.55, 2.24, | .7 | ,and 1.47 A lcompare Fig. 2a and 2b with Fig. 2c and 2d).The SAED patterns of these particles resemble that offeroxyhite (d-FeOOH); hence we assume that their struc-ture is similar to that of feroxyhite.

The MnK RDF for synthetic 6-MnO, resembles that oftodorokite (Fig. la). A similarity in structural chemicalfeatures of these two minerals was also observed in thepreedge and main-edge K XANES spectra reported inPart I. MnlK RDF for natural vernadite (Fig. 3a) alsoresemble that of todorokite, even to the point of display-ing a weak but significant (Mn,Fe) contribution at 3.4 A.

The most striking result from the vernadite study is theclose similarity of the FeK RDF for iron vernadite andferoxyhite (Fig. 3b). This likeness is confirmed by com-paring their Fourier-filtered Fe-(Fe,Mn) contributions (Fig.4a).The corresponding Fe-Fe contributions for goethiteare considerable different in phase and amplitude. Thesevernadite spectra were best frtted by assuming two near-est Fe-Fe pairings at 2.89 and 3.03 A, as in d-FeOOHGig. ab). Mn and Fe have similar backscattering ampli-

(Na,Ca,K)o tu ?(Mn,Mg)"O,,. nHrOa-MnO,

lron oxideso-FeOOHB-FeOOH"y-FeOOHd-FeOOH5FerOr.9H,Oa-FerO.

Cobalt oxideCoOOH

12345

o7

8o

1 01 11 21 3

1 4

Note:1 : Giovanoli (1980b); 2 : Post and Vebten (1 990); 3 : Drits (1 987);4 : Wadsley (1 953), Post and Appleman (1 988); 5 : Chukhrov et at. (1 97Bb),Giovanol i (1980a); 6: Post and Bish (1988); 7: Bystr6m (1949);8: Szytula et al. (1968); 9: Szytuta et at. (1970); 10: Ewing (1935);1 1 : Patrat et al. (1983); 12 : Towe and Bradtey (1967); 13: Btake etal. (1966); 14: Delaplane et al. (1969), Deliens and Goethats (1973).

as reference compounds. The minerals are listed in TableI along with their ideal chemical formulae.

Samples. The minerals under study include syntheticsodium birnessite, two natural birnessite samples (sampleBl which contains some Mg, Ca, K, and Na, and sampleMl2, which contains some Ni and Cu), synthetic vernad-ite (D-MnOr), a series of natural iron vernadite samples,natural manganese goethite, two natural cobalt nickel as-bolanes (samples asb and Mn-ox), and natural cobalt as-bolane. Details concerning the nature, origin, and chem-ical composition of the samples, as well as some selectedcrystallographic data, can be found in Manceau et al.(1992) (hereafter referred to as Part I). A summarv ofthose data is given in Table l.

Resur,rsReference minerals

Figure la shows the MnK RDF obtained for severalreference tetravalent manganates. The first three peaks inthe spectra of chalcophanite, ramsdellite, and todorokitecorrespond to contributions from the nearest O atomicshell and from the nearest two Mn shells, respectively(Manceau and Combes, 1988). For all manganese oxides,the first Mn shell is located at approximately 2.90 A andcorresponds to MnOu octahedra linked by edges. Thispeak is always present because MnO2 structures are builtof single or multiple octahedral chains (Fig. 2; Part I).The second Mn shell at approximately 3.4-3.5 A (anow)is characteristic of minerals that have MnOu octahedralinked by corners. Of interest, here, is the correlation ofthis next-nearest cation contribution with the type of Oframework. Two dimensional (2-D) layer structures (suchas buserite and lithiophorite) have no corner-sharing oc-tahedra while 2-D' structures (such as chalcophanite,

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II tt47

Fig. 2. SAED patterns and EDS spectra for Fe-bearing vernadite particles. (a), (b) Representative of most of the particles wherethere is a higher concentration of Mn than Fe. SAED pattems contain only two hk bands; (c), (d) particles where Fe is moreabundant than Mn. SAED pattern resembles that of feroxyhite.

tudes and phase shifts functions, so EXAFS cannot dif-ferentiate between the two; thus, a good fit would alsohave been obtained by assuming Mn instead of Fe nearestneighbors. However, the presence of Fe in the vicinity ofFe is supported by the fact that Mn and Fe atoms possessa different local structure.

Mn-bearing goethite

The XRD pattern for manganese goethite shows broad-ened hkl reflections that have approximately the same dvalues as those of goethite (Fig. 5). No manganese oxide

impurity was detected; a lack of even trace quantities ofgroutite is proved by the absence ofreflections near 2.81,2.38, and 230 A (see arrows). Figure 6a is a TEM imageof some sample particles; EDS analyses show that theyall contain similar amounts of Mn and Fe, although theconcentration ofFe is usually slightly greater (Fig. 6c). Inspite of a high Mn content, SAED patterns are charac-tenzed by the same reflections as those expected for goe-thite, which confirm the monomineralic nature of thismanganese goethite at the scale sampled by diffractiontechniques (Fig. 6b).

b n Fe K-edge

@

l 148 MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

U)zUJ

z

-LL

RvF|.r

0 2 4 6 0 z " 4 6R ( A )

- R ( A )

Fig. 3. EXAFS data for iron vemadite samples. (a) MnKRDF.Solid line : R7, dotted line : 77 K. Data were analyzed usinga Kaiser window with z : 3.0. (b) Compare shape of FeK RDFto that of feroxyhite. Data were analyzed using Hanning + flatwindow (Manceau and Combes, 1988). RDF have not been cor-rected for phase shifts.

The FeK RDF for this sample (Fig. 7a) is identical tothat of goethite (Manceau and Combes, 1988). The MnKRDF contains well-defined second and third peaks whoseintensities are comparable to those of todorokite (Fig. 7).The similarity in the local Mn environment in goethiteand in todorokite has been verified by Fourier-filteringMn-O and Mn-Mn contributions. The Mn-(O,OH) dis-tance is equal to 1.90-1.92 A, a value consistent with atetravalent state lor Mn ions.

Asbolane

Previous results showed that none ofthe asbolane sam-ples were completely monomineralic (see description inPart I). Sample asb and the cobalt asbolane from Ger-many also contain small amounts of cobalt buserite andcalcium buserite, respectively. Periodicity along the c axisfor asbolane and buserite are not the same; they are 9.5and 7 A, respectively (under vacuum conditions). SampleMn-ox contains lithiophorite as an impurity.

SAED patterns for most of the Co,Ni-bearing asbolane

r0 20 30 40 50 60 'to 802e degrees (Cu Koc radiation)

Fig. 5. XRD pattern for manganese goethite. No MnO, im-puriry is detected. Arrows correspond to the (130), (l I l), and( I 40) reflections of groutite.

particles of samples asb and Mn-ox contain two incom-mensurate hexagonal networks of hk} reflections and aunique series ofbasal 00/ reflections (Fig. 8). The struc-tures ofthese particles can be described by two sublatticeshaving different hexagonal unit-cell parameters in the abplane (a, : b , :2.84 A for sublat t ice l , and a, , : bu:3.01 A for sublattice II) but a common periodicity per-pendicular to them along c (c : 9.40 A;. Our SAED re-flection intensity distributions are the same as those forthe cobalt nickel asbolane and nickel asbolane particlesdescribed by Chukhrov et al. (1980). In each reciprocalsublattice the 100 reflection is about half as intense asthat of I l0 (Fig. 8), but for each given hk} reflection,reciprocal sublattice I reflections are stronger than recip-rocal sublattice II reflections (e.g., compare I l, and I1,,in Fig. 8a, 8c). EDS spectra show that Ni concentrationvaries in the particles of cobalt nickel asbolane. Our ob-servations show that with smaller Ni content, the hk0reflection intensities for sublattice II are weaker (Fig. 8).The particles containing only Mn and Co (Fig. 8e, 8f)show 00/ reflections with approximately the same peakposition and intensity distribution as those of Co,Ni par-ticles, but they show only one hexagonal network of hk}reflections corresponding to sublattice I.

The SAED pattern of the cobalt asbolane was describedby Chukhrov et al. (1982). In spite of an absence of Ni,the particles displayed patterns with two incommensu-rate hexagonal networks of hk} reflections that they at-tributed to two sublattices (I and II), with a, : 2.84 Aand arr: 3.15 A. ttre 100, I 10, and 200 reflection inten-sities belonging to sublattice I were the same as those forcobalt nickel asbolane particles [I(l l0) : 2I(100)] where-as 100 and I 10 reflections for sublattice II had approxi-mately the same intensity.

Co and Mn RDF for the asbolane samples are plottedin Figure 9a. The strong second peak attests to the pres-ence of a Me shell at 2.9 A. The appearance of a weakthird peak in the CoK RDF for sample asb was shownby analysis in the k space to result from a few corner-sharing (Co,Mn) octahedra. Such linkages are undetect-able at the MnK edge by both EXAFS and preedge spec-troscopy (Part I). One possible explanation for this effectis that the impurity in the cobalt buserite has a 2-D'struc-

0 0 2

o 0 1

Go€th i te0 0 1 5

o 0 r

o oo5

s o

0 0 0 5

- o 0 1

- o 0 1 5

a o

, o o 1

- o 0 21 0 1 2 4 6 8 1 0 1 2

K(4.) WAVEVECToR k(A 1)

Fig. 4. Fourier-filtered nearest Fe-Fe contributions toEXAFS. (a) Experimental data for goethite, feroxyhite, and ver-nadite sample 1268. (b) Fit for vernadite sample 1248. Noticethat the wave beating at about 9-10 A I is reproduced well byassuming two Fe shells at 2.89 and 3.03 A, corresponding to faceand edge linkage, respectively.

? 4 6 aWAVEVECTOR

o

o

.t_Y I

3=

I

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II 1149

>avFtL

642o

Fig. 6. (a) A TEM image of Mn-bearing goethite particles.(b) A SAED pattern of one of the particles; it resembles that ofpure goethite. (c) A representative EDS spectnrm.

ture. Analysis in the k space of the Co-O contributionsfor asbolane shows in phase backscattered waves (Fig.9b). The coincidence ofzero crossings proves that the Codistance is preserved from one sample to another. Man-ceau et al. (1987) reported 1.92 A for this Co-O bondlength. In order to test spectroscopic sensitivity to talCo3+,

a simulated Co-O contribution at 1.77 L has also beenplotted in Figure 9b. This wave is out of phase in the high

R (A)

b /,"1i Manganesegoethite

todorok i te

\ . \

R (A)Fig.7. EXAFS data for manganese goethite. (a) Comparison

of Mn and FeK RDF. (b) Similarity of the Mn-rK RDF of todo-rokite and manganese goethite. Notice their third peaks (arrow).RDF have not been corrected for phaie shifts.

k range (2kAR = zr) rendering the overall wave envelopeparticularly sensitive to the presence of a small amountof I4lCo3+. A set of simulations shows that the detectionlimit of fourfold-coordinated Co is less that l5-20010. NiKEXAFS spectra of the two cobalt nickel asbolane samplesare almost identical to that of Ni(OH)' (Manceau et al.,l 987) .

DrscussroN

Birnessite

Two-dimensional vs. 2-D' structure. According to Gio-vanoli and coauthors (Giovanoli et al., 1969, 1970a,1970b; Giovanoli and StehH, 1970), synthetic sodiumbirnessite has a two-layer, orthorhombic unit cell. Its

=><l<

FLL

6420

a it - Mn K-edge. . .FeK-edge

t\

Manganesegoethite

{/ i,Vl\iV \r",

I 150 MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II I l 5 l

structure was thought to be related to that ofchalcophan-ite. Both structures have edge- and corner-sharing octa-hedra, but in the former, Mn was proposed to substitutefor Zn it the interlayer region. With this 2-D' model, athird RDF peak typical of corner-sharing octahedra isexpected. Because of Mn substitution for Zn, rttetlayetMn atoms would absorb X-rays at the MnK edge, result-ing in a stronger peak than that which occurs for chal-cophanite (Fig. la). The detection limit of Mn neighborsbeing better than one atom at low temperature (Manceauand Combes, 1988), the absence of Mna+-Mno+ pairingsat 3.4-3.5 A (Flg. lb) indicates that no Mna* ions occurat lhe Zn position. However, the presence of some low-valency Mn atoms (e.g., Mn2* ) cannot be excluded fromthis study, as it would lead to larger Mn-Mn distances.The presence of Na, which has a small backscatteringpower, in the Zt position appears unlikely, since the ad-sorption of monovalent ions is usually nonspecific (Kin-niburgh and Jackson, 1981), i.e., Na* is adsorbed at min-eral surfaces by electrostatic attraction in the electricaldouble layer without establishing a chemical bond withthe surface.

On the basis of the calculation of XRD patterns fordifferent structural models, Chukhrov et al. (1985b) con-cluded that his birnessite sample (Bl) had a 2-D' struc-ture. However, with this method, it was impossible todetermine the distribution of isomorphic Mn and Mg at-oms among layer and interlayer octahedral sites. By com-parison with the chalcophanite structure, Chukhrov et al'(l9S5b) proposed that interlayer sites above and belowoctahedral layer vacancies would more likely be filled withMn than with Mg atoms. Now, in light of the preedgeresults (Part I) and the EXAFS results (no Mn cornersharing in sample Bl), we suggest that Mn4+ atoms aresolely located within octahedral layers, whereas Mg at-oms and possibly Mn2+/Mn3+ occupy interlayer octahe-dral sites. Because divalent cations have higher specificadsorption capacity than monovalent cations, Mg and, toa lesser extent, Ca seem to be better candidates than Na*for filling octahedral sites above and below vacancies inMnO, layers.

The very small peak on the right side of the secondRDF peak for sample Ml2 results from few Mno*, Co,Ni, or Cu neighbors ar 3.4-3.5 A (fig. lb). The presence

of Mn-Me corner-sharing octahedra at this distance canbe accounted for by the absence of Na, Ca, and Mg cat-ions and the presence of 3d elements. Thus, one mayconclude that both natural and synthetic birnessite struc-tures consist of octahedral layers containing Mna+, butthey have different layer stackings. These structures, then,

o .o5

0 . 0 3

0 . o 1a

- o . 0 1

- o . 0 3

- o . o 53 6 9 1 2

WAVEVECTOR k(A- t)

Fig. 9. EXAFS data for asbolane. (a) Comparison of nor-malized Mn and CoK RDF. Notice the presence of a small peakon the right side of the CoKRDF (arrow) indicating the presenceof corner-sharing octahedra. (b) Comparison of Fourier-filteredCo-(O,OH) contributions of New Caledonian and German as-bolane and simulation assuming 4 O neighbors at 1.77 A. Noticethe phase shift.

may have Mn-Me corner-sharing octahedra that are de-tectable by EXAFS except when the metal site is filledwith light cations like Mg or Ca. Where interlayer Na ispresent, strong corner sharing is unlikely.

Layer stacking. One of the interesting problems in de-termining the structural chemistry of birnessite is deter-mination of the factors responsible for different layerstackings. One might suppose that the nature of ex-changeable cations might have an effect. To test this hy-pothesis, let us consider birnessite structures in terms of

:-ltl

IL

F-

Fig. 8. SAED patterns and EDS analyses for cobalt nickel asbolane particles (sample asb). For the particles having relatively

high content of Ni, SAED patterns consist of two incommensurate lk reflection rings. The smaller the Ni content, the lower the

iniensity of the inner ftk reflections. The SAED pattern for particles that contain only Mn and Co displays only one /rk reflection

ring. In all cases, 1(10), = 1(1 I ),, where I (hk) is the intensity of the /rk reflection, and t : I, II is for one of the two incommensurate

ftk reflection rings.

a - C o K -. . . M n K - e d g e

I

\\\

, t

- Asbolane Ural' . . A s b o l a n e N - C- - S i m u l a t i o n

| 152 MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

cations would decrease, and electrostatic interactions be-tween the remaining cations would increase. This pre-dicted change of superperiodicity has actually been ob-served by Post and Veblen (1990).

The question is now the following: are there vacantsites in the layers of sodium birnessite? If there are novacant sites, the two-dimensional superperiodicity woulddepend only on the degree ofnegative charge in the layerand on the valency ofthe exchangeable cations. Iflayersdo contain vacant octahedral sites, divalent cations wouldbe located above and below those vacancies irrespectiveof the layer charge. In the latter situation, superperiod-icity displayed in SAED would reflect the actual distri-bution of vacant sites within layers. According to postand Veblen (1990), exchange of Na for Mg promotes thereorganization of layer stacking and the arrangement ofMg ions and HrO molecules such that Mg ions nearlylocate above and below layer octahedral sites. Such anarrangement of interlayer Mg is possible only if the un-derlying octahedral sites are empty. A second argumentfavoring the existence of layer octahedral vacancies comesfrom a description of the mechanism of Cr(III) oxidationto Cr(VI) by sodium birnessite. In Manceau and Charlet(1992) it is shown that aqueous free Cr3* ions diffuse toand fill in layer octahedral vacancies where the Cr3+ toMna+ electron transfer takes place.

Vernadite

Synthetic vernadite (0-MnOr). Vernadite has a rela-tively large number of corner-sharing MnOu octahedra,as is shown by the shape of its RDF and by the XANESdata (Part I). It is not yet clear whether the vernaditestructure is closer to that of tectomanganate or phyllo-manganate of the 2-D' type. In the first hypothesis, ver-nadite would be built of multiple octahedral chains joinedalong their lengths by sharing vertices. In the second, itwould have a layered structure with a few Mn atoms ran-domly removed from the layer plane to lie above or be-low vacant sites, as in the chalcophanite structure. In ouropinion, vernadite should be regarded as a solid solutionextending from phyllomanganate to tectomanganate. It ismore conveniently pictured as a 3-D (O,OH) framework,where cubic and hexagonal close-packing arrangementsalternate at random and where octahedral sites are ran-domly filled, but where two adjacent Mn(O,OH)u octa-hedra cannot share faces. Such a model can be viewed asa mosaic of single and multiple octahedral chains of vari-able length ranging from I to r octahedra, and the chainsare joined at corners. Because an equal average numberofcations are located between every two adjacent anioniclayers, neither XRD nor SAED patterns of 6-MnO, showbasal reflections near 7, 3.5, or 2.35 A. Thus, 6-MnO,cannot be regarded as a c-disordered birnessite as pro-posed by Giovanoli (1980a). Birnessite has either a 2-Dor 2-D' structure, depending on the existence of octahe-dra that share corners, whereas d-MnOr, like todorokite,has a 3-D anionic framework structure. As a result ofbirnessite's layered structure, its diffraction pattern should

close-packed models. For natural one-layer hexagonalbirnessite, the structure can be written symbolically as

A ,AbCb'C,b'AbC (Chukhrov et al., 1985b)

local charge compensation because the charge of Mna+ions is balanced by two divalent cations. In addition HrOmolecules are arranged in the interlayer in such a waythat, first, they form Mg octahedra with O located on oneside of the interlayer (CA, and C,A stacking) and, second,they form empty prisms with O on the opposite side (A,Aor CC' stacking). The HrO molecules and O atoms thatform empty prisms in interlayers can then be connectedby H bonds. Thus, the arrangement of divalent cationsand HrO in natural birnessite compensates for local chargeand enhances the cohesion oflayers.

In synthetic birnessite, Na alone cannot provide localcharge compensation. Therefore, the presence of Na inbirnessite is expected to change the interlayer structurein such a way that favorable conditions are created fordelocalized charge compensation and stabilization of lay_ers. We will show that the sodium birnessite structurefulfills this expectation. By using the refinement achievedby the Rietveld method (post and Veblen, 1990), the ide_alized one-layer monoclinic structure can be transformedinto a three-layer rhombohedral close-packed model, re_sulting in the following sequence: AbCC,CaBB,BcAA,AbC. . . where A, B, C represent sites for layer O atoms; A,,B', C' are sites for HrO molecules and Na cations of theinterlayers; and a, b, c are layer octahedral sites. We seefrom the symbolic notation that HrO molecules formempty prisms with O atoms from two adjacent layers(AA'A, BB'B, and CC'C stackings). In the absence of localcharge compensation, the arrangement of the interlayerspecies favors the formation of the H bonding betweenHrO molecules and O atoms. H bonding partly explainsthe difference between the observed structure and the ide_alized interlayer structure described above. It may alsoexplain the distortion observed in the hexagonal sym_metry of the layers (the value of a/;/3 in the monoclinicstructure of sodium birnessite is much different than thevalue of the D parameter). Thus, H bonding is thought tobe responsible for decreasing the space-group symmetryfrom 3R to C2/m or lower.

Our model suggests that the two-dimensional super_periodicity observed in SAED patterns from sodium bir_nessite in the past (Giovanoli et al., 1970a; Chukhrov etal., I 989; Post and Veblen, I 990) and in the present studyresults from a periodic distribution of Na atoms in theinterlayer region, which is thought to result form electro_static interaction. Under these conditions, the replace_ment of divalent cations by monovalent Na would leadto a new superperiodicity; i.e., the number of interlaver

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II I 1 5 3

contain basal reflections, even in the extreme case wherethe coherent scattering domains contain only a few layers,i.e., it is c disordered (Reynolds, 1980; Drits and Tchou-bar, 1990). The XRD curve of an extremely disorderedbirnessite sample would then contain only two-dimen-sional hk bands and basal reflections whose positionswould depend on the mean thickness ofthe coherent scat-tering domain (Drits and Tchoubar, 1990).

Iron vernadite. Simulation of the XRD pattern of hy-drous oxides led Chukhrov et al. (1988) and Drits et al.(in preparation, a) to conclude that the absence of ltklreflections ( + 0) is probably associated with (l) the mix-ing ofcubic and hexagonal anionic packings, (2) the smallsize ofcoherent scattering domains, and (3) channels hav-ing different size and distribution patterns. The absenceof basal reflections results from similar or identical cat-ions within each pair ofanionic layers. The observationby SAED of feroxyhite-like patterns simply indicates thathexagonal anionic packing prevails in Fe-rich particles.Data from XAS show that Fe and Mn atoms have differ-ent local structures and thus are segregated in domains.At the nanometer scale and on an average basis through-out the bulk, Fe octahedra belonging to Fe domains havethe same interpolyhedral connections as in feroxyhite,i.e., face, edge, and corner sharing. Instead, Mn domainspossess a 2-D' local structure. SAED and EDS resultsindicate that Mn- and Fe-containing domains are notspace-separated phases, but they coexist within a uniquecoherent scattering domain.

Burns and Burns ( I 975, 197 7) and Ostwald ( I 984) sug-gested that natural vernadite possesses a hybrid structurein which Mn and Fe atoms are segregated within irregu-larly intergrown monoatomic clusters. Miissbauer spec-troscopy (Murad and Schwertmann, 1988) and EXAFS(Chukhrov et al., 1988; Manceau and Combes, 1988)demonstrated Fe and Mn segregation, but the intergrowthmechanism remains to be proved. EXAFS spectroscopyis unable to demonstrate whether epitaxial intergrowthoccurs with either corner- or edge-sharing linkage or both,since it cannot distinguish Fe from Mn neighbors. How-ever, chemical identity can be indirectly determinedthrough the short-range order sensitivity of the preedgestructure (cf. Part I). The low intensity ofthe preedges offemrginous vernadite suggests a layerwise structure forMn domains (Fig. 3c in Part I). In addition, third peaksare visible on MnKRDF (Fig. 3a), indicating corner shar-ing within Mn domains. These corner linkages are logi-cally assigned to MnO'-O-Fe(O,OH)5 oxygen bridges.

Mn-bearing goethite

Comparison of the Mn and Fe RDF for goethite dem-onstrates that Mn and Fe have different local structures;XANES (Part I) and EXAFS data show that Mn atomsare tetravalent. Thus the diference in oxidation state be-tween Mn and Fe is the main deterrent to their randommixing among available octahedral sites. On the otherhand, X-ray and electron diffraction data indicate thatthis mineral's structure is similar to that of homogeneous

goethite with a unique three-dimensional periodicity. Withthese limitations, determination of the structural locationand distribution pattern of Mna* ions becomes a partic-

ularly interesting problem.The mismatch in cell parameters between the isostruc-

tural minerals ramsdellite and goethite is less than l0o/o(Burns and Burns, 1975, 1977). One could then reason-ably imagine an intimate intergrowth of these two phases.However, the Mn RDF of the manganese goethite samplemore closely resembles those of todorokite and buseritethan that of ramsdellite (Figs. la, 7b). This means that,unlike Fe, Mn octahedra mostly share edges with neigh-boring octahedra. Thus, we reject the hypothesis ofa closeramsdellite-goethite association at the scale of the unitcell. The difference of the local environments of Mn andFe, the relatively large number of edge linkages, and theweak preedge spectra that is characteristic of a 2-D struc-ture (Fig. 3 in Part I) indicate that Mn atoms are segre-gated within a discrete phyllomanganatelike local struc-ture.

The question remains: do Mn clusters occur as a sep-arate phase or do they have a definite structural relation-ship with the goethite lattice? A physical mixture of irongoethite and manganese phyllomanganate within thesample would have been detected during electron andX-ray diffraction studies. The apparent discrepancy be-tween information provided by EXAFS and the preedge

spectroscopies (Part I) helps to determine how Mn clus-ters relate structurally to the rest of the mineral. EXAFSshows that MnOu octahedra share as many corners as intodorokite (Fig. 7b), and from XANES we know that Mnclusters possess a roughly buseriteJike local structure (Fig.

3 ofPart I). This paradox is solved ifone considers thatthe third MnK RDF peak is in part due to the existenceof MnO,-O-Fe(O,OH)5 oxygen bridges. Although theseresults do not exclude the existence ofedge linkages be-tween Mn and Fe octahedra, one may postulate that somedomains exist where octahedral Mn and Fe chains arejoined only by corners. This local structure would resem-ble that already known in nsutite (7-MnOr), where py-

rolusite and ramsdellite domains are joined by corners.Thus, the structural model for manganese goethite may

be described as follows (Fig. 1 0a). The coherent scatteringdomains are determined by three-dimensional anionicframeworks that have a strong tendency to be hexagonallyclose-packed, resulting in hkl reflections in SAED andXRD patterns. Each coherent scattering domain includesMn and Fe clusters linked through MnOr-O-Fe(O,OH)jbridges. The cationic arrangement of Fe clusters is thesame as for goethite, whereas Mn is distributed layerwisein Mn domains. The absence of basal reflections is ac-counted for by approximately the same number of cat-ions between each O layer pair. The average structure canbe defined in terms of a goethite-like cell because of therelatively regular distribution of filled and vacant octa-hedral sites. However, the vacant sites of the iron goethiteunit cell must be replaced in the manganese goethite unitcell by some Mn, the quantity of which depends on the

I 1 5 4

- ( ) o . o H o M nV

Unit cell of goelhite

r- '@----- 8------[ i--- --@ -'

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

@ F e

Unit cell ol Mn-containing goethite

i -a ---()-- -o--. -a.i

o o @i

O fii i33i Qunlozy

Fig. 10. A structural model for the Mn-bearing goethite. (a)Hexagonal anionic closed packing where domains of goethiteand phyllomanganate are intergrown. (b) Distribution of cationsover available octahedral sites in the unit cell ofgoethite and inthe average unit cell ofmanganese goethite.

ratio between Fe and Mn atoms. Figure lOb shows acomparison of unit cells for pure goethite and a mineralcontaining 67 aIo/o Fe and 33 atolo Mn.

Asbolane

The two incommensurate reciprocal sublattices de-scribed in the result section arise from the regular alter-nation of two layers. Their unit-cell parameters vary be-cause of their different chemical compositions orstructures. The ratio I(l l0)/I(100) has approximately thesame value for reflections belonging to both sublattices,indicating that these layers have an identical structuralmotif. According to calculations made by Chukhrov etal. (1980), this intensity ratio is typical ofoctahedral lay-ers.

We can combine data from all of the techniques(XANES, EXAFS, SAED, and EDS) to determine rhestructure of asbolane. From XANES, we learned that Coatoms are trivalent and sixfold coordinated, which agreeswith a co-(o,oH) distance of l.g2 A (Manceau et al.,1987). This distance suggests a low-spin configuration, ashigh-spin t6lco3+ ions would result in a distance close to1.99 A (Chenavas et al., 197l). EXAFS data indicate thatMn and Co atoms possess an identical 2-D local struc-ture, and it is impossible to use this method to determinewhether Co substitutes for Mn in MnO, layers or formsseparate CoOOH layers. Nevertheless, in the SAED pat-tern of Ni-free particles frorn the cobalt nickel asbolanewe see 00/ reflections with d(001) : 9.4 A. This indicatesthat two layers with different scattering powers alternateregularly and leads to the conclusion that one layer typecorresponds to MnO, and the other to CoOOH or

Co(OH)r. The Co and Mn layers have the same size andare commensurate in the plane. Sublattice II correspondsto Ni(OH), layer fragments that are free of both Co andMn, explaining the progressive weakening of SAED re-flections as Ni content decreases.

In cobalt asbolane, the appearance of two hk} rcflec-tion networks in combination with a unique series of 00/reflections with d(001) : 9.4 A corresponds to two reg-ularly alternating incommensurate layers. The distribu-tion of hkl reflection intensities for sublattice II is quitedifferent from that of cobalt nickel asbolane. Reflectionintensity calculations led Chukhrov et al. (1982) to con-clude that this mineral contains alternating tetrahedraland octahedral layers, but the similar scattering powersfor Co and Mn prevent an unambiguous assignment ofcations to particular sites. SAED, XANES, and EXAFSdata make it clear that there are no tetrahedral CoOOHlayers in cobalt asbolane. The a value of sublattice II (a,,: 3.15 A) is consistent with t11Mn2+ layers and cannot beattributed to any known r6lco3+ and t61Mn4+ layer struc-ture. The presence of talMn2+ layers is then supported butnot proved by our results. The relative concentration ofMn2+ in the bulk sample is below our XANES detectionlimit, and a lack of spacial resolution with EXAFS pre-vents observation ofsuch a structure. Therefore, we pro-pose a structure consisting of octahedral MnO, layers al-ternating regularly with structural fragments of Mn(OH)r;the position of Co atoms remains unknown.

Atom segregation

One of the most interesting results of this study is therevelation of domain structure in natural hydrous oxidesthat contain some quantity of "foreign" cations. Do-mains having a different chemical composition or oxi-dation state of cations may have different structural re-lationships. The limiting case is represented by asbolane,where the domains are completely isolated. In cobaltnickel asbolane, MnOr-N(OH), and possibly CoOOHlayers coexist within the same crystal. Each layer is anindependent structural fragment that alternates regularlyin the c direction. Such a mixed-layer structure can bethought of as a polysome (Veblen, l99l), where slabs aresheet oxides of Mn and Ni. A second example is the puremanganese mixedJayer asbolane recently recognized byButuzova et al. (1990). Here, Mn is presumably presentin three incommensurate layers as Mn4+, Mn3+, and M1z+.Likewise, the cobalt asbolane from Germany is thoughtto have alternating layers of MnO, and Mn(OH)r.

The second class of domain structure is represented byiron vernadite and manganese goethite, where domainsofdifferent structure and cation composition are epitax-ially inter- or overgrown within the same anion frame-work. In both minerals, Mn4+ ions have a strong tenden-cy to combine by edge sharing of octahedra, but theirmain structural differences comprise the following: (l) theanion framework of manganese goethite is a hexagonalarray, but in iron vernadite coherent scattering domainspossess both hexagonal and cubic anionic packings; (2)

I I V I @ @ E i a)3L-@------ E-------E----,-@-,i

r l'+>

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

the Fe domains in the two minerals have a different cat-ion distribution over available octahedral sites, resultingin a feroxyhitelike local structure in iron vernadite anda goethite-like local structure in manganese goethite.

Structure of two-line hydrous oxides

Knowledge of the local structure of hydrous oxides isneeded to understand the role of short-range cation or-dering in the formation of long-range ordering. Indeed,in their highest disordered state, pure Mn and pure ironhydrous oxides usually display only two weak, broad ftkbands (or "two lines") on X-ray and electron diffractionpatterns. EXAFS data suggest at least five distinct localstructures that can correspond to such a diffraction pat-

tern; the actual local structure depends on how the gel

formed. Ferric gels that are freshly precipitated after com-plete hydrolysis and oxidation offerrous sulfate or chlo-ride solutions at neutral pH show a lepidocrocitelike^y-FeOOH local structure (Combes et al., 1986). How-ever, gels synthesized from a nitrate or chloride ferric

solution show a goethite-like (a-FeOOH) or akaganeite-like (p-FeOOH) local structure, respectively (Combes et

al., 1989). When a goethite-like gel is heated (to 92 "C)for a few hours, it transforms into a feroxyhite-like, two-line gel, and subsequently into hematite (Combes et al.,1990). This last local structure has been observed in all

of the natural two-line Fe gels studied up to now (the so-called protoferrihydrite; Combes, 1988). It is interestingthat the interpolyhedral linkages that are typical ofwell-crystallized Fe minerals are already present in their pre-

mineral gel. Thus, the structural formation of ferric ox-yhydroxides (c.-, g-, r-FeOOH) is undoubtedly decidedvery early in the polymerization process. On the otherhand, the structure of two-line Mna* hydrous gels (6-

MnOr) seems not to be strictly related to that of the sub-sequent, well-crystallized polymorph. Mn gels most prob-

ably consist of a 3-D framework of randomly distributededge- and corner-sharing MnOr octahedra; that might ex-plain why they seem able to transform into a large variety

of structures.It is evident from this discussion that diffraction effects

from extremely disordered Mna* and Fe3+ hydrous ox-ides have low sensitivity to changes in local structures.The similarity of XRD patterns for all two-line com-pounds results from the fact that their coherent scatteringdomains have extremely small values (a few tens of

thousands of cubic ingstrdms) and usually are composedof mixed cubic and hexagonal anionic packings, where

each pair ofanionic layers contains, on average, the samenumber of cations. Structural interpretations of these fea-

tureless diffraction patterns are hardly feasible and at leastnot always unique. The idea of adding about 330/o tetra-hedrally coordinated Mn and Fe to fit XRD profiles(Chukhrov et al., 1988; Eggleton and Fitzpatrick, 1988,1990) is then to be treated with caution; the concept is

not supported by Mtissbauer spectroscopy (Cardile, 1988)

and contradicts XANES and EXAFS results at both the

Mn and FeK edges (Manceau et al., 1990b; Waychunas

et al., 1993 Part I). New structural models that satisfy

both XRD and XAS results have been recently proposed

for Fe hydrous oxides (Manceau and Drits, in prepara-

tion; Drits et al., in preparation' a, b) but remain to be

established for Mn comPounds.

CoNcr,usroNs

l. Natural magnesium birnessite has a oneJayer hex-

agonal unit cell; synthetic sodium birnessite has a mono-

clinic subcell with structural similarity to chalcophanite

but lacking corner-sharing Mna+ -1\414 + octahedra. Struc-

ture of the interlayer and layer stacking are controlled by

the nature of the interlayer cation. The structures of mag-

nesium and sodium birnessite can be described using an-

ionic close-packed layer models-2. Synthetic vernadite (6-MnOr) is not c-disordered

birnessite. Its structure probably consists of a mosaic of

single and multiple octahedral chains having variable

length and width. The chains are joined by corners in a

3-D anionic framework. That there are the same mean

number of cations in each two adjacent anionic layers

explains the absence of00/reflections. The absence oflzkl

reflections results from structural disorder in the distri-

bution of chain sizes, the small coherent scattering do-

mains, and the mixture of anionic cubic and hexagonalpacking.

3. The X-ray-amorphous hydrous iron manganese ox-

ide phase called iron vernadite that occurs on the ocean

floor has a hybrid structure with coexisting Mna* and

Fe3* domains. In one domain, Mn atoms are distributed

layerwise, as in phyllomanganate, whereas in the other,

Fe octahedra are linked as in feroxyhite.4. Natural manganese goethite, previously described

as goethite-groutite, consists of Mn4+ and Fe3+ domains

that have cation distributions similar to phyllomanganate

and goethite, respectivelY.5. Cobalt nickel asbolane is formed by layers of MnOr'

N(OH)r, and possible CoOOH or Co(OH)r, alternating

regularly along the c axis. In cobalt asbolane, Mn2+ tet-

rahedral layers are probably regularly mixed with layers

of MnOr. However, additional experimental evidence for

this structure is needed.This study reemphasizes the complex structure of some

low-temperature minerals. Structural variability and im-

perfection occur in the stacking oflayered mineral struc-

tures such as birnessite. Much about their structure re-

mains unknown, and one may foresee that birnessite with

different degrees of periodicity in its two'dimensional layer

stackings as well as with structural imperfections in the

interlayer region will be found in natural environments'

A second class of imperfection concerns the hybrid struc-

ture of low-temperature minerals at a very fine scale' Two

types of structural situations have so far been identified:(li a simple deviation from the random distribution of

atoms in a solid solution series (Manceau, 1990), and(2)

the existence of intimately mixed discrete phases. On one

hand, these phases can entertain no structural relation-

ship (except for the presence of H bonding) such as is

1 1 5 5

I 1 5 6 MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II

Geology Review, 1983, 25,838-847 (translated from Izvestia Akade-mie Nauk, SSSR, Seriya Geologicheskaya, 5, 9l-99).

Chukhrov, F.V., Gorshkov, A.I., Drits, V.A., and Dikov, y.p. (1985a)Structural varieties of todorokite. International Geology Rewiew, 1985,27 , 148l-1491 (translated from Izvestia Akademie Nauk, SSSR, SeriyaGeologicheskaya, 1l , 6l-72).

Chukhrov, F.V., Sakharov, B.A., Gorshkov, A.I., Drits, V.A., and Dikov,Y.P. (1985b) Crystal structure of birnessite from the pacific Ocean.International Geology Review, 1985, 27, 1082-1088 (translated fromIzvestia Akademie Nauk, SSSR, Seriya Geologicheskaya, S, 66-73).

Chukhrov, F.V., Gorshkov, A.I., and Drits, Y.A. (1986) Crysrailochem-ical systematics of Mn-minerals with tunnel structures. Izvestia Aka-demie Nauk, SSSR, Seriya Geologicheskaya, 10, 3-18.

-(1987a) News in crystallochemistry of manganese hydroxides.Zapiski Vsesoyusnogo Mineralogicheskaya Obchestva, I 16, Zl0-222.

Chukhrov, F.V., Sakharov, B.A., Gorshkov, A.I., Drits, V.A., and Bari-nov, N.N. (1987b) New mixed-layer phase from oceanic Fe-Mn nod-ules. Lithology and Row Materials, 5, ll2-12D.

Chukhrov, F.V., Drits, V.A., Gonhkov, A.I., Sakharov, 8.A., and Dikov,Y.P. (l 987c) Structural models of vernadites. Izvestia Akademie Nauk,SSSR, Seriya Geologicheskaya, 12, 198-204.

Chukhrov, F.V., Manceau, A., Sakharov, B.A., Combes, J.-M., Gorsh-kov, A.I., Salyn, A.L., and Drits, V.A. (1988) Crysral chemistry ofoceanic Fe-vernadites. Mineralogicheskii Zhurnal, 10, 7 8-92.

Chukhrov, F.V., Gorshkov, A.I., and Drits, V.A. (1989) Supergenic man-ganese hydrous oxides, 208 p. Nauk, Moscow.

Combes, J.-M. (1988) Evolution de la structure locale des polymdres etgels ferriques lors de la cristallisation des oxydes de fer. Application aupi1geage de I'uranium. Thdse de Doctorat, Universit6 paris 6, paris.

Combes, J.-M., Manceau, A., and Calas, G. (1986) Study of the localstructure in poorly-ordered precursors of iron oxi-hydroxides. Journalde Physique, C8, 697-701.

Combes, J.-M., Manceau, A., Calas, G., and Bottero, J.-y. (1989) For-mation offerric oxides from aqueous solutions: A polyhedral approachby X-ray absorption spectroscopy. I. Hydrolysis and formation offerricgels. Geochimica et Cosmochimica Acta. 53. 583-594.

Combes, J.-M., Manceau, A., and Calas, G. (1990) Formation of ferricoxides from aqueous solutions: A polyhedral approach by X-ray ab-sorptron spectroscopy. II. Hematite formation from ferric gels. Geo-chimica et Cosmochimica Acta, 54, 1083-1091.

Delaplane, R.G., Ibers, J., Ferraro, J.R., and Rush, J.J. (1969) Diffractionand spectroscopic studies of the cobaltic acid system HCoO,-DCoO,.Journal of Chemical Physics, 50, 1920-1927.

Deliens, M., and Goethals, H. (1973) Poll'typism of heterogenite. Min-eralogical Magazine, 39, | 52-1 57 .

Drits, V.A. (1987) Electron diffraction and high-resolution electron mi-croscopy of mineral structures. Springer-Verlag, Berlin.

Drits, V.A., and Tchoubar, C. (1990) X-ray diffraction ofdisordered la-mellar structures. Theory and application to microdivided silicates andcarbons. Springer-Verlag, Berlin.

Drits, V.A., Petrova, V.V., and Gorshkov, A.L (1985) Manganese mrn-erals ofFe-Mn nodules from the sediments ofthe central pan ofpacificOcean and their postsedimentation transformation. Lirhology and RawMaterials. 3. 17-39.

Eggleton, R.A., and Fitzpatrick, R.W. (1988) New data and a revisedstructural model for ferrihydrite. Clays and Clay Minerals, 36, I I l-124.

-(1990) New data and a revised structural model for ferrihydrite:Reply. Clays and Clay Minerals, 38, 335-336.

Ewing, F.J. (1935) The crystal strucrure of lepidocrocite, T-FeOOH. Jour-nal ofChemical Physics, 3, 420-427.

Giovanoli, R. (1980a) Vemadite is random-stacked birnessite. Mineral-ium Deposita, 15, 251-253.

-(1980b) On natural and synthetic manganese nodules. In I.M. Va-rentsov and G. Grasselly, Eds., Geology and geochemistry of manga-nese, vol. l, p. 159-202. Schweizerbart'sche Verlagsbuchhandlung,Stuttgart.

Giovanoli, R., and Sr?ihli, E. (1970) Oxide und Oxidhydroxide des drei_und vierwertigen Mangans. Chimi424, 49-61.

Giovanoli, R., Stiihli, E., and Feitknecht, W. (1969) Uber Struktur undReaktivitiit von Mangan (IV) Oxiden. Chimia, 23, 264-266.

observed for asbolanes. On the other, they can be epitax_ially intergrown, as is true for iron vernadite, manganesegoethite, and iron diaspore (Hazemann et aI., lg92). Allthese types of crystal defects result from nonequilibriumcrystallization in the thermodynamic and kinetic condi-tions that prevail at the Earth's surface. For instanca, ep-itaxial overgrowths are thought to result from a hetero-geneous nucleation in conditions of undersaturation(Charlet and Mancean, 1992). Deciphering the actualstructure of these and other complex minerals is an im_portant first step toward an advance understanding oftheir conditions and mechanisms of formation.

AcxNowr,nocMENTs

The authors thank R. Giovanoli for providing the sodium birnessitesample and the staffof LURE for the synchrotron facility. A critical read_ing of a preliminary version by G.A. Waychunas is acknowledged. Theauthors express theL gratitude to Susan Stipp, who helped improve theEnglish. This research was supported by CNRS/INSU through the DBTprogr:rm, grant 90 DBT 1.03 (contribution no. 414).

RnrnnnNcrs crrEDBlake, R.L., Hessevick, R.E., Zoltai, T., and Finger, L.W. (1966) Refine-

ment of the hematite structure. American Mineralogist, 51, ll3_12g.Bums, R.G., and Bums, V.M. ( I975) Mechanism for nucleation and growth

ofmanganese nodules. Nature, 255, 130-131.-(1977) Mineralogy of manganese nodules. In G.p. Glasby, Ed.,

Marine manganese deposits, p. lg5-24g. Elsevier, New york.Butuzova, G.Yu., Drits, V.A., Morozov, A.A., and Gorshkov, A.I. (1990)

Processes of formation ofiron manganese oxyhydroxides in the Atlan_tis II and Thetis Deeps of the Red Sea. Special publication of theInternational Association of Sedimentologists, I l, 57_72.

Bystrdm, A.M. (19a9) The crystal structure of ramsdellite, an orthorhom-bic modification of MnOr. Acta Chemica Scandinavica, 3, 163_173.

Cardile, C.M. (1988) Tetrahedral Fer+ in ferrihydrite: 57Fe Mtissbauerspectroscopic evidence. Clays and Clay Minerals, 36, 537_539.

Charlet, L., and Manceau, A. (1992) X-ray absorption spectroscopic studyofthe sorption ofCr(II! at the oxide/water interface. II. Adsorption,coprecipitation and surface precipitation on ferric hydrous oxides.Journal of Colloid and Interface Science, I 4g. 425-442.

Chenavas, J., Joubert, J.C., and Marezio, M. (1971) Low-spin - high_spin transition in high pressure cobalt sesquioxide. Solid State Com-munications, 9, 1057-1060.

Chukhrov, F.V., Gorshkov, A.I., Sivtsov, A.V., and Berezovskaya, V.V.(1978a) Structural varieties of todorokite. Izvestia Akademie Nauk,SSSR, Seriya Geotogicheskaya, 12, 86-95.

Chukhrov, F.V., Gorshkov, A.I., and Rudnitskaya, E.S. (l97gb) On ver_nadite. Izvesria Akademie Nauk, SSSR, Seriya Geologicheskaya, 6, 5_19.

Chukhrov, F.V., Gorshkov, A.I., Sivtsov, A.V., and Berezovskaya, V.V.(1979) New data on narural todorokires. Nature,27g, 631_632.

Chukhrov, F.V., Gorshkov, A.I., Vitovskaya, I.V., Drirs, V.A., Sivtsov,A.V., and Rudnitskaya, Ye.S. (1980) Crystallochemical nature of Co_Ni asbolane. Intemational Geology Review, 24,5,599-604 (tanslatedfrom Izvestia Akademie Nauk, SSSR, Seriya Geologicheskaya,6, T3_8 l ) .

Chukhrov, F.V., Gorshkov, A.I., and Sivtsov, A.V. (l9gl) A new srruc-tural variety of todorokite. Izvestia Akademie Nauk, SSSR, SeriyaGeologicheskaya, 5, 88-9 l.

Chukhrov, F.V., Gorshkov, A.I., Drits, V.A., and Sivzev, A.V. (19g2)New structural varieties ofasbolane. Izvestia Akademie Nauk. SSSR.Seriya Geologich eskay a, 6, 69 -7 7 .

Chukhrov, F.V., Gorshkov, A.I., Drits, V.A., Shterenberg, L.ye., Sivtsov,A.V., and Sakharov, B.A. (1983) MixedJayer asbolane_buserite min_erals and asbolanes in oceanic iron-manganese nodules. Internatronal

-(1970a) Uber Oxyhydroxide des vierwertigen Mangans mitSchichtengitter. l. Natrium Mangan (II,III)-Manganat (IV). HelveticaChimica Acta, 53, 209-220.

-(1970b) Uber Oxidhydroxide des vierwertigen Mangans mitSchichtengitter. 2. Mangan (Ill)-Manganat (IV). Helvetica Chimica Acta,53,453-464.

Hazemann, J.L., Manceau, A., Sainctavit, P., and Malgrange, C. (1992)Structure of the aFe,Al,-"OOH solid solution. Evidence by polarizedEXAFS for an epitaxial gowth of hematiteJike clusters in diaspore.Physics and Chemisrry of Minerals, 19, 25-38.

Kinniburgh, D.G., and Jackson, M.L. (1981) Cation adsorption by hy-drous metal oxides and clay. In M.A. Anderson and A.J. Rubin, Eds.,Adsorption of inorganics at solid-liquid interfaces, p. 9 I - I 60. Ann Ar-bor Science, Ann Arbor, Michigan.

Manceau, A. (1939) Synthetic l0-A and 7-A phyllomanganates: Theirstructurss as determined by EXAFS-Discussion. American Mineral-ogist ,74, 1386-1389.

-(1990) Distribution of cations among the octahedra of phyllosili-cates: Insight from EXAFS. Canadian Mineralogist, 28,321-328

Manceau, A., and Charlet, L. (1992) X-ray absorption spectroscopic studyof the sorption of Cr(III) at the oxide/water interfac€. I. Molecularmechanism of Cr(IID oxidation on Mn oxides. Journal of Colloid andInterface Science. 148. 443-458.

Manceau, A., and Combes, J.-M. (1988) Structure of Mn and Fe oxidesand oxyhydroxides: A topological approach by EXAFS. Physics andChemistry of Minerals, I 5 (3), 283-295.

Manceau, A., Llorca, S., and Calas, G. (1987) Crystal chemistry of cobaltand nickel in lithiophorite and asbolane from New Caledonia. Geo-chimica et Cosmochimica Acta. 51. 105-l13.

Manceau, A., Combes, J.-M., and Calas, G. (1989) Chernical and struc-tural applications of X-ray absorption spectroscopy in mineralogy.Journal de Chirnie Physique, 86, 1533-1546.

Manceau, A., Buseck, P.R., Miser, D., Rask, J., and Nahon, D. (1990a)Copper site occupancy in lithiophorite from a Mn banded ore. Amer-ican Mineralogist, 7 5, 490-494.

Manceau, A., Cornbes, J.-M., and Calas G. (1990b) New data and a re-vised model for ferrihydrite: A comment on a paper by R.A. Eggletonand R.W. Fitzpatrick. Clays and Clay Minerals, 38, 331-334.

Manceau, A., Gorshkov, A.L, Drits, v.A. (1992) Structural chemistry ofMn, Fe, Co, and Ni in manganese hydrous oxides: Pan I. Informationfrorn XANES spectroscopy. American Mineralogist, 77, 1133-1143.

Miura, H. (1986) The crystal structure of hollandite. Mineralogical Jow-nal . 13. 119-129.

Murad, E., and Schwertmann, U. (1988) Iron oxide mineralogy of somedeep-sea ferromanganese crusts. American Mineralogist, 7 l, 1395-1400.

Ostwald, J. (1984) Femrginous vernadite in an Indian Ocean ferroman-ganese nodule. Geological Magazine, l 2 l, 48 3-48 8.

Patrat, G., De Bergevin, F., Perret, M., and Joubert, J.C. (1983) Structurelocale de 6 FeOOH. Acta Crystallographica, B39, 165-170.

Post, J.8., and Appleman, D.E. (1988) Chalcophanite, ZnMn,O, 3HrO:

tt57

New crystal-structure determination. American Mineralogist, 7 3, 14OI-1404.

Post, J.E., and Bish, D.L. (1988) Rietveld refinement of the todorokitestructure. American Mineralogist, 73, 86 l-869.

Post, J.E., and Veblen, D.R. (1990) Crystal structure determinations of

synthetic sodium, magnesium, and potassium birnessite using TEMand the Rietveld method. American Mineralogist, 75,477-489.

Post, J.8., Dreele, R.B., and Buseck, P.R. (1982) Symmetry and cationdisplacements in hollandites: Structure refrnements of hollandite, cryp-tomelane and priderite. Acra Crysrallographica, B38, 1056-1065.

Potter, R.M., and Rossman, G.R. (1979) The tetravalent manganese ox-ides: Identification, hydration, and structural relationships by infraredspectroscopy. American Mineralogist, 64, | 199-1218,

Reynolds, R.C. (1980) Interstratified clay minerals. In G.W. Brindley andG. Brown, Eds., Crystal structures of clay minerals and their X-rayidentification, 495 p. Mineralogical Society, London.

Sainctavit, P., Petiau, J., Manceau, A., Rivallant, R., Belakhovsky, M.,and Renaud, G. (1988) A two-mirror device for higher harmonic re-jection. Nuclear Instruments Methods, 27 3, 423-42E.

Stoufi P., and Bouldgue, J. (1988) Synthetic 10 A and 7 A phytloman-ganates: Their structures as determined by EXAFS. American Miner-afogist, 73, 1162-1169.

Strobel, P., Charenton, J.C., and L.englet, M- (1987) Structural chemistryof phyllomanganates: Experimental evidence and structural models.Revue de Chimie Min6rale, 24, 199-220.

Szytula, A., Burewicz, A., Dmitrijevic,2., Krasnicki, S., Rzany, H', To-

dorovic, J., Wanic, A., and Wolski, W. (1968) Neutron diffraction stud-ies ofc-FeOOH. Physica Status Solidi, 26, 429-434.

Szltula, A., Balanda, M., and Dimitrijevic, Z. (1970) Neutron diffractionstudies ofd-FeOOH. Physica Status Solidi, 3, 1033-1037.

Teo, B.K. (1986) EXAFS: Basic principles and data analysis, 349 p-

Springer-Verlag, Berlin.Towe, K.M., and Bradley, W.F. (1967) Mineralogical constitution of col-

loidal "hydrous ferric oxides." Journal ofColloid and Interface Science,24,384-392.

Turner, S., and Buseck, P.R. (1981) Todorokites: A new family ofnatu-rally occurring manganese oxides. Science, 212, lO24-1O27.

Veblen, D.R. (1991) Potysomatism and polysomatic series: A review andapplications. American Mineralogist, 76, 80 I -826.

Vicat, J., Franchon, 8., Strobel, P., and Tran Qui, D. (1986) The structureof K,,.Mn.O,u and cation ordering in hollandite-t1pe structures. Acta

Crystaf lographica, B'42, | 62- | 67 .Wadsley, A.D. (1953) The crystal s t ructure of psi lomelane,

(Ba,H,O),MnrO,o. Acta Crystallographica, 6, 433-438.Waychunas, G.A., Rea, B.A., Fuller, C.C., and Davis, J.A. (1993) Surface

chemistry offerrihydrite. I. EXAFS studies ofthe geometry ofcopre-cipitated and adsorbed arsenate. Geochimica et Cosmochimica Acta,m press.

Mnxuscmrr REcEIVED Aucusr 1 5, 1990MaNuscnrvr AcCEPTED JuLv 6,1992

MANCEAU ET AL.: MANGANESE HYDROUS OXIDES II