STRUCTURES OF ALUMINUM HYDROXIDE.pdf

8/12/2019 STRUCTURES OF ALUMINUM HYDROXIDE.pdf

1/35

THE AMERICAN MINERALOGIST, VOL. 55, JANUARY-FEBRUARY, i97O

STRUCTURES OF ALUMINUM HYDROXIDE ANDGEOCHEMICAL IMPLICATIONS1Rononr ScnoBNaNo Cuenr.BsE. RonBnsoN, U.S. Geo-logical Suntey, Menlo Park, CaliJornia 94025.

AgsrnacrSynthesis experiments in the alumina-water system at room temperature indicate thatthe gibbsite polymorph precipitates slowly from solutions whose pH is below the point ofminimum solubility (pH 5.8), and the bayerite polymorph precipitates rapidly from solu-tions whose pH is above 5.8. Nordstrandite, the third polymorph of aluminum hydroxide,forms from bayerite during aging at intermediate to high pH values. fn solutions of inter-mediate pH, both gibbsite and bayerite form, but with aging, early-formed gibbsite disap-

pears as more bayerite forms During aging, the pH's of the mother liquors decrease ifgibbsite precipitates and increase if bayerite precipitates.The principal structural difierence among the three aluminum hydroxide polyrnorphsis in the mode of stacking successive ayers. This, in turn, is controlled by the shape of,

and charge distribution on, hydroxyl ions located at the surfaces of the layers. The extentof polarization by the aluminum cations determines whether the hydroxyl ions will possesscy'lindrical symmetry as n bayerite or tetrahedral symmetry as in gibbsite. We suggest thatthe extent of polarization of the hydroxyl ions within the solid phases is inherited from thepolarization attained by the hydroxyl ions in aqueous aluminum complexes in the motherliquors.The usual abundance of the silicate ion in natural alkaline environments may explain

the scarcity of nordstrandite, for the silicate ion may favor the precipitation of alumino-silicate minerals rather than aluminum hydroxide. Near absence of bayerite in naturemay indicate that it is metastable and will invert to nordstrandite in alkaline solutions.

INrnoluctroNThe role piayed by aluminum during weathering of the earth's crustis an enigma. The claim by Rankama and Sahama (1950) that "Thecycle of aluminum is simple, and its details are well known." is no longeradequate. As our understanding of the geochemistry of other common

elements mproves, gaps n our knowledge of the geochemistry of alumi-num stand out.For example, the presence of abundant authigenic aluminosilicateminerals in sedimentary rocks means that large quantities of the con-stituent elementswere in solution. Aluminum is one of the few elementswhose solubility is so low in most natural waters as to appear incapableof taking part in the formation of vast amounts of authigenic alumino-silicates. Similarly, the synthesisof aluminosilicates n the laboratory atroom temperature and under conditions resembling natural environ-ments is not easily attained. Bauxite deposits, the primary ore of alu-minum, form by residual concentration of hydrous aluminum oxidesduring weathering of aluminosilicate minerals. We infer that silica andother cations, with the exceptionof aluminum, go into solution. But thisI Publication authorized by the Director, U. S. Geological Survey.

43


2/35

ROBERT SCHOEN AND CHARLES E, ROBERSONmechanism fails to explain bauxite depositsformed on exceedingly ow-aluminous rocks. Nor does the common presenceof concretionary piso-Iites in bauxite deposits strengthen our conviction in the absolute im-mobility of aluminum during weathering. The dogma of aluminum's lowsolubility at intermediate pH values leads biologists to propose thataluminum is nonessential n life processesn spite of its wide distributionin organisms (Hutchinson, 1945).

Our principal gaps in understanding the geochemistry of aluminumarise rom the lack of detailed knowledge of the controls on solubility aswell as the kinds and amounts of substancesn solution. In addition, thealuminous solids that precipitate from supersaturatedsolutions must beadequately characterized. A fuller knowledge of the geochemistry ofaluminum will help clarify the origins of the most abundant minerals incrustal rocks, as well as form an acceptable heory of genesis or bauxiteore deposits, and perhaps even shed Iight on critical economic problemssurrounding the formation of alumina catalysts.

Recent studies of the alumina-water system at room temperature bythe U. S. GeologicalSurvey aim to clarify and fill some of these gaps inknowledge.This paper, an outgrowth of the studies by Hem and Rober-son (1967) and Hem (1968), and of their continuing research nto thenature of aluminum in natural waters, describes he problems of identi-fi.cationof small amounts of fine-grained synthetic aluminum hydroxideprecipitates formed in dilute solutions. The results of our X-ray diffrac-tion study provide a basis for suggesting a mechanism controlling theformation of various polymorphs of aluminum hydroxide. An explanationof the distribution in nature of the minerals gibbsite, nordstrandite, andbayerite follows as a consequence f the application of this mechanism.

Puon Svnrnnsrs rurpsThere has been extensive research for many years on factors that control the formation

of compounds in the alumina-water system in the laboratory. We give here only a summaryof the most recent work as background for the part of this paper dealing with synthesis ofaluminum hydroxide.

A chapter by Rooksby (1961) on the oxides and hydroxides of aluminum and iron sum-marizes earlier work on the alumina-water system. According to Rooksby, gibbsite(AI(OH)3) is the only naturalll' occurring pol1'morph of aluminum hydroxide as, at thattime bayerite (AI(OH)3) and the unusual form nordstrandite (Al(OH)a) were known only asproducts of synthesis. A year later, Hathaway and Scblanger (1962) reported nordstranditeas a mineral from Guam, and Wall and others (1962) from Sarawak (Borneo). And in1963, Bentor and others described the first natural occurrence of bayerite corroborated byX-ray difiraction. Regarding conditions of synthesis, Rooksby stated that the slow acidifi-cation with COz of a sodium-aluminate solution results in precipitation of aluminumhydroxide with the gibbsite crystal structure iI carried out at temperatures near 100'C andresults in the bayerite clystal structure if the temperatule is nearer room temperature.Agrng of this bayerite, however, may convert it to gibbsite. He believed that nordstrandite


3/35


4/35

46 ROBERT SCHOEN AND CHARLES E. ROBERSONThis last ionic species,AI6(OH)IzF, constitutes a hexagonal closed ring. Further polymeriza-tion can yield higher polymers such as A n(OH)zz8+ or Alrs(OH)36e+.But, as the positivecharge on the poly'rner increases, further poly'rner'uation slows down because of eiectricalrepulsion. Only when sufficient hydroxyl ions form to discharge the polymers, will theycluster and form crystalline AI(OH)a.

Hem and Roberson (1967) proposed a somewhat difierent mechanism to provide anexplanation of the observed behavior of aluminum hydroxide precipitates at all pH values.At pH values two or more units below the point of minimum aluminum solubility (aboutpH 5.8), the aluminum ion is coordinated by six polar water molecules. The strong positivecharge of the aluminum ion repels the two protons of each water molecule and, as pH in-creases,eventually drives off a proton forming the monomeric complex ion AI(OH) (OHtb'+.At about pH 5, this complex ion and hydrated aluminum are in equal abundance withmore of the hydrated aluminum at lower pH values and more of the complex ions at higherpH values. Two of these complex ions may join by losing two water molecules to form adimer:

2A1(OH)OHt * : er,1On),(oH,);+ 2H,O.Furttrer deprotonation, dehydration, and polymerization of monomers and dimers yields aring structure of six octahedrally coordinated aluminum ions with the formula:

A16(OII)rr(OHr)16l'Coalescence of rings into layers by further growth and stacking of layers results in theformation of gibbsite (AI(OH)B). Above the point of minimum solubility, precipitation ofbayerite must involve the anion Al(OH;ot-, generally considered to be the dominant alu-minum species n alklaine solutions. The reaction may be:

Al(oH)i * H'* : Al(oH)3+ H,o.Reading this reaction to the right, crystalline Al(OH)s (bayerite) precipitates u'ith an ac-companying increase in pH of the solution.

This summary of some of the previous work on the synthesis of aluminum hydroxidesshows the confusion that exists, especially in the earlier work, regarding the environmentin which the various solids precipitate as well as the identity of these solids. As Hsu (1967)points out, conflict and confusion in the literature on the alumina-water system stems fromthe metastability of intermediate products and siow reversibility of reactions. In addition,we would mention tle problem of incomplete identification of solid products. This paper isconcerned primarily with clarifying the identification of synthetic aluminum hydroxideprecipitates.

Mnrnon or SYNTTTESTsWe prepared most of the synthetic aluminum hydroxides described in this report by

mixing three aqueous solutions. Solution I contained ions of hydrogen, perchlorate, sodium,and aluminum (4.53X10-4 molar). Solution II contained ions of sodium, perchlorate, andhydroxyl (6.53X10-s molar). Solution III contained ions of sodium, perchlorate, and alu-minum (9.06X 10-a molar). The desired amount of solution II (20.0 to 36.5 ml) was added to25.0 ml of solution I in a polyethylene bottle with vigorous stirring. An amount of solutionIII equal in volume to solution II was added at the same time to maintain the concentra-tion of aluminum at 4.53X10-a molar. The rate of addition was about 1 ml per sec. Wechose the perchlorate anion for this study because of its freedom from a tendency to com-plex with aluminum. The perchlorate concentration was 0.01 molar in all solutions in order


5/35

STRUCTURES OF ALUMINUM HYDROXIDE 47to maintain a relatively constant low ionic strength (and, therefore, a constant activitycoefficient for reacting ions).

The mixed solutions were stored in a COz-free atmosphere at2soc+2o for periods rang-ing from 16 hours to 10 days. At the conclusion of aging, we measured the pH of the solu-tions with glass and calomel electrodes and then filtered the solutions through plastic mem-brane (Millipore) filters having pore diameters of 0.45 pm (in a few experiments 0.10 pmfilters were used). Immediately after filtration, we affixed the piastic membrane filter to aglass slide with rubber cement and X-rayed it with a Norelco or Picker diffractometer andcopper Ka radiation.

The amount of precipitate often was small (-5 mg) and the X-ray diffractograms were,therefore, incomplete. In order to make positive identification, we prepared larger amountsof solution which yielded as much as 50 mg of precipitate. This was then filtered, dried atroom temperature, scraped off the filter, ground lightly, and X-rayed in a standard alu-minum holder to provide a complete X-ray difiractogram with reduced preferred orienta-tion. Samples for electron microscopy were prepared in a standard manner (Kay, 1965) byallowing a crop of mother liquor, containing some precipitate, to dry on Formvar plasticfilm and using carbon shadowing.

We made all our measurement of X-ray peak intensity on peak height above back-ground rather than on total area. Where the half-widths of X-ray peaks difiered consider-ably, we made no attempt to compare reported intensities quantitatively.We did not chemically analyze any of the aluminum hydroxide precipitates describedin this report. The matching of X-ray diffraction peak positions and intensities between

standards and precipitates was considered adequate for identificat ion. We are aware ofreports proposing that small amounts (usually less than one percent) of sodium and potas-sium found in aluminum hydroxide are necessary to stabilize the lattice (Saalfeld and others,1968). The crystallochemical dificulties of incorporating sodium or potassium ions into thealuminum hydroxide lattice, and the kinetic efiects of other foreign ions, mentioned else-where in this paper, lead us to believe that purity of precipitate is cf no consequence to thisstudy.

Rrsur-rs AND INTERPRETATToNTable 1 lists the results of the X-ray study of the synthetic aluminumhydroxides. They were prepared under a variety of conditions, and the

sample identifier listed in the first column of Table 1 can be used todetermine what these conditions were. The starting pH (related to theamount of hydroxyl-containing solution II added) varied from acid insamples with Arabic numeral 1 to alka line in samples with Arabic nu-meral 5. The Ietters A, B, and C in the sample identifi.er refer to agingperiods of 16 hours, 2 days, and 10 days, respectively. In order to assessthe reproducibility of the results, we prepared a secondgroup of sampleswith almost identical starting pH values. Roman numeral II identifiesthis secondgroup of samples,and Roman numeral I the first group.Column 2 of Table 1 lists the ratio of the hydroxyl ions bound toaluminum, to the total aluminum present. This representsan averagefor the number of hydroxyl ions bound to each aluminum ion in thesystem, including both aluminum in solution and in the precipitate. Aratio of 3.00 indicates that just enough hydroxyl ions were added to


6/35


7/35

STRUCTURES OF ALU A,IINUM HYDROXIDE

Fro. 1. X-ray difiractograms of tlpical synthetic products developed during short-termaging in tle alumina-water system (A, B, C, D) and Al(OH)s standards (E, F, G).

49

lBcq

iqR,ffiiIIIElI

IFiIiclrlu 112 ?eq ees'[ 2


8/35

50 ROBERT SCHOEN AND CHARLES E. ROBERSONTerr.r 1. X-ne.v DrrlnectoN Rrsutts ox SvNrrrnnc AI(OH)3

Sampleldentifier

1 A IB IC I

Age

16 hrs2 d,ays10 days

l'inalpH4 . 7 34 . 6 64 . 5 6

Spacing and Intensityd G ) I d ( b r o . j 0 ,

1 A I IB I IC I I

2 . 3 22 . 3 22 . 3 2

16hrs2 days10 da1'g

4 . 9 44 . 8 34 . 7 0

4 . 70 20 4 . 37 24

2 I t 7B IC I2 . 7 92 . 7 92 7 9

16 hrs2 days10 days5 . 2 85 . 1 74 . 8 3 4.90 28 4.40 24

2 A I IB I IC I I2 . 7 92 . 7 92 . 7 9

16 hrs2 d,ays10 days< a A5 .1 05 . 0 2 4.82 20

3 A IB IC I2 . 9 02 . 9 02 . 9 0

16hrs2 days10days6 . 1 55 936.95 4.88 130 4.37 60

3 . A ' I IB I IC I I2.902 .902 .90

16hrs2 days10days5 8sq 6 06.45 4 . 8 7 3 6 4 . N 2 0

3C II-1 2 . 90 10days 6 .+5 4.90 76 4 42 524 A I

B IC I3 .1 53 .1 5J . I J

16hrs2 days10days

8 8 79 . 0 18 95

4.77 28 4.40 80 2.23 244 . 7 7 14 4 . 3 6 188 2 . 2 2 3 64.M 160 2.2+ 28

4 A I IB I IC I I3 .003 .003.00

16hrs2 days10 days6 . 3 76.498.48 4.42 r+0 2.23 M

5 A IB IC I3 . 5 2? ( ,3 . 5 2

16hrs2 days10 days9 . 4 89 . 6 99.54

4.7s 20 4 37 I20 2.23 364.77 16 4.37 24O+ .23 48+.37 280+ 2.22 485 A I IB I IC I I

3 . 5 2s . 5 23 . 5 216 hrs2 days10days

9 . 2 39 . 4 79.644 . 37 48 2 . 23 164.40 r52 2.23 324.37 232+ 2.22 40


9/35

STRUCTURESOF ALUMINUM HVDROXIDE 51crystal may have an alkaline effect on the pH of the solution. Becausesurface area per unit volume decreases uring crystal growth, hydroxylions in the crystals should exert a decreasingeffect on the pH of thesolution as the crystals grow larger. Other Iikely explanations for theobserved downward pH drift are the deprotonation reaction duringpolymer growth as well as the dissociation of water proposed by Hsu(1e66).

The preceding hypotheses, however, do not explain the upward pHdrift o{ solutions containing bayerite. Hydroxyl ions on the surface ofbayerite crystals should have little efiect on the pH because he crystalsgrow rapidly in the alkaline environment and quickly achieve arge size.X-ray diffraction and electron microscopy indicate that bayerite crys-tallites attain sizes n excessof 1,000 A within 24 hours. Instead, theobservedupward drift in pH probably is due to the reaction:

Al(oH)3 + Al(oH)l- : 2A1(OH)3+ (OH)'-in which aluminate ions are added to bayerite crystals as slow growthcontinues during aging.All samples produced some very broad X-ray peaks. There was atendency for a broad peak at about 9 A to migrate toward 7 A with in-creasingcrystallinity of the sample. This broad maximum may indicatethe incipient formation of pseudoboehmite,a monohydrate of aluminum.Another broad peak near 4 A tended to migrate toward 3 A in morecrystalline samples,and may also be a pseudoboehmitepeak.Filtration of sample 3CII through a 0.45 trrmmembrane produced anopalescent iltrate so we refiltered the filtrate through a 0.10 pm plasticmembrane filter. Listed in Table 1 under identifier 3CII-1 are the resultsof the X-ray difiraction study of this 0.45 pm-0.10 pm size material.Comparisonwith the data l isted for sample 3CII indicates thatmuchcrystalline material was less than 0.45 pm and therefore was missed nthe study of other samples isted in Table 1. Subsequentstudies using0.10 pm plastic membranefilters, however, did not result in any changesin the identification of phases.Figure 2 depicts the results of one of the later studies using 0.10 pmplastic membrane fi.lters.Samplesprepared at (OH)s to Al ratios oI 2.7,2.8, and 2.9, and aged for periods of 7 days, 7 days, and 8 days, respec-tively, gave inal pH valuesof the mother l iquors ol 4.74,4.80,and 4.86,respectively. Figure 2, A and B show no crystalline material by X-raydiffraction. Figure 2C, ol material prepared with an (OU)u to Al ratioof 2.9, exhibits a broad peak at 4.8 A identified as gibbsite by otherlong-term aging experiments. X-ray oI the 2.9 ratio sample t hour afterfiltration (Figure 2D) shows a large increase n intensity of the broad


10/35

52 ROBERT SCHOEN AND CHARLES E. ROBERSON

AIIIII

2 -7 f resh I

D e g r e e s 2 e ( C u K c )FIc. 2. X-ray diffractograms showing the development of synthetic gibbsite during aging

and drying as a function of tle ratio (OH)B/AI.


11/35

STRUCTURES OF ALUMINUM HYDROXIDE 53gibbsite peak. X-ray ol the 2.7 2.8, and 2.9 ratio samples after 2 days,2 days,and 1 day, respectively,of drying on the plastic membrane filtersresulted in Figure 2, E, F, and G respectively. These diffractogramsindicate that gibbsite formed in all solutions though drying was neces-sary to make the gibbsite in the 2.7 and 2.8 ratio samplesdetectable byX-ray diffraction. Apparently, many low pH solutions prepared earliercontained gibbsite that was too fine to be trapped on 0.45 prm filters.This may account for the erratic distribution of gibbsite peaks reportedin the upper half of Table 1.The diffractograms presented n Figure 2 reveal several other thingsabout the gibbsite formed. Diffraction from the (002) or basal planesofthe gibbsite crystals causes he single X-ray peak. The great breadth ofthis peak indicates that the crystallites are relatively small in the basal(crystallographic c-axis) direction. Calculation of this dimension usingthe Scherrer ormula (Klug and Alexander, 1954,p. 512) results in 45 Afor samples 2.7 and, 2.8, and 80 A for sample 2.9. Measurements ofcrystallite shadowson electron micrographs corroborate thesedimensions(Ross Smith, oral commun., 1968). But at least one dimension of thesecrystallites must exceed 1,000 A because hey were trapped on 0.1 pmfilters. Figure 3 shows a hypothetical gibbsite crystal possessing hesedimensionsand the hexagonal form seen n electronmicrographssuch asFigure 4. Apparently, the growth habit of gibbsite favors the formationof thin platelets, whose upper and lower surfaces are parallel to thebasal planesresponsible or the observedX-ray peak. According to VanNordstrand and others (1956,p. 7t2), ". . . gibbsite usually grows asplatelets parallel to the layers of its layer lattice." (basalplanes).

Ftc.3. Sketchof hypotheticalgibbsitecrystal formed during synthesis.


12/35


13/35

STRUCTURES OF ALUMINUM IIYDROXIDE 55the crystal planes responsible for these three difiractions. The 4.35 Apeak originates from (100) prism planes in the crystal and the 2.22 hpeak orginates rom (111) pyramid planes.The 4.71A peak originatesfrom (001) basal planes comparable to those responsible or the strong4.85 A peak in gibbsite. Absenceof rhe 4.71 A (001) peak in newly-formed bayerite implies a poor development of basal planes n the crys-tallites. Crystallites with well-developed(100) prism facesand lesswell-developed (111) pyramid faces might look like Figure 5. This agreeswith the reported habit of bayerite as .'. . . rods or tapered rods the longdirection of which appears roughly perpendicular to the layers of thelayer lattice." (basal planes) (Van Nordstrand and others, 1956,p. 714).Results of electron diffraction by Lippens (1961,p. 69-70) support thisperpendicularity between the basal layers and the longitudinal directionof the bayerite particles. An electronmicrograph of bayerite aged for 10days (Fig. 6) proves the generalcorrectnessof this concept of the shapeof newly-formed bayerite crystallites.

Frc. 5. Sketch of hypothetical bayerite crystal formed during synthesis.


14/35

JO ROBERT SCHOEN AND CEARLES D. ROBERSON

Frc. 6. Electronmicrographf synthetic ayerite rystals,probablyncludinga fewnordstrandite rystals.There are at least three reasonswhy the intensities of X-ray peaksfrom newly-formed bayerite should not match those of aged bayerite.First, the particle shape, which is elongated parallel to the c-axis, tendsto produce only (hk0) diffractions because he basal planes at right an-gles o the c-axisare small initially. But after ten days aging (Fig. 6), thebasal planes are large enough to produce sharp X-ray diffractions. Ab-

senceof thesepeaks n our 10-day old samplesmust be a consequence fthe elongatedparticle shapeand the resultant preferred orientation oc-curring during filtration. Basal planes of bayerite crystallites on the plas-tic membrane fiIters, though large enough'are not in a favorablepositionto diffract X-rays. Figure 1F, the difiractogram of bayerite scraped romthe filter and mounted to minimize preferred orientation, exhibits allpeaks with proper intensities. Finally, the small amount of aluminumhydroxide precipitate deposited on the plastic membrane f,lter is notsufficient to satisfy the X-ray requirements or a sampleof infinite thick-ness(Alexander and Klug, 1948).X-raypeaks at high two-theta angles,therefore, are especia llysubject to diminution of intensity with respectto lower angle peaks. This explains the relative weaknessof the 2.2 Apeak.

The growth of gibbsite n acid solution and bayerite in alkaline solution


15/35

STRACTURES OF ALUMINUM HYDROXIDE 57occurs by the addition of units of aluminum hydroxide. These unitsattach themselves at a greater rate to the edgesof gibbsite crystallitesthan to the top and bottom surfaces.The result is that the slowestgrow-ing faces (that is, the top and bottom surfaces) eventually form theprominent crystal facesof gibbsite. Conversely, he aluminum hydroxideunits attach themselves preferentially to the basal planes of bayeritecrystallites with the result that the slower growing side faces (prism andpyramid) eventually form the dominant crystal faces.To complete our understanding of the crystal form of the three alu-minum hydroxide polymorphs, we must examine the electron microscopeand e lectron diffraction data of Lippens (1961,p. 73-74) on nordstrand-ite. He, as well as Hauschild (1963, Fig. 2) found that nordstranditecrystallites take the form of long rectanglesor more rarely long para llelo-grams. The few small rectanglesand parallelograms n Figure 6 corre-spond closely to those figured by Lippens and Hauschild and may benordstrandite.

The cross section of nordstrandite crystallites is hexagonal, and, as inbayerite, the c-axis parallels the long direction of the crystallites. Butthe D-axisalways appears to coincide with the width of the crystallitesindicating a relatively poorer development of the o-axis direction. Lip-pens found the o-axis direction to range between one-half and one-sixththe length of the D-axisdirection. Figure 7 is a sketch of a hypotheticalnordstrandite crystallite embodying thesecharacteristics.

b - a x i s

Frc. 7. Sketch of hlpothetical nord-strandite crystal formed during s1'nthesis.

c - a x t s


16/35

58 ROBERT SCHOEN AND CHARLES E. ROBERSONNone of the samplesdescribed hus far showedX-ray evidenceof con-

taining nordstrandite, the third polymorph of aluminum hydroxide. Asalready described, much confusion exists in the literature regardingconditions of synthesis of nordstrandite. Ideal pH conditions seem torange from "slightly acid to neutral" (Barnhisel and Rich, 1965) to "pH13" (Pap6eand others,1958).

To help resolve these nconsistencies, r at least to provide additionaldata, we prepared a series of four solutions for long-term aging whoseinitial pH ranged from 9.46 to 12.0.Concentrations of ions were greaterin these solutions than in those previously described,and in one solutionpotassium and sulfate were present for use in another experiment.

We had hoped to fi,nd the aluminum hydroxide polymorphs bayeriteand nordstrandite restricted to well-definedpH regions. Or, failing this,we hoped to discern a distinct increase n the amount of one phase anddecrease n the amount of the other during aging. The X-ray results, todate, are equivocal. Pure bayerite without admixed nordstrandite formedinitially in the pH 12.0 solution. But within 90 days aging, X-ray detect-able nordstrandite formed as the pH increased o 12.82.The other threeless alkaline solutions produced bayerite admixed with nordstranditealmost from the start. Nordstrandite X-ray peaks ncreased n intensityduring two and one-half years of aging, but so did the bayerite peaks.To date, peak ratios show no distinct trend.In contrast to these results, an electronmicrograph (Fig. 8) of materialsimilar to that shown n Figure 6, but aged or two years, showsa definiteincrease in the size of nordstrandite crystallites relative to bayerite.Figure 8 also shows rounding of the edges of bayerite crystallites andthe development of etch pits parallel to the basal cleavage.These eaturesindicate resorption of the bayerite into the mother liquor.

These results imply that slightly alkaline environments promote theprecipitation of nordstrandite and strongly alkaline environments theprecipitation of bayerite. But bayerite is unstable with respect tonordstrandite in alkaline solutions.

Srnucrune on Ar-uurxuu Hvonoxrpe Por-vuonpnsThe preceding experimental results corroborate the observations ofBarnhisel and Rich (1965) that acid environments favor the crystalliza-

tion of gibbsite, neutral environments nordstrandite, and alkaline en-vironments bayerite. In order to understand why the pH of the aqueousenvironment controls the structure of the aluminum hydroxide poly-morph, we must review what is known about the detailed structures ofthe three phases.


17/35


18/35

60 ROBERT SCHOEN AND CHARLES E. ROBERSON

S et at 1t3;2/3@ at at 1/6;5/6Fre. 9. Diagrammatic representation of

idealized gibbsite layer structure pro-jected on a plane perpendicular to the o-axis (from Megaw, 1934).

Incomplete filling of the interstices of the gibbsite layer results in the"holes" being less electrostatically positive than surrounding portionsof the cation sheet.This asymmetry of charge distribution has an affecton the orientation of the oxygen-proton bond within each hydroxyl ion.The protons, instead of being oriented straight up away from the cationsheet, end to be displaced oward the neutral holes.

In addition, the very high polarizing power of the aluminum ion pro-duces a more signifi.cant distortion of the hydroxyl ions. Bernal andMegaw (1935) proposed that this polarization produces a hydroxyl ionwith tetrahedral symmetry analogous o the water molecule.Two alumi-num ions bind two of the negatively chargedcornersof the four-corneredion. The remaining negatively charged corner and the positively chargedcorner containing the proton are available for the formation of ionicbonds with other hydroxyl ions, both within the gibbsite layer and be-tween layers. A distribution of both positively and negatively chargedcornersof tetrahedral hydroxyl ions on the surfacesof the gibbsite layersfavors the buildup of superposedgibbsite layers. A bond forms betweenthe negatively charged corner of one hydroxyl ion and the positivelycharged corner of a hydroxyl ion in the superposed ayer. There is oneso-calledhydroxyl bond for eachhydroxyl ion.

TIIIIo\o)l lGoIIII-iL-


19/35

STRUCTURES OF ALUMINAM HYDROXIDE 61The constraints imposed by this type of bonding result in the hy-

droxyl ions of each superposed ayer of gibbsite lying directly on top ofthe hydroxyl ions of the Iayer below in a configuration resembling open-packing (Fig.9). The effective radius of hydroxyl ions in one layer tohydroxyl ions in a superposedayer (that is, B-B or A-A in Fig. 9) isonly 1.39 A, close to the 1.34 A effective radius between hydroxyl ionsand the cations (Table 2). This indicates that distortion of the hydroxyl

Tastn 2. Ellrcrrvn INrsnroNrc DlsraNcns tN Grnrsnr lrro Bnucrrr(from Bernal and Megaw, 1935)Efiective hydroxylradius toward hydroxylin the next layer

Effective hydroxylradius toward cation

1 . 341 . 39 1 . 3 e1 . 6 1

ion by polarization produces an "open-packing" in which the centers ofions are almost as close as in true close-packing.This stacking arrange-ment of double-sheetgibbsite layers can be describedby the symbol:

AB I ee I nn andso orthl lwhere A and B denotehydroxyl sheets n different positions as shown inFigure 8, and the vertical dashes represent the boundaries betweendouble-sheet ayers.Bayeri.te.The absenceof large, single crystals of bayerite with which tomake structural determinations, renders the detailed structure of thispolymorph uncertain. Montoro (1942) concluded from X-ray powderanalysis-that bayerite possessed hexagonal attice with a:5'01 A,c:4.76 A, containing 2 molecules f AI(oH)3. This is a layer structuresimilar to gibbsite but without the distortion of the sheet of anionsaround each vacant cation site that makes gibbsite pseudo-hexagonal.Instead of the hydroxyl ions in adjacent layers being in a state of open-packing as in gibbsite (Fig. 9), the hydroxyl ions interdigitate in a stateof close-packing Fig. 10). This proposed structure for bayerite is identi-cal with the structure of brucite (Mg(OH)), except that in brucite everycation position is filled and in bayerite one-third of the cation positionsare vacant.Milligan (1951) and Kroon and Stolpe (1959) took exception to the

iseI


20/35

62 ROBERT SCHOEN AND CHARLES E. ROBERSONbrucite-like structure proposedby Montoro. Other workers (Yamaguchiand Sakamoto, 1958; Lippens, 1961) using material of greater puritythan Montoro's, confirm the generalcorrectnessof the brucite model forbayerite. Lippens showed that bayerite is slightly distorted from hexag-onal symmetry and is more l ikely orthorhombicwith o: 8.67A, D:5.06A, and c:4.71A. For comparison, he hexagonal rucite modelof Yama-guchi and Sakamotoyields an orthohexagonalo-axisof 8.74 A.The brucite structure consistsof two sheetsof close-packed, egativelycharged hydroxyl ions held together by positively charged magnesiumions occupying all the intersticesbetween the hydroxyl sheets.These twosheetsof hydroxyl ions, together with the intervening sheet of magnesiumions, constitute a layer of the mineral brucite, and this layer is the smallestunit that can exhibit the propdrties of brucite. Larger brucite crystalsform by the superpositionof layers on top of one another, as well as bythe lateral growth of all the sheets.Lateral growth occurs simply by theaddition of magnesium and hydroxyl ions in a configuration where everymagnesium on binds six hydroxyl ions with bonds of one-third strengthand every hydroxyl ion binds three magnesium ions. These bonds areionic and three, one-third strength magnesium-hydroxyl bonds com-pletely satisfy the chargeof a hydroxyl ion. Therefore, there are no ionicforces available to bond together the hydroxyl ions of superposedbrucitelayers. In addition, the proton-oxygen bond within each hydroxyl ionpoints straight up at right angles o the sheetsof ions and away from thesheet of magnesium cations (Elleman and Williams, 1956). This resultsin a sheet of protons on the upper and lower surfacesof each layer ofbrucite, and it producesa tendency toward cylindrical symmetry in thehydroxyl ions.Growth of brucite by the superpositionof layers, therefore, must takeplace without the help of any ionic bonding and must overcome theelectrostatic repulsion of sheets of adjacent protons. The bonding be-tween brucite layers is due to van der Waals forces supplemented byforces related to the interaction of the rotating hydroxyl dipole ions. Asa result of all these constraints, the hydroxyl ions in successive rucitelayers interdigitate in such a way as to keep adjacent protons as farapart as possible.This interdigitation, similar to that found within thebrucite layer, has been called "close-packing" but it is much more openthan that. Table 2 shows that the effective hydroxyl ion radius towardhydroxyl ions in the next layer in brucite is 1.61A (t.SZ A in bayeriteaccording to Lippens, 1961), whereas the effective hydroxyl ion radiustoward a cation is 1.39 A. tfrir indicates how much more open this so-called "close-packing" between ayers of brucite really is.This discussionof the structure of brucite (Mg(OH)z) applies directly


21/35

STRUCTURESOF ALUMINUM HYDROXIDE 63to the structure of bayerite (AI(OH)a) if we keep n mind the substitutionof two aluminum ions and one vacancy for every three magnesium ons.The presenceof holes n the cation sheet ntroduces somedistortion intothe bayerite structure compared with the perfect hexagonal symmetryof brucite. Bayerite, however, does not suffer the additional distortioncausedby hydroxyl bonding within the layers that affectsgibbsite'

Comparison of Figures 9 and 10 shows that the simplest unit of struc-ture of gibbsite and bayerite is identical (a double sheet of hydroxylions), and the difference between the two polymorphs is solely in themanner of stacking superposed ayers. The symbol AB IAB IAB andso forth, describes the stacking scheme for bayerite as depicted inFigure 10.

$""minumFrc. 10. Diagrammatic representation of brucite-bayerite layer structure.

Nordstrandite.Becauseof the lack of goodsingle-crystalsand the relativerecency of the discovery of nordstrandite, both as a synthetic product(Van Nordstrand and others, 1956) and as a naturally occurring mineral(Hathaway and Schlanger, t962; Wall and others, 1962), the detailed

B

AT-I+I IOu B

A- . / \


22/35

ROBERT SCHOEN AND CHARLES E. ROBERSONstructure of nordstrandite remains obscure.By analogy with the X-raypatterns of gibbsite and bayerite, Van Nordstrand and others (1956)proposedthat nordstrandite (called "bayerite II" by them) possessedlayered structure with a spacing between layers equal to the averageofthe gibbsite and bayerite spacings.They proposed that the new phasemight be a screwdislocationpolymorph in which the gibbsitepacking andbayerite packing alternated and in which the Burgers vector was twolayers high.Lippens (1961) examined the structure of nordstrandite as well asgibbsite and bayerite. His powder samples for X-ray study containedgibbsite, bayerite, pseudoboehmite,and nordstrandite in varying pro-portions, requiring him to ignore X-ray lines attributable to the otherthree phases.For this reason, as well as the lack of single-crystal data,Lippens' results are only a reasonableapproximation to the structure ofnordstrandite.

Lippens (1961) reported that his synthetic nordstrandite was mono-clinicwith a:8.634, D: S.OtA, c 19.12 , andB 92.000. he c-dimen-sion of I9.l2L indicates that the unit cell of nordstrandite contains twolayers of gibbsite (4.S5A each) and two layers of bayerite (4.71A each).Thesemay be stacked n either of two ways: I-bayerite, gibbsite, gibbsite,bayerite; or ll-bayerite, gibbsite, bayerite, gibbsite. By an analysis ofthe 00/-diffractions, Lippens concluded that arrangement II was mostprobable. This coincideswith the arrangement proposedby Van Nord-strand and others (1956), and is depicted in Figure 11. The symbol forthis stacking sequences:

ABluelt lAn attempt at single-crystal studies on the natural material fromGuam by Hathaway and Schlanger 1965), met with fai lure when allnordstrandite crystals selected proved to be multicrystalline or dis-torted. A similar attempt on the natural material from Sarawak ap-parently was successful Saalfeldand Mehrotra, 1966).As recently reported by Saalfeldand Jarchow (1968),nordstranditeis t r ic l in ic wi th o:8.752 A, b:5.069 L, c : t0 .244 A; o:109.33o, B:97 66",?:88.340. Theseauthors report that "From the structural point

of view nordstrandite can be regarded as intermediate between gibbsiteand bayerite." But unlike the stacking sequenceof nordstrandite pro-posedby Lippens (1961),Saalfeldand Jarchow (1968), n their Figure 3,depict the unusual stacking sequence:B A I A CI

BAiIABiI and so forth.

CB and so forth


23/35


24/35

66 ROBERTSCHOENAND CHARLESE. ROBERSONThe validity of this model requires urther elaboration and confirmation.Pending a complete and precise description of the nordstranditestructure, the regular interlayer model (Fig. 11) has much in its favor.By analogy with another large group of layer structures, the clay min-erals, interlayering is a common phenomenon in response o environ-mental changes.And, as in the clay mineral group, there seems o be adefinite tendency for a 1i1 regular interlayered structure to be morestable than either irregular interlayering or other proportions of thecomponent layers. In addition, a single layer of aluminum hydroxidepossesses marked proclivity to interlayer, both in nature (the chloritemineral group) and with other clay minerals in the laboratory (Hsu andBates ,1964b) .

Sranrr,rrv or AlutrrNulr HynnoxrnB PorvuonpgsSeveralsuggestions ave been offered to explainwhy gibbsiteor bayer-ite may be favored during precipitation of aluminum hydroxide. VanNordstrand and others (1956) suggested hat if nordstrandite is a screwdislocation polymorph with a Burgers vector of two layers, bayerite canbe considered o be a screw dislocation ploymorph in which the Burgers

vector is one ayer. They suggested hat unspecifiedconstraints imposedby this type of growth might prevent the gibbsite-type open-packingfrom developing. Barnhisel and Rich (1965) believed that some of theirdata showed a "seeding" effect by foreign mineral surfaces hat mightcontrol the structure of the aluminum hydroxide polymorph. Hsu (1966),realizing the great difference n the rate of precipitation of gibbsite andbayerite, suggested that rapid precipitation may favor the bayeritestructure, and slow precipitation the gibbsite structure. None of thesesuggestions, owever, specified he mechanismof stability control.Our study leads us to conclude that a difierent control from thosealready proposed s responsibleor the.structuresof the aluminum hy-droxide polymorphs. Comparison of Figures 9 and 10 shows hat the onlydifference between gibbsite and bayerite is in the mode of stacking thealuminum hydroxide layers. And the two different modes of layer stack-ing reflect the shapesof the hydroxyl ions on the opposingsheet surfaces.If mild polarizing forcesaffect the hydroxyl ions, they assumea cylindri-cal shape, and the proton-containing ends of the ions move as far aspossibleaway from the aluminum ions between the sheetsof hydroxylions. When layers of aluminum hydroxide with this hydroxyl ion con-figuration superpose, he repulsive forces of the opposing protons forcethe layers to interdigitate. The brucite-bayerite type structure results.When stronger polarizing forces affect the hydroxyl ions, the cylindricalsymmetry breaks down and tetrahedral ions form. The distribution of


25/35

STRUCTURES OF ALUMINUM HYDROXIDE 67electrostatic chargeson these tetrahedral ions permits an ionic bond toform with tetrahedral hydroxyl ions in the superposedaluminum hy-droxide layer. The strong attraction of these ionic hydroxyl bonds re-quires that the hydroxyl ions arrange themselvesvertically above oneanother perpendicular to the layers. The gibbsite structure is the result.It is probable then that formation of gibbsite or bayerite dependsuponthe distortion, and consequently the degree of polarization, of the hy-droxyl ions.

But how does the pH of the mother liquor cause such a difference npolarization of the hydroxyl ions in the precipitate? And what can ac-count for the maintainence of a large disparity in degreeof hydroxyl ionpolarization in two solid phasesof the same composition? The answersto thesequestionsconstitute what is, at present, only a reasonableguess.Additional specific investigations must be undertaken to evaluate thefollowing hypothesis.

First, we must remember that aluminum is amphoteric and can existin solution as a complex cation or anion depending on the pH. Proofof this statement is afforded by the observation that electrolysis of asolution of potassium aluminate causes deposition of aluminum hy-droxide on the anode, whereas electrolysis of a solution of aluminumnitrate deposits aluminum hydroxide on the cathode (Berges, 1947). Asdescribedpreviously, Hem and Roberson (1967) showed that the pre-dominant cationic aluminum complex between pH 4 and pH 5.8 isAl(OH)(OHr)r'+. In this complex ion, the lull polarizing power of thealuminum ion affects a single hydroxyl ion. Polymerizatiort of the mono-meric AI(OH)(OH)rrt+ brings the polarizing power of additional alumi-num ions to bear, and deprotonation of the water moleculescontinueswith the formation of strongly polarized hydroxyl ions. The precedingchain of events would be expected in acidic solutions. Above pH 5.8,however, the dominant aqueousaluminum species s the complex alumi-nate anion AI(OH)41-. We would expect the excessnegative charge tonullify the polarizing power of the aluminum ion and form cylindricalhydroxyl ions. Therefore, the phase hat precipitates from acid solutionsshould contain highly polarized hydroxyl ions becauseof their abundancein the acidic mother liquor, and the phase hat precipitates from alkalinesolutions should contain weakly polarized hydroxyl ions becauseof theirabundance in the alkaline mother liquor. And this is exactly what weobserve.Judging from the preponderanceof gibbsite in nature, we can inferthat strong polarization of the hydroxyl ion in aluminum'hydroxide isthe most stable state under conditions prevailing at the earth's surface.It terms of crystal energy (Pauling, 1960,p. 509), the longer hydroxyl-


26/35

68 ROBBRT SCHOEN AND CHARLES E. ROBERSONhydroxyl bond in bayerite, caused by the weak polarization of the hy-droxyl ions, creates a smaller crystal energy than strongly polarizedhydroxyl ions do and, therefore, a lessstable structure. Bayerite shouldtend to undergo a slow increase n the polarization of its hydroxyl ionsand recrystalize to gibbsite.

Our experimental results, however, do not support this. What formsduring long-term aging of bayerite is not gibbsite but nordstrandite, amineral of intermediate structure containing both strongly and weaklypolarizedhydroxyl ions. Nordstrandite, which is probably a regular inter-layering of the gibbsite and bayerite modes of stacking, may be the stablepolymorph of aluminum hydroxide in an alkaline environment.A probable qualitative explanation for this is that even though thecrystal energy associated with weakly polarized hydroxyl ions is less

than that associatedwith strongly polarized hydroxyl ions, the relativeactivity of strongly polarized versus weakly polarized hydroxyl ions inthe solution also afiects the stability of the solid phase. If the solutionis alkaline, the activity of weakly polarized hydroxyl ions (AI(OH)a1-)is much greater than that of strongly polarized hydroxyl ions (AI(OH)(OH)r)6'+)(Hem and Roberson, 1967). This ratio of ion activities maymake a structure containing both strongly and weakly polarized hy-droxyl ions (nordstrandite) more stable than onecontaining only stronglypolarizedhydroxyl ions (gibbsite).

In other words, strongly polarized hydroxyl ions make a crystal cagethat is more efficient at containing the internal energy of the enclosedatoms. But this crystal cagemust be in a state of dynamic equilibriumwith the solution surrounding it. If the surrounding solution contains anabundance of weakly polarized hydroxyl ions, the crystal cage mustcompensate or this . It does this by assuming a structure made up ofstrongly polarized hydroxyl ions for high efficiency in containing theinternal energy of its atoms, and embodying weakly polarized hydroxylions in response o the high relative activity of these ons in the surround-ing environment.

GBocsnursrnv or AluurNUM HyDRoxrDE MTNERALSWe now can compare the laboratory observations and hypothesis forpolymorph stability with the natural occurrenceof aluminum hydroxideminerals to see f we can obtain additional eeochemical nsieht.

Gibbsite n nature. The most abundant aluminum hydroxide mineral isgibbsite. It is the principal mineral in most bauxite ore deposits, theprimary sourceof aluminum metal. Gibbsite is alsoabundant in laterite,the ferruginous analog of bauxite. Harder (1952) described he geology


27/35

STRUCTURES OF ALAMINUM HYDROXIDE 69of many bauxite deposits and ascribed their genesis o the interplay ofmany factors.

Bauxite (and its dominant mineral gibbsite) is principally a residualweathering product of any kind of aluminum-containing rock. There is acomplete spectrum of parent rock types from alkali-rich, aluminousigneous rocks through clays, basalts, and metamorphic rocks to alumi-num-poor limestones. In addition to an easily soluble aluminous rock,warm climate, high rainfall, moderate topographic relief, and freedomfrom erosion are all necessary or the optimal development of bauxitedeposits.The advent of extensive erosionwill, of course, emove bauxitefaster than it f orms. And a change n the delicate balanceof other factorswill alter the geochemical onditions and end the formation of bauxite.Some believe that bauxite deposits form by solution, transport, andprecipitation of aluminum. Theobold and others (1963) summarized thebackground of this hypothesis and presented a small-scalenatural ex-ample. They postulated the transport of aluminum in acid solution and(or) as a sulfate complex, derived from the oxidation of disseminatedpyrite in the parent rock.Many bauxite deposits of unquestioned residual origin show evidenceof local solution and precipitation. It is still uncertain, however, whetherthis mechanism can produce large bauxite deposits some distance awayfrom the sourceof aluminum.

One type of bauxite deposit, whosegenesis till is an enigma, developson aluminum-poor limestones. It seems quantitatively impossible toconcentrate the one tenth of one percent or less of Al2O3 n the insolubleresidue of these imestones o form thick bauxite deposits.Perhaps theseunusual bauxites represent accumulations of aluminous material derivedfrom a wide area and washed nto sinkholesand other Iocal basins.Dur-ing filling of a sinkhole, large amounts of water percolate through thedepositedsediments.And, after filling, peat bogs frequently develop atthe surface resulting in continued percolation through the sediments ofwaters rich in humic acid (Clark, 1966; Bushinsky, 1964).In addition to gibbsite, boemite (AIO(OH), one of the monohydrateminerals of aluminum, frequently occurs n tropical bauxite depositsnearthe surface where the heat of the sun may causedehydration of gibbsite(Harder, 1952, p. 56). Although the temperature for this transitionranges rom 1300C o 200oC n the laboratory, the presenceof salts in thesoil profile may reduce the activity of water suficiently to make thisdehydration possible at lower temperatures. Hsu (1967) showed thathigh salt concentrations would delay almost indefinitely the conversionof pseudoboehmite o bayerite. Boehmite also s the principal aluminousmineral in bauxite depositsaround the Mediterranean Seawhere regional


28/35


29/35

STRUCTURES OF ALUMINUM HYDROXIDE 7Itions above pH 7 that nordstrandite formed in our Iaboratory experi-ments thus supporting the inferred conditions of genesisof natural nord-strandite.Another interesting aspect of the distribution of nordstrandite onGuam is the minor amounts found in limestone some distance from thecontact with volcanic iocks. Hathaway and Carroll (az Schlanger,1964,p. Daa) speculate hat it may be detrital gibbsite that recrystallized tonordstrandite in the limestone environment. The results of our synthesisexperiments support the reasonableness f this hypothesis.

Nordstrandite from west Sarawak, Borneo, occurs as rounded pelletsin terra rossasoil on the edgeof a sinkhole n limestone (Wall and others,1962). The authors suggest an origin related to weathering of probableoverlying dacitic sills . Although the data given are meager, we may sur-mise that the limestone host terrain probably limited nordstrandite-precipitating solutions to a pH above 7.Bayeri,te n Nature. The only well-documented occurrence of naturalbayerite known to us is in Hartruim, fsrael (Bentor and others, 1963;Gross and Heller, 1963). The investigators used X-ray diffraction toidentify the bayerite, which occurs with calcite and gypsum in veinletscutting sedimentary rocks of late Cretaceousage. The rocks are com-posedof calcite and spurrite (a calcium-carbonate-silicate sually formedby contact metamorphism) in a ratio of about two to one. The veinletsalso contain vaterite, portlandite, tobermorite group minerals, thauma-site, and ettringite. Several of these minerals connote genesisat a veryhigh pH (Gross and Heller, 1963), and this agreeswith the conditionsunder which pure bayerite ormed n our experiments pH l2+).

A recent report of bayerite in a weathering crust of amphibolites andserpentinites in Russia (Khorosheva, 1968) is corroborated by X-raydiffraction. The associatedminerals, in this obviously non-equilibriumenvironment, are gibbsite, diaspore, and possibly nordstrandite.Bayerite-nord.stranditen'igma. The relative ease and speed with whichbayerite, and to a lesser extent nordstrandite, form in the laboratorycontrast with the apparent paucity of these phases n nature. Gibbsite,the slowest orming phase n the laboratory, is by far the most abundantof the three aluminum hydroxide polymorphs in nature.

We have shown that bayerite and nordstrandite form only in alkalinesolutions, and the field evidence mplies that the few natural samplesofthese minerals also ormed in alka line environments .Perhaps he scarcityof aluminum-bearing alkaline ground waters is part of the reason or thescarcity of bayerite and nordstrandite in nature. In addition, because


30/35

72 ROBERT SCEOEN AND CHARLES E. ROBERSONof the abundanceof silica in most natural alkaline waters, it is temptingto speculate on the importance of silica in favoring the formation ofaluminosilicate minerals rather than aluminum hydroxide.If silica inhibits the formation of aluminum hydroxide, why doesgibb-site form in the pH range from 4 to 7 where silica greatly exceeds lumi-num in most natural waters?The answermay depend upon the dominantform of silica in solution. At pH values below 9, most of the silica is inthe monomeric uncharged form HaSiOr. Above a pH of 9, increasingamounts of silica go into solution as the silicate ion (HsO)Si(OH;ut-(Iler, 1955). Perhaps the charged silicate on is required to form alumino-silicate minerals, and the virtual absenceof this ion at pH values belowabout 7 (except when the silica concentration is high) allows aluminumin solution to precipitate asgibbsite.If the silicate ion inhibits the formation of bayerite and nordstrandite,the few known natural occurrences of these minerals must representalkaline environments with very little silica n solution. The environmentof nordstrandite deposition on Guam, described by Hathaway andSchlanger (1965),probably possessed high pH and a low silica concen-tration. Initially, acid ground water in the underlying basalt and tuffshould react with the unstable primary silicates to form secondarysili-cates, increase the pH of the water, and take aluminum into solutionas the aluminate ion (Al(OH)rl-). Little silica should go into solutionbecause hese rocks contain almost no uncombined silica (Stark, 1963),and the solubility of the secondary silicates s low in alkaline solutions(Pickering,1962).

From what we can deduce about the nordstrandite depositional en-vironment in west Sarawak, it resembles that on Guam. Wolfenden(1961) described most of the bauxite deposits in Sarawak as formingfrom intermediate to basic igneous rocks, generally with little quartz.The immediate environment of the nordstrandite pellets on the edgeofa sinkhole in limestone is roughly equivalent to that on Guam.Barnes and others (1967) reported that ground waters high in pH(lI-12) and low in silica (


31/35

STRUCTURES OF ALAMINUM EYDROXIDE 73by reacting with and destroying rock minerals The long-term through-flow of meteoric waters must. result in the destruction of the rock in anenvironment that is slightly acid most of the time. The vastly lowersolubility, in these waters, of aluminum relative to other metals, leadsto the residual enrichment of aluminum hydroxide. And the generallyacid nature of the environment through time causes he gibbsite struc-ture to form.

Our laboratory aging of bayerite-nordstrandite mixtures, althoughunderway for only 2 years, seems o indicate that during aging nord-strandite forms at a greater rate than bayerite (Fig. 8 versus Fig. 6). This,together with the almost complete absenceof bayerite in nature, leadsusto propose hat bayerite is a metastablephase hat will eventually recrys-tallize in alkaline solutions to the stable phasenordstrandite. How longit takes for completerecrystallization of bayerite to nordstrandite we donot know, though we detected increased amounts of nordstrandite injust a few months.If this proposal is correct, the natural bayerite from Hartruim, Israel(Bentor and others, 1963; Grossand Heller, 1963),must be metastable.This natural bayerite formed, presumably, many hundreds or thousandsof years ago. We can speculatethat its slow rate of recrystallization tonordstrandite may be due to the extreme dryness of the region in whichit occurs.Hartruim is located on the southwesternedgeof the Dead Seain a region with a mean annual rainfall of less than 50 mm (Orni andEftat, t964). 1lhe presence of vaterite (a metastable polymorph ofCaCO) in these bayerite-bearing rocks attests to their dryness, forvaterite is unstable in the presenceof water at room temperature andpressure McConnell, 1960;Johnstonand others, 1916).The uncertainties of extrapolating from conditions of laboratory syn-thesis to natural conditions must always be kept in mind and the pos-sibility of reinterpretation left open. fn the alumina-water system theeffectof most natural foreign ions has been horoughly studied and foundto slow, but not change, the courseof crystallization. Our experiments,and those of others, show that pH is the overriding control on crystalstructure. And, as the preceding discussionshows,the best pH that canbe inferred for the environment of precipitation of natural aluminumhydroxides approximates the pH found in the laboratory.

SuulranyThe mostimportant conclusionsderiving from this study are:1. The pH of the mother liquor controls the structure of the aluminumhydroxide precipitate.


32/35

ROBERT SCEOEN AND CHARLES E. ROBERSON2. The gibbsite structure forms when the pH is less than about 5'8.

The bayerite structure forms when the pH is greater than 5.8.3. The nordstrandite structure forms slowly at pH's greater than5.8. Becausenordstrandite seems o form at the expenseof previously-

formed bayerite, it may be the stable polymorph of aluminum hydroxidein alkaline solutions.

4. Though the structures of bayerite and nordstrandite are imper-fectly known, they seemto differ from gibbsite and each other solely inthe manner of stacking successiveayers of aluminum hydroxide.5. This difierence in stacking is directly attributable to constraintsimposed by polarization-induced distortions of the hydroxyl ions.6. The strong and weak distortion of the hydroxyl ions in acid and inalkaline solutions, respectively, s inherited by the gibbsite and bayeritesolids that precipitate from theserespectivesolutions.

7. The structure of nordstrandite seemsbest explained as a 1:1 regularinterlayering of the gibbsite and bayerite modes of stacking.

8. The suggestedstability of nordstrandite relative to bayerite mayreflect a compromise between the higher crystal energy afforded bystrongly polarized hydroxyl ions and the dominance of weakly polarizedhydroxyl ions in the surrounding solution.9. The scarcity of nordstrandite and abundance of gibbsite in naturereflects, n part, the dominantly acid nature of most weathering environ-ments.

10. The scarcity of nordstrandite, even in alkaline environments , mayindicate a preferenceby aluminate ions to form aluminosilicateswith theusually abundant silicate ions.

The conflict in earlier work regarding the conditions of stability of thealuminum hydroxide polymorphs arose or two reasons: 1) small quanti-ties of often poorly crystallized material subject to strong preferredorientation favored misidentification, and (2) depending upon the pres-ence of foreign ions and other experimental variables, metastable solidsfrequently formed. The conversion of these metastable phases o stablephases s often slow, especially f the metastable phasesare well crys-Lallized.Metastable products therefore have frequently been consideredto be stable.

We found that synthesesusing perchlorate ons, though they bear ittlerelation to natural systems,decrease he effect of metastableprecursors.Though the effect s decreasedt is not eliminated as shown by our dift-culty in converting well-crystallized bayerite to nordstrandite duringreasonableperiods of aging.

Though a good number of our conclusionsrepresent hypothesesandspeculationsbasedon imprecise data, we feel that theseproposalsshould


33/35

STRUCTURES OF ALUMINAM HYDROXIDE 75be made here so that they may guide future work aimed at their aban-donment, revision, or acceptance.

AcxxowreoournrsW. L. Polzer and Carol Lind producedthe electronmicrographsand J. C. Hathawayprovidedus with the X-ray difiractogram oI nordstrandite.We appreciate he efiortsof R. A. Gulbrandsen, . C. Hathaway,J. D. Hem, and P. H.Hsu, all of whomreadthe manuscriptand suggested hangesor its improvement.We are especiallygrateful for numerousdiscussions ith J. D. Hem that frequentlysowed he seedsrom which our morespeculative roposalsater blossomed.

RrlmrrcrsAmxAwonn, L. E., axo H. P. Kr,uo (1948)Basicaspects f X-ray absorption n quantita-tive di.firaction nalysisof powdermixtures.Anal. Chem.,20, 8G889.BAnNts, IvaN, V. C. LaMencrru, Jn., eNo Glnn Hru*rsnno (1967)Geochemical vi-denceof present-day erpentinization.Sci.ence56,830-832.BAnNursnl, R. I., .tlro C. f. Rrcrr (1965)Gibbsite,bayerite, and nordstrandite ormationas afiectedby anions,pH, and mincralsurfaces. roc.Soil,Sci..Soc.Amer.,29r531-534.Bnxton, Y. K., S. Gnoss,er.roL. Hnr.r,ur (1963)Someunusualminerals rom the "mot-tled zone"complex, srael.A tner.M inera1,., 8, 924-930.Bnncns, Menrue (1947)Sur un modede formation d'alumine hydrat6epar 6lectrolyse.C.R. A cail.Sci.,P wis 225,241-2+3.Bnnrel, J. D., .lwn H. D. Mncaw (1935)The function of hydrogen in intermoleculartorces.Proc.RoyalSoc.Lond,on 51A,384420.BusnrNsrv, G. I. (19fl) Types of karst bauxitedepositsand their genesis,n Symposiumsw lesbaurites,oxyd.est hyd.roryiles,'aluminum,lrZagteb, AcaddmieYougoslave esSciencest desArts,93-195.Bvr, G. C., eno J. G. RoarrsoN (1964)Crystallizationprocessesn aluminum hydroxidegels.K ol,l,oi.d-. Z.Jiir P olym.198,53-60.CrAer.n,O. M., Jn. (1966)Formation of bauxiteon karst topography,Eton. Geol.6l, 903-916.Dnun, W. 4., R. A. Howm, axo J. ZussueN(1962)Rock-Jormingmi,nerals,S,on-si.licates.New York, JohnWiley and Sons,371p.ErtruAn, D. D.,.Lwo D. Wrr,raus (1956)Proton positions n brucite crystals.J. Chem.Phys.25,742-74.Gnoss,S., arro L. Hnr.r,nn 1963)A natural occurrence f bayerite.Mineral..Mag.33r 723-

a 1 ^HAroEn, E. C. (1952)Examplesof bauxite deposits llustrating variations n origin, i.nProbl'ems JClayand LateriteGenesis. ew York, Am. Inst. Mining Metall. Engineers,35-&.Ileruewrv, J. C., axo S. O. Scnr,eNcnn 1962)Nordstrandite from Guam. Noture 196,265-266.- (1965)Nordstrandite (Atroa.3HrO) rom Guam. Amo. M,i,neral. 0, 1029-1037.IlAuscrru;o,Urmcn (1963)Uber nordstrandite, Al(OH)s. Z. Anorg.undAll,g.Chem.,324,15-30.Hntt, J. D. (1968)Graphicalmethods or studiesof aqueous luminum hydroxide, luoride,and sulfatecomplexes.U. S. Geol.Sora.Water-SuPflry ap. 1827-81 3p.-r.tN'D C. E. Rosrnsox (1967)Form andstability of alumiriumhydroxidecomplexesin dilute solution.U. S. Geol.Sura.W ater-Suppl,y ap. 1827-A,55p.


34/35

76 ROBERT SCHOEN AND CHARLES E. ROBERSONHrn.ar.r.ox,A., aNo M. C. Gasrucsr (1962)Synthdse t gendsede I'hydrargillite. C. RAcad,.Sci,.P aris, 254,1105-1107.Hrr,r,,V. G. (1955)The mineralogyandgenesis f thebauxitedepositsof Jamaica,B. W. I.A ma. M ineral'. 40, 676-688.Hsu, P. H. (1964)Efiect of anionson the formation of Al(OH)a labstr.l Program13thNorth Amer. ClayMinuds ConJercnce.adison,Clay Minerals Society,13.- (1966)Formation of gibbsite from aging hydroxy-aluminum solutions.Soil' Sci.Soc.Amn. Proc.,30,17 -17 .- (1967)Efiects of salts on the formation of bayerite versuspseudo-boehmite.Soi,Scz'., 03,101-110.Hsu, P. H., ern T. F. Bnrns (1964a)Formation of X-ray amorphous nd crystallinealu-minum hydroxi des.M i,ner l. M ag.,33,749-768.- (1964b)Fixation of hydroxy-aluminum pol1'rners y vermiculite. Soil' Sci'. Soc.

Ama. Proc., 28, 763-T69.Hurcnntsor.r,G. E. (1945)Aluminum in soils,plants,andanimals.Soi'l ci,.,60,29-40.Irrn, R. K. (1955)TheColloid.alChemistryoJil,haand, ilicates.Ithaca,CornellUniversityPress, 24p.JonNsror, J., H. E. MnnwrN, axl E. D. Wrr.lnusolt (1916)The severalorms of calciumcarbonate.Amer. J. Sci.,Ser.4, 4\ 473-512.Krv, D. H. (1965)Techniquesor El.ectron icroscoy'y. xford,BlackwellScientiic Publica-tions,560p.Kronosueva, D. P. (1968)Baierite boksitovogogoizonta srednego ridneprov'ia.Dokl'.Akod..Nauh.,S,SSR. 82, 434-436.Kr,uc, H. P., aNnL. E. Ar,nxexonn (1954)X-ray DffiaclionProcdures. New York, JohnWiiey and Sons,716p.Knoow, D. J., lwo C. v. d. Sror,lr (1959)Positionsof protons n aluminum hydroxidesderived rom proton magnetic esonance. atwe 183,9M-945.LrrreNs, B. C. (1961)Stru,cture nd,TertureoJAlumi,nas.ProefschriftDelft, Netherlands,I79 p.McCorlmr.r, J.D.C. (1960)Vaterite from Ballycraigy,Larne,Northern Ireland. Mineral.M ag., 32,535-544.Mncew, H. D. (1934)The crystal structure of hydrargilIite A1(OH)8. . Kri'statrl'0gr.87,185-204.Mrr-r-rcaN,W. O. (1951)RecentX-ray difiraction studieson the hydrous oxidesand hy-

droxides. . Phys. anL Colloi'd' hem.55, 497-507MoNroro, V. (1942)Struttura cristilfina dellabayerite. RicercaSci. Prog. Tecn.Roma.,13,565-571.Onwr, E., aNo Er-rsrreE. (1964)Geography J vael,. Jerusalem, srael Programfor Sci'Transl. 335p.Plrfn, D., R. Trntr*m, erroR. Brers (1958)Recherchesur a constitutiondesgelset deshydratescrystallisds 'alumine.Bul'l'.Soc.Chim.Fr.,1958' 1301-1310.PAurrNG,L. (1960) The Nattae oJ he Chemi,cal ond'. rhaca, Cornell University Press,644p.Prcxrnrxc, R. J. (1962)Someeachingexperiments n three qtartz-free silicate ocksandtheir contributionto an understanding f laterizatton.Econ.Geol,.,S7r185-1206.R.rNre.uA, K. aNn T. G. Sen.$re (1950)Geochemi'stry.hicago,University of ChicagoPress,912.Roorsnv, H. P. (1961)Oxidesand hydroxidesof aluminum and iron, in The X-ray lilen-tif,cati'onond' CrystoJStrwctwres J Clay Minaals. London, Mineralogical Society.35+-392.


35/35

STRUCTURES OF ALUMINUM HYDROXIDES,Leunm, H. (1960)Strukturen desHydrargiilits und der Zwischenstufen eim Entw?is-sern.Neues ahrb.Mineral'.Abh.95, l-87.-t ANDB. B. Mrnr.ornl (1966)Zur Struktur von Nordstrandit A-I(OH)3.Natwwiss.,s3,L28-129.ScntANGrn,S. O. (1964)Petrology of tle limestonesof Guam. U. S. Geol'.Swt. ProJ.Pap.,4O3-D,2p.Saar.rnlo, Honsr, nr,mO:rro Jencrow (1963)Die Kristallastruktur von Nordstrandit,AI(OH)3.Neues ahrb.Mineral.Abhond'l'.109'85-191.Hnrnz MettruEs, AND Munnm Kenu.lr

STRUCTURES OF ALUMINUM HYDROXIDE.pdf

Documents