American Mineralogist, Volume 64, pages 1219-1226, IgTg

Mineralogy of manganese dendritesand coatings

Russeu M. PorrEn' AND GBoRcB R. RossuaN Division of Geologicaland planetary Sciencef Califurnia Institute of Technology Pasadena,California 9I I 25

Abstract Infrared spectroscopic study of desert varnish, manganese dendrites,river deposits,and other manganese oxide concentrations of the terrestrial weatheringenvironmenthas shown that the morphological distinctions among them have a sound basis in di.fferences in their mineralogy' The manganese oxide in manganese dendritescollectedin surfaceexposures is either romanechiteor a hollandite-groupmineral. Theseare mixed with varying amountsof silicateminerals,which are a passive substrate for the oxide deposition.Dendritescollectedin undergroundmine workings are todorokite. Manganese streamdepositsare generallybirnessite with minor amounts of silicate minerals;one nsutite streamdiposit has been identitred. Crack deposit mineralogy resembles that of manganese dendrites.Cave and subglacialdeposit mineralogyresembles that of manganese streamdeposits. Although dendriteshave long beenconsidered to be pyrolusite,no exampleof pyrolusitemineralogyhasbeenidentified.

range order, it providesmineralogicalinformation on Manganeseoxide concentrationsare a cornmon even the most disordered and finely-particulate product of terrestrial weatheringprocesses. Its successful application to determinationof Concen- phases. trations with variable morphology, chemis_ desert varnish mineralogy (Potter and Rossman, try, and physical propertiesoccur in diverseenviron_ 1979a)led us to systematicallyinvestigateits usefulments throughout the world: desert varnish in arid nessfor determinativemineralogy of the manganese and semi-arid regions, manganese dendrites yilhin oxides(Potter and Rossman,1979c). This paper presentsthe results of a mineralogical and on the surfaces of rocks, deposits within cracks in rocks, and associated oxide concentrawith rivers, caves,and gla- study of a wide rangeof manganese tions of the terrestrial weathering environment. Inciers. Determination of manganese indicates distinct differencesin oxide mineralogy of frared spectroscopy thesematerialsby X-ray difraction has beendifficult mineralogy which can be correlated with the moror impossibleowing to the finely particulate and dis- phology of the deposit and its depositionalenvironordered nature of the manganeseoxides and the ment. presenceof silicate phasesas major componentsof Experimental the deposits.For this reason the structural and genetic relationsamongthem are poorly understood. We prepared powdered samples as follows: milliInfrared spectroscopyis particularly useful for gram quantities of the manganese oxide concentramineralogical analysesof manganese oxide concen- tions were removedfrom their substrates by scraping trations. It eliminates ambiguity frequently caused with a tungstenneedleunder 20x magnification.The by silicate components(such as the confusion of ka- resultant material was examined under 40x, lightolinite and birnessite). Because it is sensitiveto short_ colored contaminants removed, and the material ground under toluene to a fine powder in a boron carbidemortar and pestle.When carbonatewas presI Present address: Owens-Corning FiberglasCorporation,Gran_ ent in the sampl, it was removed by treatment with ville. Ohio 43023. 1.0 M acetic acid for 5 minutes prior to any further 2 ContributionNo. 3153. analysis.Manganeseand iron oxides were extracted
w3-o04x/7 9/ I I I 2-12l 9$02.00

Introduction

1220

POTTER AND ROSSMAN: MANGANESE DENDRITES

from the samplesby the sodium dithionite procedure to a which was shortened (Mehra and Jackson,1960), single 5-minute extraction to minimize the effect on and their the silicates.Infrared spectraof the samples with a were obtained residues dithionite-extracted on 0.50 Perkin-Elmer Model 180 spectrophotometer mg of powdered sample dispersedin 200 mg KBr and dried overnight at 120"C under vacuum to remove water adsorbedon the KBr. This treatmentaffected only those infrared features due to adsorbed water. In the figures we have presentedspectra at of their true intensitiesto facilivarious percentages tate comparison.The intensity for any sample may in its parvary by a factor of2 or 3 due to differences ticle sizeand degreeof dispersionin the KBr. We oboxides in tained infrared spectra of the manganese by computergenerationof differencespectra samples (for example,Fig. l). Absorption due to silicateswas renoved from a spectrum (Fig. lC) by subtracting from it that fraction of the extracted residue specto null the absorptionnear 1000cm-' trum necessary The result(Fig. lD), which is due wholly to silicates. I <ntcnoNs I^IFVELENGTH 8101525

ing differencespectrum(Fig. lB) is that of the manganese and iron oxidesin the sample.For desertvarnish samples, where Mn and Fe are present in comparable concentrations, features attributable to both manganeseand iron oxide are present. The other depositsstudied are either free of Fe or have it in sufficiently lower concentration than Mn that it does not interfere with the determination of manganese oxide mineralogy. We used a scanning electron microscope equipped with an energy-dispersiveXray analyzerfor semi-quantitative chemical analyses. X-ray powder diffraction was done with a DebyeScherrercamera using vanadium-filtered chromium radiation. Results Desert varnish The results of a detailed study of samples l-3 (Table l) have been reported elsewhere(Potter and The additional varnish sam1977,1979a). ior.-"tt, ples listed in Table I confirm the generality of those iesults. Desert varnish invariably consistsof birnessite in intimate mixture with hematite and large amountsof clay minerals.The clays are generally of the illite-montmorillonite type with minor kaolinite, but in severalsamplesthis concentrationratio is reversed. Manganesedmdrites We define manganese dendrites as those manganese oxide concentrations which exhibit a dendiitic pattern. They may be within the rock of the rock, matrix (internal), along fracture surfaces or on its surface. oxide minerals (Burns Ring-structure manganese of all three types and Burns, 1977\ ate characteristic exception(No' a single With dendrites. of manganese the dendrites present in mineral 18) the manganese (Ba, HrO)rMn'O,o romanechite has been found to be or one of the following hollandite-group minerals: hollandite, BaMnrO,u; cryptomelane, KMnrO,u; coronadite, PbMnrO,u. Figures I and 2 ate representative of the spectral data on which mineralogical identification is based, although in generatio; ofdifference spectrawas not necessary distinguish to analysis We used chemical uil "n."t. hollandite-group minerals because their di-fferent infrared spectra are identical in the 4000 cm-' to 200 cm-' regi,on. Distinguishing features between the romanechite of Figure I and the hollandite of Figure 2 are the position of band near ?00 cm-', the relative intensity of the band near 600 cm-' (which is

II

^l
LJl

t_!

zCE co E. o a co ctr

D
1200 1000 800 600 q00

<cn-1: I^IRVENUMBEB
Fig. t. Infrared spectra identifying romanechite in manganese dendrite No. 17:(A) romanechitestandardNo.34,41% intensity' In this and subsequentfigures the standard number refers to Table I in Potter and Rossman(1979c\;(B) differencespectrumof C and intensity; (D) D, lWo intensity; (C) unextracteddendrite, 1007o l7o intensity. 5 dendrite, extracted

POTT:ER AND ROSSMAN: MANGANESE DENDRITES Table l. Sampleinformation

Accessory mineraloBy Chemist!y coatlnS 2 3 coatinS on andesire o. sMll boulderg ourcrop

t22l

quartzlte

coatlng on deserE pavehent rhyollre and quartz cobbles srreaked "stain" qurrz coarlng consloneraEe canyon walls coatlng on snall sandsrone outcrop and desert pavenenE dacite porphyry cobbles coatidg on basalr ourcrop coatlng coatlnS coaElnS coaEing coating coathg coatlnt coarins on dese!r on colluvial on alluvial on deselt on snall on basatc on cherr on basalt pavenenc gneiss llnesron pavenenr banded llon bouldels arrifacr cobble 1n d.serr basalt cobbtes

1 8 9 LO I1 L2

and hornfels cobbles rock

volcanlc fotutlon

manganese

dendriEes

l5 r6 17 18 19 20 2l 22 23
Beechrorth, oro Auscralia Argus Peak, Ca. AfEon Canvon, Ca SouLhuescern UniEed States

h,

Pb/cal/zn, cu,At. si ,re Cal/ Si, Al, Fe

h/K, h,

surficial dendlites up ro I alluvial sandsEone cobble < 100 !n internal dendrites

rhlck ln

on

thlck

s i, BalA1,Cal/Mg, Pb, Ie,Na Mn,si ,41/ ca,Ti/ /Fe,K, Pb ca,Mn, si/Bal /K,Mr, Ar h, si , BalK//Ca,Al ,Pb

fracture surface and intern6l dendraEes < 100 !h chlck on pe8mrltlc feldspar surftclal and internal dendrlt.s up ro I m rhick on peghatit tlacture surfac dendlites up ro I m thick on siltslone ourclop inrernal dendriric staln 1n sllt.tone inEernal dendriEic stain lb calc.r.oue claystone alluvial boulder surficial dendriles up to I m rhlck on altuvial volcanlc !ock surfi.ial dendrites up to 1 6 thlck on lnrernal dendrlEes < 100 !n rhick ln micaceous tanellae of quartz nlca schlst veins of purer macerial ln concrerlone oxide and clay fron a dry arreanbed coaring on pebbles of varled llrhology fron a dry streanbed coatins r 100 !h thtck on.trarbcd qua!rzice cobbles coaEing < 100 um thick on streahbeo gneiss.obbles < 100 !m rhick on badelt cobbles coating in splash zone concrerio.s in st..am a1luvlum con.reElons in strean aIluvlun probably associated of

un, si,M8lA1, Bal/ Ca Mn,Bal/si,ca,K Mn/K//Ba, sl, 41,Ms Mn/Bal K,A1, St / /ca

PreEo, Brazil

25 26 27 2A 29 30 l1 )2

Soucheastern Uniced Scaces Sou.heasEern Unired SraEes

Hn, re ,Cal S1/ /K, Ba Un, Si ,Fe, Calcl,A1l/Mg,Na

S 1 , M n ,I e , A 1 / c a , M 8 / / h, Fe/cal /A1, si/Ni, Ba

Si,Al,h,Ca,Iel/'Ti,K,Mg h/ /Al, k, S i , C a, K , F e

rh, S1,A1, fe/Cal/K, zn Mn/si,K//AI

mouncainslde deposic Fith a sprinS

niscellaneous subBlacial

sanDles deposlE

ll

Jackson clacier,

Xont
Qtz/ /chL

S1,A1 ,h,

Fe,ls/Ca ,R//

coating < 100 !n rhi.k on sltrstone surfnce unde. sta.ier .oaElng < I00 !m on silrston su.fu.e under elacier

ouccrop

14

Khuhbu Resion,

Nepal

Si,Al, Fe,h/K,

Ti ,Cal /M8

ourclop

35 16

Sierra

Narl

Foresr,

ca

rohanecbite

M n ,S 1, A 1 / / K , C a , U g , F e M n .B a l S l . A l . M q / / (

Cady Mtns

, Ca

coaEing up to 1 nF thlck near a water seeP coaclng up to I ton chick

on granite on votcanic

outcrop rock

counrerfelt

Mn

37 18 39 40 4l

5ier.a

NaEl

Folesr, Ca , Ca

Ca.

Rock Cieek, Sumir ilteor

Meadow Lk Crarer, A2

Fe/At,Mn.s1//

coatlnS up to I m ln a eacer seeP coatlnS up to I m ln strean splash .oating up to 1 m outlet stranbed dendrltlc grourh on

ahick thlck zone chi.k

on tlantre on Srantre on alan1te

oucclop boulders 1n lake

shocked

sandsrone has rh

Anrarcrlca

pollshed surface of hornfels whlch appearance of des!t varntsh

*{Nolse drrecc tAn iron

qhtch are nearty tndistinaulshable laye! clay, illire, or nuscovlte! by = carbonace; Fld = fetdspari Cta = unldenrlfied ctay nlnelat: e used to indlcate rouShly che retatlv anouncs.t r."""..." Dlnerars and slash are presenr in such h18h concentrarton rhar rhey dm1;are the lnfrared ient concenr!at1on that rhey sl8nlflcanrly affecr the spectrh. A.cessory oscopy but are present in loo 1os a concentratlon ro s1gn1f1canc1y alEer mlnant consrituentsi chose bereeen slDgle and doubl stashes ar hhor . This ls based roughly on peak heiShr frofr Ehe enelty dlsperslve X-ray s and sanple localirles ale found 1n Appendlx B, potrer (1979). incroduced bv the Seneration of conparison Eo other speccra of oxide Phase ls Present. Its difference spect.a 1s Eoo large Eo alloe deteminarlo. of nanganese nlnraloty rhe sane phenomenon.rtl accessory m1nerals. "rnri.. sas Doc determined, buc 1t is presuDed co be bematiEe on the basts of hstead, cotor. 1r 1s inferred flo'

nlnelalo8y

1222

POTTER AND ROSSMAN: MANGANESE DENDRITES

MIcEoNS) N R V E L E N G T( H 8101525

l
lJl

1
E O U) G

z. CE co
co

necessarily involved in the dendrite formation dendritesof sampleNo' The fracture-surface process. 17 ,pp"u, to be a local staining of clay minerals ptet*t throughout the fractures.Accessoryminerals of th" ittt"-al dendrites are in many casessimply the minerals of the rock matrix in which the dendrites have formed. Manganesedendrites do not generally have a high degree of chemical homogeneity. The manganese concentration varies greatly throughout the stained area in those dendriteswhich are not pure oxide' In most dendrites areastens of microns in size can be found which are predominantly manganese oxide' The chemical analyses in Table I are from these areas. In other areas the dendrite chemistry is dominated by the accessoryminerals. Manganese oxide sesfingson dendritesappear to be manganese minerals' We a substratecomposedof the accessory have found only trace quantities of Fe in manganese dendrites. This distinguishes them from other manganeseconcentrations, which generally contain Fe as one of the major constituents. Stream deposits Our stream deposit samples include coatings on stonesfrom both the streambedand the splashzone, and concretions from stream alluvium. The mineralogy of thesesamplesis dominant manganese birnessite,(Na,Ca,K)MorO,o'3HrO. Figure 4 is representativeof the spectral data on which this 'is based' The two mineralogical identification intense, poorly resolved bands near 500 cm-' are

1200 1000 800 600 tl00 > <cu-t hIRVENUMBER

Fig. 2. Infrared spectraidentifying hollandite-groupmineral in dendrite No. 24:.(A) cryptomelanestandard No. 28' manganes intersity; (B) difference spectnrm of C and D, lWo intensity; 28Vo (C) unextracteddendrite, 1007o intensity; (D) extractcd dendrite' 75% intensity.

prominent in the hollandite-groupminerals and only of the a shoulder in romanechite),and the presence In romanechite. for cm-' 450 weak band near at bands HrO absorption has romanechite addition, (Fig. 7), which cm-' 1605 and 3470 cm-', 3520cm-', oxides. Our distinguish it from all other manganese Ba (Table high concentrations l) show data chemical in those sampleswith romanechitemineralogy and thereby support the infrared results. The single example of a non-channel structure manganese oxide, No. 18, was unambiguously identified as todorokite, (Mn,Ca,Mg)MnrOr'H'O (Fig. 3). We have found no dendriteswith pyrolusite mineralogy. The general term "pyrolusite dendrite" is a misnomerand should be discontinued. Manganese dendrites have no characteristic accessorymineralogy such as the high concentration of clay mineralsin desertvarnish. Somedendritesare nearly pure manganeseoxide. In others the oxide is mixed with various carbonateor manganese silicate minerals characteristic of the weathering environment. These accessoryminerals are not

<ntcnoNsr NRVELENGTH
8101525

I (nl -l J ul
(J

z ct
@ E O

a co

CE

1200 1000 800 600 q00 <cr-1> I,'IRVENUMBEB

Fig. 3. Infrared spectra identifying todorokite in manganese intensity; dendrite No. l8: (A) todorokite standardNo. 55, 1367o intensity. (B) unextracted dendrite, 1007o

POTTER AND ROSSMAN: MANGANESE DENDRITES

1223

characteristicofbirnessite and distinguishit from all NRVELENGTH <rrcnoHs I other manganese oxides (Potter and Rossman. 8101525 1979c), although it is easily confused with some todorokite samples. Most todorokites have spectra distinct from that of Figure 4C (Fig. 3,{ for example), but the spectrum of highly disordered todorokite (Fig. aA) is similar. The nearly equal intensity of the two major absorption bands of Figure 4C and, the positionsof the weak shouldersat higher wave number indicate a birnessiterather than a todorokite mineralogy. eualitative chemical analysesof the river deposits support a birnessite mineralogy. Invariably Ca and/or K are present in L.IJ minor amounts.Nowlan (1976) has found birnessite z. in one streamin Maine and noted that many samples G co from his other Maine localities(amongwhich are our cc O samplesNo. 30 and 3l) have the two broad X-ray U) co lines characteristic of 6-MnOr, which is thought to be CE a disorderedbirnessite(Buseret al.,1954). We have identified sample No. 3l as nsutite. Mn(O,OH)r.xHrO, on the basis of its infrared spectrum (Fig. 5) and chemistry. The spectrum of sample No. 3l is characteristicof nsutite in the broadnessof its band near 580 cm-' and in the 1200 1000 800 600 q00 positions of the weak bands at energieslower than h R V E N U I 4 B< Ec B N> -t 400 cm-'. However,the reliability of identification on Fig.4. Infrared spectraidentifying birnessitein stream deposit this basisalone is questionablebecause of the strong No. 3 l: (A) disorderedtodorokite standardNo. 57, l20Vointensity; similarity betweenthe infrared spectraof sampleNo. (B) birnessite standard No. 47, l007o intensity; (C) difference 31 and the hollandite-group minerals. eualitative spctrum of D and E, lWo intensity; (D) unextracteddeposit, intensity;(E) extracteddeposit, l47ointensity. chemical analysis which indicates only traces 1007o of cations other than Mn confirms the nsutite caves(Moore and Nicholas, 1964),where they form mineralogy. Because of differences in cation ab_ coatings on stream cobbles and soot-like layers on sorption properties,this mineralogicaldiference may walls and other cavesurfaces. have profound effectson the use of trace elementsin manganese oxides for geochemical Miscellaneous exploration. Streamdepositsdo not, in general,exhibit the high Subglacial manganese deposits:We have examined degreeof separationof manganese from iron seenfor samplesfrom two areaswhere a recedingglacier has the manganese dendrites.We found Fe to be a major exposed bedrock surfaces stained with manganese constituent of many of the stream samples.More oxides. The infrared spectrum of the manganese extensivechemical analyses(Carpenter et al., 1975; oxide is attributable to either a disorderedbirnessite Nowlan, 1976)indicate that this is generally true for or a disordered todorokite (Fig. 6). The latter is a Turkey Creek, Pyrite Creek, and Shawn Creek. better match, although the intensity of the lowerNowlan's report of nearly equal amountsof Mn and energybirnessiteband is variable.We suggest (potter Fe in Pyrite Creek depositsrepresentsthe average and Rossman, 1979c)that birnessiteand todorokite composition;our analyses are of the manganese-rich disorder to the same structural entity. If this areas. suggestion is correct,a continuum existsbetweenthe two structures, and as disorder increases the Cave deposits distinction between the two minerals becomes The two cave deposits examined are both meaningless. Except for a slight increase in the predominantly birnessite and resemble the river amount of chlorite, the bulk of the stained material depositsin all respects, which is not surprising since does not differ from the rock matrix, which is they came from dry streambeds in caves.Birnessite predominantly quartz. In both samplesFe, probably and romanechitehave previously beenreportedfrom as iron oxide, is a major constituent.It appearsthat

POTTER AND ROSSMAN: MANGANESE DENDRITES

<utcnoHsl NFVELENGTH 8101525

Conclusions Our resultsindicate that distinctionsbetweenmanganeseoxide concentrations based on their morphologic relations in the terrestrial weathering environmint have a sound basis in terms of the mineralogy mineroxides and their accessory of the manganese stream and dendrites, manganese varnish, als. Desert which mineralogy characteristic a have each deposits environand morphologies diferent reflects their ments. Desert varnish is composedpredominantly of illite-montmorillonite and kaolinite clay minerals with lesserand somewhatvariable amounts of hematite and birnessite. In manganesedendrites, the manganeseis generally present as romanechite or a hollandite-group mineral and is sometimesthe major constituent of the dendrites. The accessorysilicate mineralogy is variable, and iron oxide is present in oxide of a birnessite only minor amounts.Manganese phase in manganese dominant is the mineralogy and clay minoxide iron minor but deposits, stream > <utcnoNs I,,IRVELENGTH 8101525

I C\I ol
I

I
I

LtJ

z CE m ctr
O

Ln co ct

1200 1000 800 600 '100 cB n-l> h R V E N U M B< E

Fig. 5. Infrared spectraidenti$ing nsutite in river deposit No. 30: (A) cryptomelanestandardNo. 28, ll07o intensity; (B) nsutit standard No. 16, 8870intensity; (C) unextracted river deposit, intensity. 100%

t
^l -t
ol

the manganeseoxide is deposited on the rock in relatively pure form. Crack deposits:The two manganesecrack deposits dendritesin we have examined resemblemanganese and mineralogy, accessory manganesemineralogy, generalize to few samples too We have iron content. from these results. These crack deposits formed at unknown depth within the rock, and should be distinguished from more surficial crack deposits which have formed within the upper few centimeters of the crack in associationwith desert varnish at the surface and iron oxide-clay deposits deeper in the ofthese latter and association crack. The appearance varnish in desert resemble they that depositssuggests composition. and mineralogy Counterfeit manganesedeposits:Polished surfaces' organic matter, and iron oxide with traces of of a oxide can each give the appearance manganese field careful Although stain. oxide manganese examination can generally distinguish these phenomena from manganeseconcentrations, it is sometimes neoessaryto ascertain the presence of manganese oxide by laboratory tests. Several examplesare listed in Table 1.

II
I

LlJ (J

z. CE m
G. O

(n co
CE

1200 1000 800 600 r'100 > <cr-1 l,'IFVENUMBEB

Fig. 6. Infrared spectra identifying birnessite in subglacial deposit No. 33: (A) disorderedtodorokite standard No. 57' 75% intensity; (B) birnessite standard No. 47, 607o intensity; (C) difference spectrum of D and E, lOVo intensity; (D) unextracted subglacial deposit, 100% intensity; (E) extracted subglacial deposit,53%intensitY.

POTTER AND ROSSMAN: MANGANESE DENDRITES

t225

erals are generally present.The diverse minslnlsgy previous workers have reported for manganese soil concretions(Taylor et aI., 1964;Gallaher et al.,1973: McKenzie, 1977; Ross et al., 1976; potter and Rossman, 1979b) contrasts with desert varnish. manganese dendrites, and stream deposits. This reflects the wide range of environmental conditions possible in soils. Data on several subglacial deposits and crack deposits suggeststhat the former resemble river deposits, and the latter resemble manganese dendrites. This is reasonable in terms of the environnental conditions. The characteristic differences may reflect differencesin the formation mechanismsof the manganese deposits and are therefore useful in further understandingthesemechanisms. The variable mineralogy and quantity of manganese-dendriteaccessoryminerals and their frequent similarity to minerals of the rock matrix argue strongly that they are simply passive contaminants.We can rule out several of the possible factors which might control a romanechite ys. hollandite-group mineralogy in manganesedendrites. The presence of Ba doesnot direct the mineralogy to romanechite, as evidenced by the presenoe of hollandite in three dendrites. Rock lithology showsno correlation with mineral species, nor does NFVELENGT <H ulcnorsr 3r-t56

the relation of the dendrite to the rock (internal, fracture surface, or surficial). Romanechite is stable to conversionto hollandite up to at least400"C (Fleischer, 1960),so that thermal history, exceptas it may influenceprimary mineralogy,is probably not an important factor in this regard. The relation between romanechite and hollandite is a closer one than distinct mineral names might suggest.Turner and Buseck (1979) have shown that intergrowth occurs between the two minerals at the level of the unit cell and have explained variations in romanechite X-ray diffraction patternson that basis.Severalof our samples have infrared spectra intermediate between those of romanechite and hollandite (Fig. 7). This may be the result of intimate structural intergrowth, although it could be explained by an intimate physical mixture. We found no evidencefor mixed phases under the scanning electron microscope. The intensitiesof the water absorption bands in the 4000 cm-' to 1400cm-' do not follow the trend set by the absorption bands in the 1400 cm-' to 200 cm-' region, which are due to vibrations of the ring framework of MnOu octahedra. This reflects the variable amounts of water in romanechite and suggeststhat water is not important in controlling the mineralogy. At the Cady Mountain locality we found desert 8 NfiVELENGTH <Nlcnorsr 25 10 15

Nl

OI

J 1
U

t!

z.

CE

m ctr a
@ CE

CE

co ctr O (n m
CE

B
a

3600

3000 2q00 1800 I ^ I F V E N U M B< Ec Bx - t :

1200 1000 800 600 rl00 NRVENUMBE <B cN-l:

Fig. 7. Infrared spectraillustrating the range betweenpure romanechite and pure hollandite structures: (A) cryptomelane (hollanditegroup) standard No. 28, 90Vo intensity; (B) creek deposit No. 36, diference slrcctra at 1057o intensity (minor quartz absorption removed); (C) manganese dendrite No.22,94% intensity; (D) romanechitestandardsNo. 38 (4000cm-,-1460'cm-t; ana No. 34 al4m cm-t-2fi) cr,n-l1,lMo intensity.

1226

POTTER AND ROSSMAN:MANGANESE DENDRITES

stitute of Technology; B. Hallet, Stanford University; J. Hayden' Tucson, Arizona; A. Jelinek, University of Arizona; B' Kisner' Hollywood, California; K. Nealson, Scripps Institution of Oceanography; G. Nowlan, U' S' Geological Survey; R' Perry' University of Washington; R. Sharp, California Institute of Technology; R. Stoessel, Exxon Production Research, Houston, Texas' A portion of the funding for this work was provided by the L' S' B' Leakey Foundation and the John A' McCarthy Foundation' References Burns, R. G., and V. M' Burns (1977) Mineralogy' In G' P' Glasby, Ed., Marine Manganese Deposits, p' 185-248' Elsevier, Amsterdam. Buser, W., P. Graf and W. Feitknecht (1954) Beitrag zut Kenntnis und des 6-MnOr. Helvetica Chim' der Mangan(Il)-manganite Acta,37,2f22-2333. Carpenter, R. H., T. A. Pope and R. L. Smith (1975) Fe-Mn oxide coatings in stream sediment geochemical survys' J' Geochem' Explor., 4,349-363. Fleischer, M. (1960) Studies of manganese oxide minerals' III' Psilomelane. Am. Mineral., 45, l'l 6-18'1. Gallaher, R. N., H. F. Perkins, K. H' Tan and D. Radclitre (1973) Soil concretions: IL Mineralogical analysis. Proc. Soil Scr" Soc' Am., 37,469472. McKenzie, R. M. (197?) The manganese oxides and hydroxides' In J. B. Dixon, F,d.,Minerals in Soil Environmenls,p' l8l-193' Soil Science Society of America, Madison, Wisconsin' Mehra, O. P. and M. L. Jackson (1960) Iron oxide removal from system buffered with sosoils and clays by a dithionite{itrate dium bicarbonate. In A. Swineford, Ed.' Proc. Seventh Natl' Conf. Clays and CIay Miner.,p.317-327. Pergamon, Oxford' Moore, G. W. and G. Nicholas (L964) Speleology' D' C' Heath and Company, Boston. Nowlan, G. A. (1976) Genesis of manganese-iron oxides in stream sediments in Maine. lJ.5. Geol. Surv' Open-File Report 76-878' Potter, R. M. (1979) The tetravalent manganese oxides: clarification of their struclural variations and relalionships and characterization of their occarrence in the teftestri^l weathering environ' ment as desert varnish and other manganese oxide concentrations' Ph.D. Thesis, California Institute of Technology' Pasadena, California. and G. R. Rossman (1977) Desert varnish: the importance of clay minerals. Science, 196, 1446-1448. and 1lQ79a) The manganese and iron oxide miner-9 alogy of desert v arnish. C hem. Geol., 25, 79 4. -(1979b) A magnesium analogue of chalcophaand nite in manganese-rich concretions from Baja Califomia' Am' M ineral., 64, 1227-1229. (1979c) The tetravalent manganese oxides: and identification, hydration, and structural relationships by infrared spectroscnpy. Am. M ineral., 64, 1199-1218' Ross, Jr., S. J., D. P. Franzmeier and C. B. Roth (1976) Minetalogy and chemistry of manganese oxides in some Indiana soils' Proc. Soil Sci. Soc.Am.,40' 137-143. Taylor, R. M., R. M. McKenzie and K' Norrish (1964) The mineralogy and chemistry of manganese in some Australian soils' Aust. J. Soil Res.,2,235-248. Turner, S. and P. R. Buseck (1979) Some manganes oxide tunnel structures and their intergrcwlh. Science, 203,456458' Manuscript received, January 2, 1979; acceptedfor publication, February l1' 1979-

varnish (No. l0) and crack deposits (No. 36) developed within a meter of each other. Several meters away were manganese dendrites (No. 22). The varnish has birnessite, and the crack deposit and dendrites are romanechite-hollandite intergrowths (Fig. 7). Clay minerals are not detectable in either the crack deposit or the dendrites. It is not known how far below the ground surface or at what time the crack deposits and dendrites formed, but the occurrence in close proximity of manganese oxide concentrations which differ markedly in mineralogy suggests that the accessory clay minerals may be an important factor in controlling manganese oxide mineralogy. Although the presence of clay minerals is not a sufficient condition for birnessite formation (some manganese dendrites have appreciable clay associated with hollandite or romanechite), it appears to be necessary. Clay minerals are an integral part of desert varnish, which invariably has birnessite. All stream deposits with birnessite mineralogy are associated with clay minerals. The only stream deposit lacking clay minerals is the nsutite from Pyrite Creek. Environmental conditions do not appear to be responsible for the distinctive mineralogy of the Pyrite Creek deposit. The Shawn Creek deposit has the normal birnessite mineralogy. These two streams are within 20 km of each other in western Maine and differ only in water trace-element chemistry (Nowlan, 1976). Pyrite Creek water is anomalously high in Pb and Zn, but we know of no reason why this should lead to nsutite formation. Burns and Burns (1977) have suggested that epitaxial growth on iron oxides may be responsible for the occurrence in marine manganese nodules of 6-MnOr, which is believed to be a disordered birnessite. With the exception of some birnessite soil concretions (Taylor et al.,1964), there is a good correlation between birnessite mineralogy and iron content in manganese oxide deposits of the terrestrial weathering environment. Infrared spectroscopy has provided mineralogical information on deposits which are too poorly crystalline, too finely particulate or too highly contaminated with other crystalline phases for successful study by X-ray diffraction. It is hoped that infrared spectroscopy will find wider application to more detailed investigation into the genetic nature and relationships of the various manganese oxide concentrations. Acknowledgments J. Bard,LawWe thank the followingfor providingsamples:
rence Berkeley Laboratory; A. Bauman, Jet Propulsion Laboratory; R. Carpenter, University of Georgia; M. Furst, California In-