Clay Minerals (1989) 24, 531-538

BAYERITE, AL(OH)3, FROM RAOUL ISLAND, KERMADEC GROUP, SOUTH PACIFIC

K. A. R O D G E R S , M. R. G R E G O R Y AND R. P. C O O N E Y

Departments of Geology and Chemistry, University of Auckland, Private Bag, Auckland, New Zealand

(Received 16 November 1988; revised 24 January 1989)

A B S T R A C T : Bayerite forms a cascading, travertine-like deposit over volcanic tufts on Raoul Island in the South Pacific where it is produced from episodic discharge of spent, caustic aluminate solutions from a hydrogen generator. Textural evidence indicates microbial activity has possibly been involved in formation of the hydroxide, as well as sodium carbonate aluminate, present in minor amounts. X-ray powder diffraction and infrared spectral studies failed to give confident identification of the bayerite, the identity being confirmed by laser Raman spectroscopy. A high quality spectrum was obtained extending from 3280 to 3680 cm -1 and from 100 to 1000 cm-'.

Bayerite occurs rarely in nature. It has been recorded in low-grade, thermally metamor- phosed sedimentary rocks from Israel (Bentor, Gross & Heller, 1963; Gross & Heller, 1963), from Hungarian brickclays (Nfiray-Szab6 & P6ter, 1967), and bauxitic soils from China and the USSR (Khorosheva, 1969). On Raoul Island, in the Kermadec Group, it forms the major component of a deposit which coats and permeates volcanic tufts over which discharge from a hydrogen generator spills intermittently.

Raoul Island (29 ~ 14'50" Slat. , 177~ " E long.) is an active stratovolcano (Brothers & Searle, 1970). A New Zealand Meteorological Service station on this island maintains a small, low-pressure hydrogen generator for filling weather balloons, of a design in common use by meteorologists throughout the world. Hydrogen is produced by reaction of aqueous caustic soda with AI filings, the quantity varying with the size of balloon to be filled. A common charge is 680 g A1 to 1100 g of caustic soda in 45-701 of water obtained from a local, groundwater-fed spring. The spent solution, diluted with further water, is flushed from the chamber and spilt over adjacent cliffs formed of dacitic pumice breccias and tufts of the Fleetwood Tephra (Lloyd & Nathan, 1981). A cascading, travertine-like deposit of soft, white bayerite has formed on the rock surface incorporating wind-blown dust and encrusting some vegetation, as well as permeating porous tufflayers, coating particles near the surface of these layers, and cementing scoriaceous basalt and pumice scree (Fig. 1).

P H Y S I C A L P R O P E R T I E S

The bayerite is white to greenish grey with a white streak and forms stalactitic flowstone and multi-lamellar, radiate botryoidal concretionary masses (Fig. 1, insert) reminiscent of bacterial shrubs described from typical calcium carbonate travertines (cf. Chafetz & Folk, 1984, Fig. 24). Cleavage is perfect on ?{001 } with the microcrystalline layers showing a semi- conchoidal fracture developed perpendicular to their surface. S.G. ~ 2.22 g/cm3; H ~ 3.5.

�9 1989 The Mineralogical Society

532 K.A. Rodgers et al.

FIG. 1. Cascading, travertine-like deposit of bayerite, encrusting tufts below the hydrogen generator outfall, Raoul Island. Insert: botryoidal habit of encrustation suggesting bacterial

shrubs. Pack and knife give scale.

Bayerite from South Pacific 533

Fracture surfaces have a vitreous lustre which inclines to resinous when fresh. The encrustation fluoresces white in both short- and long-wave ultra-violet radiation with weathered surfaces showing strongly yellow.

Laminated crusts are several mm to ~ 1 cm thick. Individual concretionary laminae (Fig. 2) seldom exceed 0.02 mm in thickness and consist of very fine-grained, low-birefringence (grey to first-order yellow) microcrystalline radiating tufts and sheaves, the whole appearing somewhat mottled between crossed polarizers. Rare, discontinuous lensoid layers attain a thickness of 0.12 mm in which comparatively coarse crystals are from 0.045 to 0.085 mm long, and up to 0.02 mm across.

FIG. 2. Microphotograph of thin section (AU 40627) through typical bayerite encrustation. Note the well-defined, finely laminate and fenestrate fabric (lower) and irregular lithic-rich layers (top right), as well as the enigmatic globular, flask- and kidney-shaped structures (centre)

of probable biogenic origin. Base of photo is 0.7 mm long.

The fabric is typically fenestrate and grummose textures are generally evident. With the incorporated volcanic detritus (Fig. 2), there are close similarities to the calcareous sinter rinds that Riding (1979, Fig. 17) has illustrated from lacustrine algal bioherms. Darker, probable organic stained bands can be likened to laminae, rich in cyanophyte filaments found in many stromatolites and also in carbonate travertine deposits (Chafetz & Folk, 1984). In thin section, textures occur akin to those of bacterial clumps with diffuse borders as described by Chafetz & Folk (1984, Fig. 22). Enigmatic globular structures (Fig. 2) resemble algal fructifications, sporangia or dasycladacean ampulla (G6not, 1984, pl. 2) and suggest that microbial processes were active during crystal sedimentation.

The laser Raman spectrum of the Raoul mineral gave the distinctive signature of bayerite (Fig. 3). Spectra were recorded using Jobin-Yvon U1000 and $3000 instruments with 514.5 nm Ar § (ca. 150roW) as the exciting line. Microscope sampling and both multichannel (diode-array) and monochannel (RCA 31034A photomultiplier) detectors were employed. Intense Raman lines were observed at 3420, 3541 and 3651 cm -1 identified as v (OH)- stretching modes of the hydroxide by Huneke et al. (1980). These authors also reported that

534

O O o

X

t .d i , i t / )

t / I I - - Z

C) t...$

1.600-

1.200-

O. 800-

K. A. Rodgers et al.

e~ -,,t c:) o3

S 0 00J 33 00 3 200 3 000 800 WAVENUMBER S / Zk(:m -I

FIG. 3. Microprobe laser Raman spectrum of Raoul bayerite from 3280 to 3680 A cm -1.

the lower frequency region of the bayerite spectrum contained only a few weak, broad bands which were generally non-reproducible due to excessive fluorescence and a high background noise level. However, in the present study a good quality spectrum was obtained of a bayerite specimen over the region 200 to 1000 cm -1, and is reproduced in Fig. 4.

The infrared spectrum of the Raoul sample (Fig. 5) closely resembles that of bayerite (cf. Van der Marel & Beutelspacher, 1976). Absorption bands at 3472 and 762 cm -1 appear distinct for bayerite. Bands at 3657, 3550, 3423, 1022, 977, 585,558, and 427 cm -1 are closer to typical bayerite values than those of nordstrandite, while nordstrandite bands at 3490, 3360, 1000, 823, and 702 cm -~ are lacking. Only one or two very weak bands were seen which have been reported in the spectra of gibbsite and/or nordstrandite and which are not coincident with bayerite bands (e.g. 912 cm -1) although scattering maxima which might be assigned to gibbsite were not seen in the high-frequency region of the Raman spectrum.

Separates from botryoidal crust coating pumice, from crustose flowstone on tuff, and from multiple layers from within a tuff horizon all showed the same X-ray powder diffraction pattern containing three extremely strong reflections at 4.79 + 0-1, 4.39 + 0.1, and 2.233 A, and strong reflections at 3.22 + 0.01 and 1-724 A. All other reflections (2.718, 2.542, 2.476, 2.375, 2.171, 2.085, 1-994, 1.653, 1.606, 1-562, 1.411, 1.393, 1.334/~) were weak to very weak.

The presence of very strong reflections close to 4-4 and 2.2 A characterizes bayerite (Schoen & Roberson, 1970) but a peak near 4.8 A is typical of nordstrandite. While the Raoul diffraction pattern as a whole conforms most closely with that for bayerite-b (Yamaguchi & Sakamoto, 1960), it differs in some aspects from that commonly published for this compound. Typically, the first three prominent reflections for bayerite occur close to 4.71, 4.36 and

c~ i,o

1 800-

o 1"600- 0 0

x 1.400-

~ 1-200-

~ 1"000-

Z 0-800-

8 0-600 -

0"~0(

Bayerite from South Pacific 535

tt~ "4' tr)

2doo ~,o'o o

(:)

to

606.0 80'0-0 10'00.0

WAVENUMBERS/z~ cm -~ FIG. 4. Microprobe laser R aman spectrum of Raoul bayerite from 200 to 1000 A cm -1.

Iii C.) Z ,<

8~ 0 tn IX) ,<

.%

0"0"

-0 .1

0"5"

i

3500"0

v~ c~ o~

3oGo.o ~ooo,o ~58oo

WAVENUMBERS / A cm "I

t~

J

i

1000.0

FIG. 5. Infrared absorption spectrum of bayerite, Raoul Island.

50'0- 0

536 K . A . Rodgers et al.

3-20 A (e.g. JCPDS file 20-11), but published values for these first three diffraction lines have ranges of 4.69-4.79, 4-32-4.38, and 3.19-3.21/~ (e.g. Yamaguchi & Sakamoto, 1960; Bentor et al., 1963; Khorosheva, 1969; Sato et al., 1969; Schoen & Roberson, 1970; ASTM 8-96, 20-11). Many, but not all, of the lower values in each range occur amongst pure synthetic samples and there seems to be some variation in the crystal chemistry of bayerite e.g. Van der Marel & Beutelspacher (1976, p. 194) referred to the mineral as 'not of constant, well-defined structure' but failed to elaborate.

C H E M I S T R Y

Spot electron microprobe partial analyses of the crust (e.g. Table 1, cols. 1-3) show minor and quite variable amounts of silica, lime, and soda and traces of potash and chlorine. While some Na and Si may be incorporated in the bayerite structure (e.g. Wefers, 1962), silt- and clay-sized inclusions of vitric and lithic tufts are dusted throughout the samples, being concentrated towards the top of each aluminium hydroxide layer and often acting as nuclei about which a concretion has developed. Several such particles had compositions close to labradorite feldspar, olivine and pyroxene.

Samples of a sinter-like coating on tuff and from layers within cemented scree showed, in addition to the XRD pattern of bayerite, a pattern which conforms closely to that of Na2AI202(CO3)2.2"9H20 (JCPDS file 34-547). Strongest reflections were at 5.73 and 2-80 A, moderate lines at 3.38, 2.25, 2.16 and 2.708 A, and weak lines at 2.62, 2.51 and 1-87/k. Other reflections noted by JCPDS for this substance are coincident with those of bayerite. Spot microprobe partial analyses of a polished thin section of the cemented scree showed discontinuous layers of subparallel microcrystallites with Na : A1 atom ratios close to 1 : 1 (e.g. Table 1, col. 4). A1203 ranged from 42.41 to 49-96 wt~, and Na20 from 25.07 to 31.62 wt~. Broken surfaces from this same tuff examined under a scanning electron microscope showed hollow clumps, clusters and tubules of radiating crystallites, possibly originating as envelopes around gas bubbles or encrustations on organic threads (Fig. 6). The habit of these conformed with layers probed in thin section whose qualitative EDAX spectra showed high levels of Na and A1.

TABLE 1. Representative partial microprobe spot analy- ses of alumina-rich matrix and sodium aluminate

carbonate, Raoul Island.

1 2 3 4

SiO2 1.67 1.04 0.91 N.D. A1203 65.88 62.13 60.45 45.81 CaO 0-16 0.25 0.04 0.45 Na20 2.81 0.30 N.D. 27"76 K 2 0 N.D. N.D. N.D. N.D.

C1- 0.02 0.08 0.06 0.46 Total 70.54 63.80 61.46 73.85

1-3 Botryoidal bayerite on pumice, AU 40627 4 Cemented scree coated with flowstone, AU 40624 N.D. - - not detected.

Bayerite from South Pacific 537

FIO. 6. Scanning electron microscope photograph of Na- and Al-rich crystal clusters associated with bayerite, Raoul Island. Scale bars are 10 #m long.

D I S C U S S I O N

fl-Al(OH)3 can be produced by ageing of aluminate gels (e.g. Elderfield & Hem, 1973) and by streaming of sodium aluminate liquor with CO2 (e.g. Sato et aL, 1969). Elderfield & Hem (1973) pointed to alkaline solutions in which the aluminate ion, AI(OH)2, is the dominant AI species as favouring formation of bayerite, the strong polarizing power of the A1 ion being balanced by the excess negative charge. During ageing, aluminate ions are added to bayerite through a protonation mechanism:

AI(OH)3 + AI(OH)2 + H + ~ 2Al(OH)3 + H20

The conditions under which the various polymorphs form are quite critical, bayerite form- ing between pH 6.7 and 11. Above pH 11 the product is nordstrandite (Franz, 1975) but laboratory studies have sometimes produced conflicting results. The manner of crystalliza- tion of intermediate gels seems to depend, at least in part, on relative reaction rates, while some reactions have been found to be slowly reversible e.g. see discussions of Hsu & Bates (1964) and McHardy & Thomson (1971). Before manufacturing techniques were refined prior to 1977, manufactured aluminium hydroxides were frequently a mixture of the ct- and fl- polymorphs.

Such considerations highlight the monopolymorphic nature of the Raoul deposit and point to formation under a restricted range of conditions, although the precise conditions are not known. The sole relevant parameter is pH measured in the waste sump of a disused low- pressure generator on Funafuti. Sediment dredged from the sump had a pH of 10.19 and contained globular bodies of alumina gel. Whether ageing or carbonation has aided formation of the Raoul deposit is unclear. The chemistry of the spilt solution could be consistent with ageing (cf. Elderfield & Hem, 1973) but the presence of the carbonate sodium aluminate suggests CO2 was present, a possible reaction being:

2Na + + 3Al(OH)2 + H + + 2CO 2 ~ AI(OH)3 + Na2AI202(CO3)2 + 5H20

In conclusion, it is perhaps pertinent to draw on the continuing debate over the relative significance of physicochemical and biochemical processes in the formation of carbonate travertine deposits (e.g. Chafetz & Folk, 1984; Emeis et al., 1987) where it is agreed that

538 K . A . Rodgers et al.

photosynthesis regulates the CO2 conten t in fresh waters and hence indirect ly helps regulate

p rec ip i ta t ion o f calcite. In s imilar fashion, photosynthes is may have cont r ibu ted to the

prec ip i ta t ion of sod ium carbona te a lumina te and hence possibly bayer i te at Raoul Island,

where layering in the depos i t reflects the episodic d ischarge of spent waters f rom the

hydrogen generator .

A C K N O W L E D G M E N T S

Field work and most equipment used were funded by the University of Auckland Leave and Research Committees. Microprobe laser Raman spectra were recorded at the Jobin-Yvon Division of ISA Instruments (France) and we are grateful to Madame H. Boyer for her assistance. The visit to Raoul Island was by permit from the Department of Conservation and with logistic support from R.V. Vulkanolog. M.R.G. is indebted to fellow expeditioners from the Institute of Volcanology, Petropavlosk, Kamchatsky. Information on the hydrogen generating plant came from J. S. Falconer of the New Zealand Meterorological Service with D. Roberts permitting sampling of the Funafuti sump. Technical assistance was provided by Prof. M. Taylor, Dr R. Sims, S. Courtney, N. Howett and D. Stringer.

R E F E R E N C E S

BENTOR Y.K., GROSS S. & HELLER L. (1963) Some unusual minerals from the "mottled zone" complex, Israel. Am. Miner. 68, 924-930.

BROTHERS R.N. & SEARLE E.J. (1970) The geology of Raoul Island, Kermadec Group, southwest Pacific. Bull. Vol. 34, 7-37.

CHAFETZ H.S. & FOLK R.L. (1984) Travertines: depositional morphology and the bacterially constructed constitutents. J. Sed. Pet. 54, 289-316.

ELDERFIELD H. & HEM J.n. (1973) The development of crystalline structure in aluminium hydroxide polymorphs on ageing. Mineral. Mag. 39, 89-96.

EMEIS K.-C., RICHNOW H.-H. & KEMPE S. (1987) Travertine formation in Plitvice National Park, Yugoslavia: chemical versus biological control. Sedimentology 34, 595-609.

FRANZ E.-D. (1975) Das System A1203-H20 bei 25~ und 1 atm Druck im pH-Bereich 5-13. N. Jb. Miner. Abh. 125, 80-90.

GI!NOT P. (1984) Calcification in fossil Neomereae (Dayscladales). Pp. 264-272 in: Paleoalgology (D.F. Tomey & M.H. Nitecki, editors). Springer, Berlin.

GROSS S. & HELLER L. (1963) A natural occurrence of bayerite. Mineral Mag. 32, 723-724. Hsu P.H. & BATES T.F. (1964) Formation of X-ray amorphous and crystalline aluminium hydroxides. Mineral.

Mag. 33, 749-768. HUNEKE J.T., CRAMER R.E., ALVAREZ R. & EL-SWALFY S.A. (1980) The identification of gibbsite and bayerite

by laser Raman spectroscopy. J. Soil Sci. Soc. Am. 44, 131 134. KHOROSHEVA D.P. (1969) Bayerite from the bauxite horizon of the middle Dnieper region. Dokl. Akad. Nauk.

SSSR Earth Sci. 182, 123-126. LLOYD E.F. & NATHAN S. (1981) Geology and tephrochronology of Raoul Island, Kermadec group, New

Zealand. N.Z. Geol. Surv. Bull. 95, 1-105. MCHARDY W.J. & THOMSON A.P. (1971) Conditions for the formation of bayerite and gibbsite. Mineral. Mag.

38, 358-368. NARAY-SZABO I. & P~TER 1~. (1967) Nachweis von Nordstrandit und Bayerit in ungarischen Ziegeltonen. Acta

Geol. Acad. Sci. Hung. 11, 375-377. RIDING R. (1979) Origin and diagenesis of lacustrine algal bioherms at the margin of the Ries Crater, Upper

Miocene, southern Germany. Sedimentology 26, 645-680. SATO T., YAMASHITA T. & OZAWA F. (1969). The preparation of bayerite. Zeit. A norg. Allg. Chem. 370, 202-208. SCHOEN R. & ROBERSON C.E. (1970). Structures of aluminium hydroxide and geochemical implications. Am.

Miner. 55, 43-77. VAN DER MAREL H.W. & BEUTELSPACHER H. (1976) Atlas of Infrared Spectroscopy of Clay Minerals and their

Admixtures. Elsevier, Amsterdam. WEFERS K. (1962) Zur struktur der Aluminiumtrihydroxide. Naturw. 49, 204-205. YAMAGUCHI G. & SAKAMOTO K. (1960) The identity of bayerite-a and bayerite-b. Can. J. Chem. 38, 1395-1396.