American Mineralogist, Volume 68, pages 972-980,I9E3
A review of the todorokitebuserite problem: implications to the
mineralogy of marine manganesenodules
Rocen G. Bunns, VrRcrNre Mee BunNs AND Henr-eN W. SrocrN,rnNrl
Department of Earth and Planetary Sciences
Massachusetts Institute of Technology
Cambridge Massachusetts 02 139
Abstract
The names todonokite and buserite are two competing terminologies for the abundant
manganese
oxide mineraloccurringin deep-seamanganese
noduledeposits.Proponentsof
the buserite phase claim it to be the parent of todorokite which is consideredto be a
mixture of buseriteand its breakdownproductsbirnessiteand manganite(rMnOOH). The
recent experimental evidencedemonstratingthe integrity of todorokite (a tekto-manganate
with multidimensionaltunnels formed by walls of edge-shared[MnOe] octahedra)has led to
an assessmentof the proposed relationships between synthetic buserite (a phyllomanganate containing layers of edge-shared[MnOo] octahedra) and its transformation
products.Inconsistenciesare found to exist in publishedelectrondiffractiondata underlying the proposed topotactic transformation mechanismof synthetic birnessite platelets to
acicular "7MnOOH" crystallitessuggestedto constitutenaturaltodorokites.The crystal
chemistry of todorokite structure'types is discussed.We suggestthat three types of atomic
substitutionoccur in the tunnel structures:first, substitutionof Mn2* by other divalent
cations (e.g., Mg2*, Niz+, Cu2*, Zn2*1in the "walls" formed by chains of edge-shared
[MnOe] octahedra,characteristicallythree octahedrawide; second,replacementof Mna*
by similar sizedcations(e.g., low.spin Co3+)in the "ceilings" or "floors" typically three,
but as many as sevenor more, [MnOe] octahedrawide; and third, in the tunnel interiors
adjacentto Mna+ vacanciesin the "ceilings." Here, a variety of large cations(K+, Ba2+,
Ag*, Pb2*, Ca2+,Na*;, H2O molecules,and hydrated transition metal cations could be
accommodated.We recommend that the name todorokite be universally adopted for the
predominant manganeseoxide mineral accommodatingdivalent cations of nickel and
copper in marine manganiferousconcretions.
of a synthetic sodium manganeseoxide hydrate (Feitknecht and Marti, 1945;Wadsley, 1950a,b) called "l0A
manganite." As a result, the phaseoccurring in manganesenoduleswas identified as either todorokite (Straczek
et al., 19ffi; Hewitt et al., 1963),or synthetic "l0A
manganite" (Buser, 1959).This dual nomenclaturehas
pervaded the literature on marine manganesedeposits
(Burns and Burns, 1977a, 1979b),d-espitethe complaint
(Arrhenius, 1963)that the term "l0A manganite"causes
confusion with the mineral manganite(rMnIllOOH).
Giovanoli et al. (1971)proposedthat the nodule l0A
phase
be called buserite, in honor of W. Buser, and
l9ro).
X-ray diffraction patterns similar to that of the nodule suggestedthat buserite has the same crystal structure as
l0A phaseare shown by two other species:todorokite, a synthetic"l0A manganite." Buserite was acceptedas a
Mn-Mg-Ca-Na-K oxide originally reported in a non- mineral name by the Commission on New Minerals by a
maririeenvironment(Yoshimura, 1934);and derivatives small majority (M. Fleischer, personal communication,
1974),and although it is listed in Hey and Embrey (1974),
the name is absentfrom Fleischer's(1980)"Glossary of
I Present address: Sandia National Laborataries, AlbuquerMineral Species." Giovanoli and coworkers(1971,1975)
que, New Mexico 87185.
further contended from X-ray diffraction and electron
0003-(M)vE3/09 l 0-0972$02.00
972
Introduction
Ono of the most common manganeseoxide phases
detected in marine ferromanganesenodules, crusts and
rnetalliferous sedimentsis that with characteristic X-rav
difraction linesat 9.5-9.8A and 4.8-4.9A. Chemicalani
electroR microprobe analyses of this nodule l0A pfiase
phase indicate that it is a hydrated Mn-Mg-Ca-Na-NiCu oxide. Its ability to concentrate Ni and Cu to several
weight percent (Burns and Burns, 1978, 1979b)makes
deep-seamanganesenodules the focus of considerable
scientificand economic interest (Glasby, 1977;Crortan,
BURNS ET AL.: REVIEW OF THE TODOROKITE-BUSERITE PROBLEM
microscopy measurementsof synthetic manganeseoxides that natural todorokites are a mixture of buserite and
its decompositionproducts birnessiteand manganite(r
MnOOH). The Swiss group also recommendedthat todorokite should be discreditedas a valid mineral (Giovanoli and Brutsch, 1979a,b; Giovanoli, 1980).However,
recent experimental evidence verifying the integrity of
todorokite includes results derived from infrared spectroscopy(Potterand Rossman,1979),electrondiffraction
(Chukhrov et al., 1978, 1979, l98l; Siegel, l98l), hieh
resolution transmission electron microscopy (snreu)
(Turner and Buseck, 1979,1981,Turner et al., 1982),and
extended X-ray absorption fine structure (exnrs) measurements(Crane, 1981)of manganese(IV) oxide minerals.
The conflicting viewpoints about todorokite have led us
to critically evaluate experimental results and literature
on buserite and todorokite. We summarizehere evidence
demonstrating that buserite and todorokite are distinct
phases, describe recent observationsof todorokite in
marine manganesenodules,and discusspossiblecrystal
structuresand site occupanciesof cations in the todorokite polymorphs. Finally, we recommendthat todorokite
takes precedenceover buseritefor the predominantl0A
phasein manganesenodules.
973
with the Cuban todorokite, either forming simultaneously
with todorokite, or resulting from its alteration. The latter
observation was later cited as evidence by Giovanoli and
coworkers (1971, 1975)for rejecting todorokite as a valid
mineral.
Chemical analysesof todorokites in continental deposits (Straczeket al.,1960; Frondel et aI., 1960;Nambu el
al., 1964)show that manganeseis present in two oxidation states, and that MnII/MnIv ratios fall in the range
0.15-0.23. Significantamounts of Mg are also present,
suggestingthat relatively small divalent Mg2* and Mn2*
ions are essentialconstituentsoftodorokite. The analyses
show that Ca2*, Na* and to lesser extents K+, Ba2*,
Ag* lRadtke et aL, 1967)and Zn2* (Larson, 1962), are
also common constituents. Several chemical formulae
have been proposedfor todorokite, including(Ca,Na,K,
Ba,Ae) (Mg,Mnz*,Zn) Mnl* O12' 3H2O(Frondel er a/.,
1960).
Most todorokites in hand specimenconsist of fibrous
aggregatesof small acicular crystals, although platey
morphologieshave been observed. Electron micrographs
(Straczeket al., 1960;Hariya, 1961;Finkelmann et al.,
1974,Chukhrov et al.,1978; Burns and Burns, 1979a)
reveal that the crystals consist of narrow lathes or blades
elongated along one axis (parallel to b) and frequently
show two perfect cleavagesparallel to (001) and (100).
A structural model for todorokite was inferred from its
Survey of the crystal chemistry of todorokite and
crystal morphology, cleavage properties and electron
buserite
diffraction data (Burns and Burns, 1977b).It was noted
that minerals of the hollandite-cryptomelaneand psilomTodorokite
elane (romandchite)groups also have fibrous or acicular
Todorokite occurs in a variety of terrestrial deposits. habits and two perfect cleavagesparallel to the fibre axis,
At the type locality in Japan, it is formed as an alteration as do a variety of synthetic Tilv oxides. Since these
product of inesite (Yoshimura, 1934).The economically phasespossesstunnel structuresconsistingofdouble and
important todorokite deposits of Cuba, such as Charco treble chains of edge-shared
[MnOo] octahedra,2Burns
Redondo in Oriente Province, were formed syngenetical- and Burns (1977b) proposed that todorokite might also
ly during the Eocene in foraminiferal oozes on the sea- have a tunnel structure basedon chains of multiple width
floor near fumerolic hot springs accompanying volcanic edge-shared
[MnOo] octahedraextending along its b axis.
activity. Subsequentto lithification, some todorokite in The correlation is borne out when comparisonsare made
the weathering zone has altered slightly to manganite, between the unit cell parametersof todorokite, psilomewith more intenseweathering yielding pyrolusite (Simons Iane, and hollandite-groupminerals. Furthermore, todorand Straczek,1958;J. A. Straczek,personalcommunica- okile and psilomelaneboth contain essentialMn2* ions,
tion, 1978).In other terrestrial localities, todorokite is and i! psilomelane these divalent cations are located in
reportedto be of secondaryorigin (Frondel et al.,1960; specific positions in the triple chains of edge-shared
Larson, 1962).X-ray difraction data show that consider- octahedra.Burns and Burns suggestedthat Mn2*, Mg2*,
able variations exist between relative intensities of com- Ni2+, etc. might also be located in analogous sites in
parable lines for different todorokite samples. Faulring todorokite. The larger Ca2*, Na*, K+, Ba2+, erc. ions
(1962)attributed these large intensity variations, as well
and H2O molecules in todorokite were envisaged to
as the diffuseness of certain reflections, to preferred occupy large tunnels in a psilomelanelike tektomanganorientation of the fibrous crystallites of todorokite studied ate structure (Burns and Burns, 1979a).The site occupanby her from Charco Redondo, Cuba. Thus, the most cies of cations in todorokite are discussedlater.
intenselines at 9.65A and 4.82A observedwhen X-rays
Recent electron diffraction and high resolution transare difracted parallel to the fiber axis were weak or
undetected for X-rays diffracted perpendicular to the
2 HRTEM measurementsshow
crystallite axis. The most intense lines in the latter
[2x2] tunnels in hollanditeorientationoccur at 2.424 and 1.424. Faulring( 1962)also cryptomelane and [2x3] tunnels in psilomelane (romandchite)
suggestedthat manganite was topotactically intergrown (Turner and Buseck, 1979).
974
BURNS ET AL.: REVIEW OF THE TODOROKITE-BUSERITE PROBLEM
al., 1970a).Syntheticbuseriteconsistsof elongatedplatelets which give an electron diffraction pattern with pseudohexagonalsymmetry and weak superstructure reflections (d : 7.38A) with a pgriodicity of three between
major reflections(d : 2.464) in the basal (001) planes
(Giovanoli, 1980).X-ray difraction measurements(Giovanoli et al., 1975; Giovanoli, 1980) reveal that the
diagnostic intense lines representing basal plane separationsoccurin sodiumbuseriteat l0.l-10.2 and 5.0-5.1A;
these distances are significantly larger than those measured in samplesof marine manganeseoxide depositsand
in todorokite. A variety of buserite derivatives can be
synthesizedby cation exchange reactions (Wadsley,
1950a,b; Giovanoliet al.,1975;GiovanoliandArrhenius,
1983),and the Mg2+, Cu2*, Ni2*, erc. derivativeshave
smaller basal plan spacingsthan the sodium-bearingparent buserite(Giovanoli, 1980).The transitionmetal derivatives appear to have greater thermal stabilities and to be
more resistantto dehydration than the parent Na buserite
(Giovanoliet al. 1975;Giovanoli and Arrhenius, 1983).
Sodium buseritereadily decomposeswhen exposedto
air. Drying over PaOlsin vacuo leadsto the formation of
sodiumbirnessite,NtuMnr+Ozz' 9H2O, which also has a
platey morphology (Giovanoli et al., 1970a;Giovanoli,
1980).The strongest basal plane X-ray difraction lines
then occur at 7.1 and 3.55A. The electron diffraction
patterns of the platelets again show pseudohexagonal
symmetry and superstructurewith only triple periodicity
(Giovanoli et al., 1970a; Giovanoli, 1980). The latter
superstructureforms the basis for postulating a vacancyordered layered structure for birnessite (Giovanoli et al.,
1970a)based on the chalcophanitestructure (Wadsley,
1955). The structure of chalcophanite, Zn2MnoOra'
6H2O, consistsof layers of edge-shared[MnOe] octahedra and singlesheetsof water moleculesbetweenwhich
the Znz+ ions are located. The octahedrallayer is only 6/7
filled by Mna+; that is, one out of every sevenoctahedral
sites is vacant. The Znz* atoms occur above and below
the vacancies in the octahedral layers, and the stacking
sequencealong the c axis is -O-Mn-O-ZwH2O-Zn-OMn-O-, so that consecutive [MnOe] Iayers are about
7.164 apart. In the structure of sodium birnessiteproposed by Giovanoli et al. (1970a),one out of every six
to
octahedrais vacant,and Mn2* and Mn3* are suggested
lie above and below vacancies in the octahedral layer.
The positionsof the Na* ions are uncertain.Buserite is
consideredto have a similar layered structure, but additional OH- ions and HzO moleculesare presentso that
the v-acancy-orderedlayers of [MnOo] octahedra are9.610.1Aapart (Giovanoli, 1980).Recentintercalationstudies of syntheticburserite (Paterson,1981;Giovanoli and
Buserite
Arrhenius, 1983)correlatewith its proposedlayer strucBuserite,which is assumed(Giovanoli et al.,l97l) to
ture.
be equivalent to synthetic sodium manganese(II, In)
When NaaMn ro,Ozt' 9HzO is refluxed with dilute HNOr
manganate(IV) hydrateor " l0A manganite",is prepared at 40'C, a sodium-free birnessite, Mn7O13' 5H2O, is
by oxidationof fresh Mn(OH)z suspensionsin cold aque- obtained (Giovanoli et al., 1970b)which again has a
ous NaOH solutions by molecular oxygen (Giovanoli er platey habit. The electron diffraction patterns of (001)
mission electron microscopy measurementsof natural
todorokites have confirmed the structural model for todorokite proposed by Burns and Burns (1977b,1979a).
Selected area electron diffractio-n(SAD) patterns of the
Cubantodorokitewith a :9.754 (Chukhrovet al..1978.
1979)revealedthree less intensereflections(d : 9.754)
correspondingto four sub-cellsalongthe a* axis between
the strongreflections(d : 2.444) forming a pseudohexagonal net (Straczek et al., 19ffi1'V. M. Burns, unpublished data). This is in contrast to SAD patterns of
synthetic buserite discussed later which contain only
three sub-cellsalong the ax axis (Giovanoli, 1980;V. M.
Burns, unpublisheddata). Hnrev imagesof the Cuban
todorokite (Turner and Buseck, l98l; Chukhrov et al.,
1978)revealed it to have a tunnel structure, in which
[3x3] tunnels with dimensions9.754 x 9.594 predominate (Turner and Buseck, l98l: Turner et al., 1982).
Chukhrov et al. (1978, 1979, l98l) studied todorokites
from other localities, including a Pacific Ocean manganese nodule, and found SAD patterns having b (2.844)
and c (9.59A) parameters identical to the Cuban todorokite, but largera parameters(14.64, 19Szland 24.384).
Trilling intergrowthsof thesetodorokitesproduceplatey
morphologies,while SAD patternsof the plateletsreveal
nets of hexagonal,triangular, and rhombic cells with four
or five weakerreflections(d : 14.64or 24.38A)between
the strong reflections (d : 2A4h along the a* axes
(Chukhrov et al., 1978;1979;Burns and Burns, 1979a).
Chukhrov et a/. proposed that a family of todorokite
speciesmight exist having a unit cell parameterswhich
are integral multiples of 4.88A. Lattice image photographsof todorokite publishedby Chukhrov et al. (1978)
also show evidenceof structuraldisorder.In addition to
the structure-types suggestedby the variability of the a
parameter, Chukhrov et al. fonnd variations in the periodicity of the (100)planes.Thus, in addition to the main
super-periodicityoftodorokites producinga = 9.75A and
14.6A, defect layers wrtn I.SZA and 12.4A were observed.Theselayersare multiplesof 2.444, the heightof
the base (i.e., dimensionsof a [1ll] axis) of a MnOc
octahedron.
SubsequentHRTEMimagesof todorokitesfrom terrestrial deposits(Turner and Buseck, l98l) and from manganese nodules (Turner et al.,1982; Turner and Buseck,
1982)revealed them to be intergrowths of tunnels of
different widths. Although [3x3] dimensional tunnels
predominatein todorokite,occasionaltunnelswith [3x2],
[3 x 4], [3x 5], [3 x 8] and higherdimensionswere observed
in crystals from both terrestrial and manganesenodule
deposits.
BURNS ET AL.: REVIEW OF THE TODOROKITE-BIISERITE PROBLEM
platelets show hexagonalsymmetry, but streaks in the
sub-cellare interpreted to be indicative of vacancy disordering in the IMnOe] octahedral layers. When
Mn7O13. 5H2O is subjected to careful reduction with
cinnamylalcohol, the plateletsbreak down to extremely
thin needles(Giovanoliet al.,l97l). Theseneedlesfailed
to give an X-ray diffraction pattern, but were believed to
be IMnOOH (manganite).Giovanoli et al. (1971) proposed a mechanismfor the topotactic transformation of
Mn7O13. 5H2O to "yMnOOH". They believedthat the
(fi)l) layers of linked lMnlvoul octahedrain birnessite
correspondto chainsof [MnrII(OH,O)o]octahedrain the
(010)plane of "7-MnOOH," and that the [100]direction
of birnessitebecamethe needleaxis (c axis) of the ,.y
MnOOH" formed topotactically. The structural correlations suggestedby Giovanoli et al. (1971)are shown in
Figure l, and are discussedlater.
These observationsfor buserite and its dehydration
products formed the basis for Giovanoli and his coworkers (1971;1975:l979a,b; 1980)to proposethat world-wide
todorokite ores actually consist of buseritepartly dehydrated to birnessite and partly reduced to needles of
manganite(Giovanoliand Biirki, 1975).Accordingto this
view, the small amount of buseritein the ores accounts
for the characteristic X-ray diffraction pattern, while a
large amount of fibrous, nearly amorphous manganite
accounts for the morphology. Apparently this "yMnOOH," which could not be detectedby X-ray diffraction (dueto its small crystallitesize),is recognizedon the
basis of electron microscopeobservationsof morphology. Thus, the argument againstthe integrity of todorokite
as a mineral depends on the analogy with the topotactic
decompositionof syntheticMn7O13. 5H2O(birnessite)to
what could be unambiguouslyidentifiedas y-MnOOH on
the basis of morphology and electron diffraction. Apparently, the morphology argument was considered stronger
than the diffraction evidence, as the electron diffraction
patternsobservedfor the hypotheticaltodorokite decay
productsdo not match those of the experimentalbirnessite decay product.
975
birnessite
m a n g a n it e
Fig. l. Schematicstructuraldiagramsrelatingthe synthetic
. 5H2Ophase(right)to its proposed
birnessite
MnTO;3
topotactic
transformationproduct manganiteyMnOOH (left) [from
Giovanoli
et al.,1971,Fig. lll.
the result of water loss alone rather than to a structural
rearrangementof the [MnOe] framework.
The infrared spectra of natural todorokites (Potter and
Rossman,1979)indicatethat this mineralis not analogous
to any syntheticphases,includingbuseriteand its cationexchangedderivatives,or to any decompositionproducts
of buseritesuch as birnessiteand manganite.Potter and
Rossmansuggestedthat the todorokite spectraare consistent with either a layer structure of linked [MnOe]
octahedracontainingvacancies,or a highly polymerized
chain or tunnel structure with quadruple chains. Their
hypothesisconforms with observationsof the multidimensionaltunnelsfound in subsequentstudiesofnatural
terrestrial and marine todorokites (Turner and Buseck,
1981,1982;Turner et al.,1982).
Discussion
The information derived from infrared spectroscopy,
electron diffraction, and Hnreu measurementsappear to
demonstrateconclusively that todorokite and buserite are
distinct phaseshaving different linkagesof edge-shared
[MnOo]octahedra.Nevertheless,Giovanoli and coworkers (1971; 1975; 1979a,b; 1980; 1983) maintain that
buserite is the primary phase and that todorokite is an
Information from infrare d sp ectros copy
assemblage of buserite and its breakdown products
Some of the problems of identification and structural birnessiteand manganite.However, closeexaminationof
correlations for todorokite and buserite have been ad- the evidencepresentedby the Swissgroup revealsappardressedby infrared spectroscopy(Potter and Rossman, ent inconsistenciesin interpretations of their data or
1979). In a study of numerous natural and synthetic questionablerelevance of the results to geochemical
manganeseoxides, Potter and Rossman(1979)demon- processes.
strated that it is possible to distinguish between different
First ofall, there are the inferencesfrom the laboratory
[MnOo] structural linkages by their spectralprofiles in the studiesthat are not correlatablewith sedimentarymineral
mid-infrared region. They have shown, for example, that formation. For example, the decompositionstudies of
the proposed layer structure of birnessite is supportedby
syntheticMn7O13. 5H2O to the presumedmanganite(y
its infrared spectrum, and have confirmed the identity of MnOOH) phase reported by the Swiss group were not
natural birnessite with its synthetic analogues.The infra- performed on buserite, but on a sodium-free birnessite
red spectra also show that synthetic buserite and birnes- under conditions (reduction with cinnamyl alcohol) not
site have analogousstructures, the shift from the l0A to found in nature. Also, the initial dehydration of Na
7A basal plane spacingin X-ray diffraction patterns being buseriteby drying over PaO16
in vacuo and conversionof
976
BURNS ET AL.: REVIEW OF THE TODOROKITE-BUSERITE PROBLEM
Na birnessiteto Mn7O13. 5H2O by refluxing with dilute
HNO3 at 40'C are unlikely geochemicalprocesses.Furthermore, the implication that todorokite in deep-sea
manganesenodulesrepresentsbuseritewhich has dehydrated,in situ to birnessite is untenable. Second, the
Swiss group did not prove that crystal morphology is
more definitive than diffraction evidence for identifying
manganite,but insteadassumedthat the fibrous pseudomorphs of decomposedMn7O13. 5H2O must be manganite, despite diffraction evidence to the contrary (see
below). Third, although some todorokite ores in the
weatheringzone in Cuba have been slightly weatheredto
manganiteand pyrolusite, there is no reason to believe
that the fibrous habit of this material is due to a large
componentof non-diffractingmanganitesincefibrosity is
an inherent property of all todorokites from other localities. Gram for gram, many todorokite ores give characteristic X-ray powder diffractionpatterns(nor containing
manganited-values)as intenseor more intensethan those
of syntheticbuserites.Such resultsare inexplicableif the
oresconsistof a mixture of non-difractingmanganiteand
undecomposedbuserite.
A more rigorous examination of the evidence for the
proposed birnessite-+manganitetopotactic relationship
(Giovanoli and Stiihli, 1970;Giovanoli, et al., 1970a,b,
l97l) reveals that the structural interpretationis incorrect. In the schematicstructureof manganite(e.g., Figure
ll in Giovanoliet a\.,1971),reproducedherein FigureI,
the authors inadvertently switched the a and b translations (with respectto the structure).In order to have the
electrondiffractionmeasurements
actuallyagreewith the
true manganitestructure (Buerger, 1936;Dachs, 1963),
the idealized structure depicted in Figure I must be
rotated90'about the c axis. In this orientation,the planar
triangular[111] surfacesof the [Mnur(OH,O)6]octahedra
in manganiteno longer coincide with similar surfacesof
the layersof [MnIvOu]octahedrain birnessite,so that the
postulatedtopotaxy is much less evident. The structure
correlationbetweenmanganiteand birnessitesuggested
by Giovanoliet al. in Figure I has anotherinconsistency
in that each row of [MnO6] octahedrain the birnessite
structure is drawn to align with two rows of octahedra in
the manganitestructureone of which is devoid of Mna+
ions, resultingin octahedrawith unusualproportions.In
addition,the structuressketchedin Figure I revealanother problem (S. Turner, pers. comm., 1982)in that the
spacing between the layers of birnessite is incorrect
(relativeto the^manganite
structure).They show-aspacing of about 5A insteadof the true value of 7 .27A.
One of the consequencesof mislabellingthe axes of
manganite in Figure I is that the indexing of electron
diffraction patterns of the break down product of
Mn7O13'5H2O purported to represent the a*-c* zero
level of manganite(Giovanoliet al.,1971, Filures 7 and
12; Giovanoli, 1980,Figure 15) does not sarisfy criteria
for the space group B21ld deterrninedfor the manganite
structure (Buerger, 1936).The difraction pattern of the
altered Mn7O13'5H2O phase (Giovanoli et al., l97l;
Giovanoli, 1980)does,however,roughlyfit that expected
for a (100) orientation for manganite(S. Turner, pers.
comm., 1982).Apparently,previousexperiencewith synthetic manganite(Giovanoli and Leuenberger,1969)led
to recognition of a manganite diffraction pattern but
incorrect indexingof it.
In summary, although the topotactic relationship portrayed in Figure I has many flaws, the overall interpretation of the experimentalwork of the Swiss group may
possiblybe correct, and syntheticplatey birnessitemight
break down to acicularmanganitecrystallites.However,
it is difficult to follow the logic presentedby Giovanoli
and Biirki (1975)that fibroustodorokitesnaturallyoccuring in marine manganesenodules and in non-marine
manganeseore deposits are mostly manganite. It becomesclear, though,that a critical stepin their identification of manganite based on crystal morphology and
electron diffraction evidenceis ambiguous.Again, it is
difficult to understand how the proposed buserite parent
in marine manganesenodule depositscould have dehydrated to birnessite and decomposedto manganitein situ
on the deep seafloor.
Finally, although past X-ray powder diffraction measurementsof marine ferromanganese
nodulesand crusts
have led to ambiguousidentificationsof todorokite or
buserite(" 10A manganite"),it is significantthat the more
recent electron microscopy techniques have identified
only todorokite in these seafloordeposits(Chukhrov er
al., 1978, 1979, l98l; Siegel, l98l; Turner and Buseck,
1982;Turner et al.,1982). Buseritehasyet to be positively identified in the natural environment. Such findings
underminethe recommendation(Giovanoli et al., l97l)
that the "l0A phase" in manganesenodules be named
buserite.
Crystal structure correlations
The recentHRTEMand infrared spectralmeasurements
describedearlier indicate that todorokites have tunnel
structures analogousto hollandite and romandchite (psilomelane).Turner and Buseck (1981)proposeda provisional nomenclatureschemefor manganese(IV) oxide
structuresin which familiesare designatedby the symbolism T(m,n), where T denotesa tunnel structureand m, n
are the widths of infinite chains of edge-shared[MnOe]
octahedraforming the walls of the tunnels.Thus, m : I
defines the nsutite family in which T(l,l) and T(1,2)
symbolizepyrolusiteand ramsdellite,respectively.Intergrowths of these fundamental units, perhaps with higher
dimensionaltunnels T(1,3), T(1,4), erc. (Burns and
Burns, 1980),characterizesyntheticyMnO2 and naturally-occurringnsutites(Turner and Buseck, 1983).Similarly, T(2,n) includes the cryptomelane-hollandite,
T (2,2),
and romandchite(psilomelane),T(2,3), groups, together
with the observed coherent intergrowths of T(2,4),T(2,7) multidimensional tunnels found in fibrous manganese oxide minerals (Turner and Buseck, 1979).Turner
BURNS ET AL,: REVIEW OF THE TODOROKITE_BTJSERITEPROBLEM
977
and Buseck (1981)suggestedthat the 7A phyllomanganate birnessitephasesrepresentthe end-memberT(2,o)
tunnel structure.
The todorokitefamily in the Turner and Buseck fl981)
classificationis representedby T(3,n) with T(3,3) the
eO
most commonstructure-type.The variouscoherentintergrowths in todorokitesrecognizedby them (Turner and
Oe
Buseck1981,1982;Turner et al., 1982)and other workers
(Chukhrovet al.,1978,1979,l98l) are designated
T(3,2),
T(3,3),T(3,4),T(3,5),T(3,6),T(3,7),etc., whltethe endmemberT(3,o) is suggestedto representthe phyllomanganatebuseritephases.
Although this nomenclaturesummarizesthe dimenO@
\t
tt
sions of the edge-shared[MnOe] octahedral chains in
todorokites,it doesnot definethe structuraldetailsinside
OO
OO
the tunnels. Some insight into the structure and crystal
chemistryoftodorokite can be deducedfrom correlations
with the known crystal structure of psilomelane(Wadsley, 1953).The "floors" and "ceilings" of the [2x3]
tunnels of psilomelaneare formed by double chains of
edge-shared
[MnOo]octahedracontainingonly Mn4+ ions
a M n 4 - ,M n 2 ' ,e t c , a t y = o
c=9 59A
c M n 4 . ,M n 2 ' , e t c , a t y = t / 2
in the M3 positions. The "walls", however, consist of
O oxygen at y=o
triple chains of edge-sharedoctahedra(definingthe psiI ' ruo
Q oxygen at y=1rz
"=n
lomelaneunit cell parametera : 9.564), in which Mna+
@ c a 2 - , N a ' , K ' , H 2 o , e t ci ,n t u n n e l s
ions in the central Ml position are flanked by larger
divalent cations in the M2 positions. In the todorokite
[3x3] structure-typeportrayed in Figure 2, the "walls"
are suggestedto be similar to those of psilomelanewith
divalent cations occupying outer M2 positions in the
triple chains of edge-sharedoctahedra and Mna+ ions
located in the inner Ml positions (definingthe unit cell
parameterc = 9.59A common to all structure-typesof
todorokite). The "floors" and "ceilings" in the most
common todorokite [3x3] structure-typecontain Mna+
ions in M3 and M4 positions of triple chains of edgeshared [MnOo] octahedra, giving rise to the unit cell
parametera = 9.754 (Fig. 2). A (001)projection of the
structure shown in Figure 2b provides another way of
viewing the "floors" or "ceilings" of the todorokite
structure.The triple chainsof linked [MnO6]run parallel
to the b axis and are separatedfrom adjacenttriple chains
by furrows of Mna+ ion vacanciesat the z : 0 level in the
(001)plane. In the (001)projecrion (Figure 2b) divalent
cationsin M2 positionslie aboveand below the furrows at
. manganese at z=O
approximatelevels z : ll4 and 314.
a=z.aqA
O oxygen
The observations of different todorokite structuretypes (Chukhrov et al., 1978, 1979, lg9l) and inrer+ /nn4* vacancies at z=O
a--9 754
(afso ivt2 a1 2=1141
growths of variabletunnel widths in todorokites(Turner
and Buseck, 1981,1982;Turner et al., l98Z)are depicted
Fig. 2. Proposedcrystal structure of todorokite. The diagram
schematicallyin Figure 3. This figure shows an intergrowth of [3x3] and [3x5] tqnnels,correspondingto the of Turner et al. (1982,Fig. l) has been redrawn to resemblethe
unit cell parametersa : 9.75A anda : 14.64, respective- psilomelanestructure-typeillustratedin Burns and Burns 1977b;
ly. Again, furrows of Mna+ ion vacanciesat the z = 0 Fig.4) so as to highlight the linkages of [3x3] tunnels in
todorokite (a) a (010)projection of the structure viewed down the
level definethe widths of the "ceilings" (or ..floors") of
[3x3] tunnels.(b) a (001)projectionshowingthe triple chainsof
the tunnels,as portrayedin Figure 3b. It becomesappar- edge-shared
[MnOe] octahedra forming the "ceilings" or
ent from the (001)projectionsshownin Figures2b and 3b "floors" of the [3x3] tunnels. A (001)projection showing the
that very wide tunnelsattainingdimensionsof [3x8] and triple chains constituting the "walls" is comparable.
BURNS ET AL,: REVIEW OF THE TODOROKITE-BUSERITE PROBLEM
978
lent cationsare bondedin postionsabove and below the
vacanciesin the layered structures (Crane, 1981).We
when discussingFigure 2b earlier that divalent
noted
a
cation-bearing M2 sites are located immediately above
a
and below the furrows of Mna* ion vacanciesin the (001)
planes of todorokite. Similar Mna* ion vacanciesmay
also be present within the "ceilings" of todorokite,
a
a
t
particularly where multiple-width edge-shared [MnOe]
domains exist. Examples are sketched in Figure 3b.
These cation vacancies not only may nucleate faults,
a
c =9 . 5 9 A
kinks, and twinning observed in nntElvt micrographs of
todorokite fibers (Turner et al., 1982), but they also
J
.1./.1 l.l.l.l.l.
would influencethe crystal chemistry and site occupancies of the tunnel interiors. As a result, three types of
atomic substitution might contribute to the crystal chem+
e
a=9.75A
*
a=14 6A
istry of todorokite.First, substitutionof Mna+ cationsby
cations of similar ionic radii (0.53-0.55A)in the "ceilings", suchas low-spinCo3* ions(Burns,1976).Second,
a+
+
.+
+
a
substitution of divalent Mn2* ions in the "walls" by
.+
+
a+
a
+.
+
Mg2*, Ni2*, Ctr2*, Zrf* and other cations having ionic
a+
a
.+
+
+.
+
radii in the range0.65-0.804, and third, constituentsof
.+
a
al
a
+
a
+
a
aa
a
a
the tunnels.The latter would consistof a variety of large
.+
.+
a
+
+
a
cations(K+, Ba2+,Ag+, Na+, Ca2*,Pb2*), H2O moleo
.+
a
+
+
a
+
cules, and smaller hyrated cations adjacent to Mna+
a
a
.+
+
.+
a
+
aa
+
vacanciesin the "ceilings."
a
a
a
.+
a+
+
+
The cation site occupanciesof todorokites and busera
.+
a+
+
+
ites proposed here also account for relative stabilities
a
.+
a+
+
+
+
toward oxidation. Todorokite and synthetic buserite are
a
both destabilizedby the presenceof substantialMn2+
.+
a+
+
+
+
a
a
a
ions in the structures, which are vulnerable to oxidation.
.+
a
.+
a
+
+
+
a
Thus, Mn2+-todorokites are oxidized to vernadite
r
aaaa*
aa*a
aa
*aa
*a
a*a
a
(Chukhrov et al., 1978, 1979),while synthetic buserites
a=14.6A
_l
are oxidized (accompaniedby dehydration) to birnesF a=9.75A -lsites. However, replacementof Mn2* by Mg2*, Zn2*,
of [3x3] and [3x5] tunnelsin
intergrowth
Fig. 3. Schematic
Ni2*,
and Cu2*, which are not susceptibleto oxidation,
(a)a (010)projection
vieweddownthetunnels.(b) a
todorokite.
stabilize
both todorokite and buserite.
(001)projectionshowinglocationsof Mna* cationsin the
Although detailed crystal structure refinements of
are
"ceilings"of the tunnels.SomesporadicMn4* vacancies
shownto existwithin the "ceilings" on the rightof the diagram. these cryptocrystalline manganeseoxides are urgently
required to establish the precise similarities and differencesbetween todorokites and buserites,it is apparent
higher(Turner and Beseck, 1982)tend towardsthe [3 x o]
that currently available evidence demonstratesthat they
Iayered stmcture postulated(Turner and Buseck, 1981) are distinct phases.
for buserite.Their predominancewould provide grounds
Conclusions and recommendations
for expecting a buserite-like phase to occur naturally in
Critical
examination of the published literature on
manganese
deposits.
marine
oxide
todorokites and buserites leads to the following concluAn importantfactor to be consideredin the "ceilings"
of the todorokite tunnels are cation vacancieswithin the slons.
(1) Naturally occurring todorokites and synthetic bubands of edge-shared[Mnlvou] octahedrawhich proba"ceiling"
serites
are crystallographicallydistinct phases.
prevalent
wider
the
the
bly become more
(2) Stability relationshipslinking the mineral todorokite
dimension.As noted earlier,Mna' cation vacanciesare
essentialfeatures of chalcophanite(Wadsley, 1955),as to synthetic buserite and its suggestedbreak down prodw e l l a s n u m e r o u s d i v a l e n t c a t i o n ( R 2 * ) - b e a r i n g ucts birnessiteand TMnOOH were deducedfrom phases
phases(e.g., Ostwald and Wampetich, 1967; preparedand degradedin the laboratory under conditions
R2+2Mn3Os
Riou and Lecerf, 1975, 1977) and possibly synthetic not encounteredin geochemicalprocesses.
(3) The validity of todorokite as a distinct mineral group
birnessite(Giovanoli et al.,1970a). In these phyllomanbeen demonstratedand a family of todorokite struchas
ganates,the Mna* vacanciesin sheets of edge-shared
ture-types
exists possessingmulti-dimensionaltunnels
divabecause
their
crystal
chemistries
dictate
octahedra
,l.l.l
l./.1.1.1.
BURNS ET AL.: REVIEW OF THE TODOROKITE-BUSERITE ZROBLEM
979
symbolized by T(3,n), where n, the dimension of the
A. Kozawa, Eds., ManganeseDioxide Symposium,Vol. 2, p.
"ceilings," is most commonly 3 (triple chains of edge97-112. Electrochemical Society Publication, Cleveland.
shared[MnO6] octahedra).However, n may rangefrom 2 Burns, V. M. and Burns, R. G. (1978)Post-depositional
metalenrichment processesin manganesenodules from the equatoto 7 or higher. With this terminology, n + @ may
rial Pacific. Earth and Planetary ScienceLetters, 39,341-j48.
represent the buserite group.
(4) Todorokite,not " l0A-manganite"or buserite,is the Burns, V. M. and Burns, R. G. (1979b)Observationsof processes leading to the uptake of transition metals in manganese
abundantMn(IV) oxide l0A phaseoccurringin manganinodules.In C. Lalou, Ed., La Gendsedes Nodulesde Mangaferous concretions and crusts including seafloor ferrondse,no. 289,p. 305-315.ColloquiumInternationaldu Centre
manganesenodule deposits.
National de la RechercheScientifique.
Buser, W. (1959) The nature of the iron and manganesecom-
Acknowledgments
poundsin manganese
nodules.In M. Sears,Ed., International
We are indebtedto severalpeoplefor helpful discussions,
Oceanography
Congress,
Abstracts, p. 962-963.
including
Drs.G. Arrhenius,
C. Bowser,p. Buseck,S. Crane,G. Chukhrov, V.,
F.
Gorshkov,A. I., and Sivtsov,A. V. (1981)A
Faulring,
M. Fleischer,
C. Frondel,M. Hey,J. Hunter,H. Jaffe,
new structural variety of todorokite. Izvestia Akademia Nauk,
R. McKenzie,U. Marvin,G. Mellors,P. Moore,J. Murray,R.
S.S.S.R.,SeriesGeology,5, 88-91.
Potter,G. Rossman,
J. Straczek,
andW. Zwicker.WethankDr. Chukhrov, F. V.,
Gorshkov, A. I., Sivtsov, A. V., and BereG. Arrheniusfor sendingus a preprintof his paperwith R.
zovskaya, V. V. (1978) Structural varieties of todorokite.
Giovanoli.Dr. S. Turnerprovidedan extremelyconstructive
Izvestia Akademia Nauk, S.S.S.R.,SeriesGeology, 12, 86reviewof themanuscript.
Research
on themineralogy
of manga95.
nesenodules
wassupported
by theNationalScience
Foundation Chukhrov, F. V., Gorshkov, A. I., Sivtsov, A. V., and Bere(grantnumberOCE-7821495)
and the I.C. SampleOffice(U.S.
sovskaya, V. V. (1979) New data on natural todorokites.
Electrochemical
Society,Cleveland
Section).
Nature, 278,631-632.
Crane, S. E. (1981)StructuralChemistryof the Marine Manganate Minerals. Ph.D. Thesis, University of California, San
Two recent publications support the conclusions
Diego.
reachedin this paper. First, todorokite is now acknowl- Cronan, D. S. (1980)Underwater Minerals. Academic Press,
New York.
edged by Perseil and Giovanoli fl982) to have several
am
featuresin commonwith the l0A phasein marinemanga- Dachs, H. (1963)Neutronen-undRdntgenuntersuchungen
Manganite, yMnOOH. Zeitschrift fiir Kristallographie, ll8,
nese nodules.This finding originatesfrom XRD, TEM,
303-326.
SEM and EMP measurementsof todorokite from the
Faulring, G. M. (1962)A study of Cuban todorokite. In M.
easternPyrenees.Second,further evidencefor the tunnel
Mueller, Ed., Advancesin X-ray Analysis,Vol. 5, p. 117-126.
structure of todorokite is describedby Arrhenius and Tsai Feitknecht,W., and
Marti, W. (1945)Uber die oxidation von
(1981),who suggestthat todorokite forms diagenetically
mangan (II) hydroxide mit molekularem sauerstoff. Uber
from buserite in marine manganiferousdeposits.
manganite und ktinstlichen braunstein. Helvetica Chimica
Acta. 28. 129-148and 149-157.
References
Finkelmann,R. B., Evans, H. T., Jr., and Matzko, J. J. (1974)
Manganese minerals in geodes from Chihuahua, Mexico.
Arrhenius,
G. (1963)
Pelagic
sediments.
In M. N. Hill, Ed.. The
Mineralogical Magazine, 39, 549-558.
Sea,p. 655-727,
Vol. 3. Interscience
Publicarion.
New york.
Arrhenius,G. O. and Tsai, A. G. (1981)Structure,phase Fleischer,M. (1980)Glossaryof Mineral Species19E0.Mineraltransformation,
and prebioticcatalysisin marinemanganate ogical Record, Arizona.
minerals.ScrippsInstitutionof Oceanography
ReferenceSe- Frondel,C., Marvin, U. B., and Ito, J. (1960)New occurrences
of todorokite.American Mineralogist,45, 1167-1173.
ries8l-28,l-19.
Buerger,M. J. (1936)The symmetryand crystalstructureof Giovanoli, R. (1980)On natural and synthetic manganesenodules.In I. M. Varentsov,and G. Grasselly,Eds., Geologyand
manganite,Mn(OH)O. Zeitschriftfiir Kristallographie,95,
Geochemistry of Manganese,vol. l, p. 159-202,E. Schwei163-174.
Burns,R. G. andBurnsV. M. (1977a)
Mineralogy
of manganese zerbart'sche Verlagsbuchhandlung,Stuttgart.
nodules.
In G. P.Glasby,Ed.,MarineManganese
p. Giovanoli, R. and Arrhenius, G. (1983)Structural chemistry of
Deposits,
marine manganeseand iron minerals and synthetic model
185-148.
ElsevierPublishing
Company,
New york.
compounds.In P. Halbach, Ed., Marine Mineral Deposits:
Burns,R. G. and Burns,V. M. (1977b)The mineralogy
and
New ResearchResults and Economic Prospects.Gluckauf,
crystalchemistryof deep-sea
manganese
nodules,a polymeEssen,West Germany.
tallic resourceof the twenty-firstcentury. In Mineralogy:
Towardsthe 2lst Century(CentenaryVolume,Mineralogical Giovanoli,R., and Briitsch, R. (1979a)Uber Oxidhydroxidedes
Society,London,p. a9-67)Philosophical
Mn(IV) mit Schichtengitter.5. Mitteilung: Stochiometrie,
Transactions
ofthe
RoyalSocietyof London,A286,283-301.
Austrauschverhaltenund die Rolle bei der Bildung von TiefBurns,R. G. andBurns,V. M. (1979a)
see-Mangankonkretionen.
Manganese
Chimia, 33, 372-376.
oxides.ln R.
G. Burns,Ed.,MarineMinerals,
p. 146. Reviewsof Mineral- Giovanoli, R., and Briitsch, R. (1979b)L'dchangedes ions de
ogy,Vol. 6. Mineralogical
Societyof America,Washington, transition par le manganate-l0Aet le manganate-7A.In C.
D. C.
Lalou, Ed., La GenEsedes Nodules de Mangandse,no. 289,
Burns,R. G. andBurns,V. M. (1980)
Recentstructural
datafor
Coloquium International du Centre National de la Recherche
(IV) oxides.In B. Schumm,
manganese
Jr.,H. M. Joseph,
and
Scientifiqueno. 289, 305-315.
Note Added
9E0
BURNS ET AL.: REVIEW OF THE TODOROKITE-BUSERITE PROBLEM
Giovanoli, R., and Btirki, P. (1975)Comparisonof X-ray evidence of marine manganesenodules and non-marine manganeseore deposits.Chimia, 29, ll4-117.
Giovanoli, R. and Leuenberger,U. (1969)Uber die oxydation
von Manganoxidhydroxid. Helvitica Chimica Acta, 52, 23332347.
Giovanoli, R., and StAhli, E. (1970)Oxide und Oxidhydroxide
des dreiund vierwertigen Mangans. Chimia, 24, 49-61.
Giovanoli,R., Biirki, P., Giutrredi,M., and Stumm, W. (1975)
Layer structured manganeseoxide hydroxides. IV: The buserite groups; structure stabilization by transition elements.Chimia,29,110-113.
Giovanoli, R., Feitknecht, W., and Fischer, F. (1971) Uber
Oxidhydroxide des vierwertigen Mangansmit Schichtengitter.
3. Mitteilung: Reduktion von Mangan (Ill)-manganat (IV) mit
Zimtalkohol. Helvitica Chimica Acta, 54, 1112-1124.
Giovanoli, R., Stehli, E., and Feitknecht, W. (1970a)Uber
Oxidhydroxide des vierwertigen Mangansmit Schichtengitter.
l. Mitteilung: Natrium-mangan(II,ilI) manganat(IV). Helvitica Chimica Acta, 53, 2W-220.
Giovanoli, R., Suhli, E., and Feitknecht, W. (1970b) Uber
Oxidhydroxide des vierwertigen Mangansmit Schichtengitter.
2. Mitteilung: Mangan (Ill)-manganat (IV). Helvitica Chimica
Acta,53, 4534@.
Glasby, G. P. (1977) Marine Manganese Deposits. Elsevier
Publishing Company, New York.
Hariya, Y. (1961)Mineralogical studieson todorokite and birnessite from the Todoroki mine, Hokkaido. JapaneseJournal of
the Association of Mineralogists, Petrologists, and Economic
Geologists, 45, 219-230.
Hewitt, D. F., Fleischer,M., and Conklin, N. (1963)Depositsof
the manganeseoxides: supplement.Economic Geology"58, l51.
Hey, M. H. and Embrey, P. G. (1974)Twenty-eighthlist of new
mineral names. Mineralogical Magazine, 39, %)3-932.
'l
Larson, L. . (1962)Zinc-beaing todorokite from Philipsburg,
Montana. American Mineralogist, 47, 59-66.
Levinson, A. A. (l%0) Second.occurrenceof todorokite. American Mineralogist. 45, 802-807.
Nambu, M., Okada,K., and Tanida,K. (1964)Chemicalcomposition of todorokite. JapaneseJournal of the Association of
Mineralogists, Petrologists,and Economic Geologists,5l, 3038.
Paterson,E. (19E1)Intercalation of synthetic buserite by dodecyclammoniumcbloride. American Mineralogist, 66, 424-427.
Perseil,E-A. and Giovanoli, R. (1982)Etude comparativede la
todorokite d'Ambollas (Pyr€n6esOrientales),des manganates
i l0A recontr6s dans les nodules polymdtalliques des oc€ans
et des products de synthCse.Compte Rendu Academic Sciencede Paris,294, 199-202.
Potter, R. M. and Rossman, G. R. (1979) The tetravalent
manganeseoxides: identification, hydration, and structural
relationships by infrared spectroscopy. American Mineralogist,64,1199-l2lE.
Radtke,A. S., Taylor, C. M., and Hewett, D. R. (1967)Aurorite,
argentian todorokite, and hydrous silver-bearinglead manganese oxide. Economic Geology, 62, 186-2M.
Siegel,M. D. (l9El) Studiesof the Mineralogy,ChemicalComposition, Textures and Distribution of ManganeseNodules at a
Site in the North Equatorial Pacific Ocean. Ph.D. Thesis,
Harvard University, Cambridge.
Simons, F. S., and Straczek, J. A. (1958) Geology of the
manganesedeposits of Cuba. U. S. Geological Society Bulletin. 1057.l-2E3.
Straczek,J. A., Horen, A., Ross, M., and Warshaw, C. M.
(1960) Studies of the manganese oxides. IV. Todorokite.
American Mineralogist, 45, ll7 4-1184.
Turner, S. and Buseck, P. R. (1979) Manganeseoxide tunnel
structures and their intergrowths, Science, 203, 456-45E.
Turner, S. and Buseck,P. R. (1981)Todorokites:a new family of
naturally occurring manganeseoxides. Science, 212, 10241027.
Turner, S. and Buseck, P. R. (19E2)HRTEM and EDS of tops
and bottoms of someNi-bearing manganesenodules.EOS, 63,
1010.
Turner, S. and Buseck,P. R. (1983)Defectsin nsutite(rMnOJ,
a dry-cell battery cathode. Nature, in press.
Turner, S., Siegel,M. D., and Buseck, P. R. (1982)Structural
features of todorokile intergrowths in manganesenodules.
Science, 296, 841-842.
Wadsley, A. D. (1950a) A hydrous manganeseoxide with
exchangeproperties.Journal of the American Chemical Society, 72, 1782-1784.
Wadsley,A. D. (1950b)Synthesisof somehydratedmanganese
minerals. American Mineralogist, 35, 458499.
Wadsley, A. D. (1953) The crystal structure of psilomelane,
(Ba,HzO)zMnsO1q.Acta Crystallographica,6, 165, 172.
Wadsley, A. D. (1955)The crystal structure of chalcophanite,
ZnMryO1 . 3H2O. Acta Crystallographica, 8, 165-172.
Yoshimura, T. (1934) Todorokite, a new manganesemineral
from the Todsoroki Mine, Hokkaido, Japan. Journal of the
Faculty of Science, Hokkaido University, Sapporo, Series 4,
YoL 2,289-297.
Manuscriptreceived,December2I, l98l;
acceptedfor publication, February 22, 19E3.