+ All Categories
Home > Documents > Revised yalues for the thermodynamic properties of ... · Revised yalues for the thermodynamic...

Revised yalues for the thermodynamic properties of ... · Revised yalues for the thermodynamic...

Date post: 29-Jul-2018
Category:
Upload: truongtruc
View: 230 times
Download: 0 times
Share this document with a friend
13
American Mineralogist, Volume 76, pages445-457, 1991 Revised yalues for the thermodynamic propertiesof boehmite,AIO(OH), and related species and phases in the systemAI-H-O Bnucn S. Hr'urNcw,l,v, Rrcn.cno A. Ronrn U.S. Geological Survey, Reston, Virginia 22092,U.5.A. JonN A. Apps Lawrence BerkeleyI-aboratory, Berkeley,California 94720, U.S.A. Ansrnlcr Heat capacity measurements are reported for a well-characterized boehmite that differ significantly from resultsreported earlier by Shomateand Cook (1946) for a monohydrate of alumina. It is suggested that the earlier measurements were made on a sample that was a mixture of phases and that use of that heat-capacityand derived thermodynamic data be discontinued. The entropy of boehmite derived in this study is 37 .19 + 0. l0 J/(mol . K) at 298.15 K. A review of recent 27Al solution NMR data and other experiments has shown that the method of preparation of Al-bearing solutions can significantly affectthe concentration of monomeric Al species in solution. Because the procedures by which Al solution concen- trations are measured in solubility studiesdetermine the quantity of so-called monomeric species present,apparent differences in calculatedgibbsite stability most likely arise from diferences in experimentalprocedurerather than from differences in crystallinity, as often suggested. A review ofpublished solubility data for gibbsite suggests that the best values that can be currently estimated from that data for the Gibbs free energies of formation of Al(OHh and Al3+are - 1305.0 + 1.3and -489.8 t 4.0 kJ/mol, respectively (Hemingway et al., 1978). Basedon our value for the entropy and acceptingthe recommendedGibbs free energy for AI(OH);, we have calculated the Gibbs free energy and enthalpy of formation of boehmiteto be -918.4 + 2.1 and -996.4 + 2.2kl/mol respectively, from solubility data for boehmite. The Gibbs energy for boehmite is unchanged from that given by Hemingway et al. (1978). fNrnonucrroN monly attributed to boehmite (Shomateand Cook, 1946) Minerals with compositions in the chemical system were based on heat-capacity data for a phase described AlrO3-SiOr-HrO are ft"settt in a wide variety of sedi- as a_ monohydrate that produced an X-ray pattern similar mentary and metamorphic rocks and in soils. Metasta- to that of bayerite' Kelley and King (1961) were the first bility of phasesin this system has resulted in disparate to describe the phase as boehmite. Subsequenttabula- experimental data and cons€quentdisparate interpreta- tions of thermochemical data have followed Kelley and tion of the phase relationships of some of the phaies in King ( I 96 I )' Recently, Apps et al. ( I 988) concluded that the system.Boehmite, AIO(OH), is an example of such u the entropJ @I 298.15 K and I bar) of boehmite was too phase. large and Berman et al. (1985) concluded the heat capac- Diaspore, AIO(OH), is generallyaccepted to be the sta- ities reported for boehmite were too large.We report here ble aluminum hydroxide or oxyhydroxide phaseat earth heat-capacity measurements for a well-characterized surfacetemperatures (e.g.,Parks, 1972; perkins et al., sampleofboehmitethatvalidatetheconcernsexpressed 1979; Hemingway, 1982). However, conflicting conclu- by Parks (1972), Hemingway et al. (1978),and Berman sions have been reachedregarding the relative stabilities et al. (1985) and confirm the prediction of Apps et al. of other aluminum hydroxides and oxyhydroxides (e.g., (1988). Sanjuanand Michard, 1987,have calculateda Gibbs en- ergy for bayerite that would make it slightly more stable Sample than diaspore). Inconsistenciesand errors in the ther- Two samples of boehmite were synthesized by K. We- modynamic data utilized by the several groups may be fers ofAlcoa. The first samplewas preparedfrom gibbsite responsiblefor some of the disparate interpretations. Parks during a 20-h hydrothermal experient at approximately (1972) and Hemingway et al. (1978) noted and expressed 473 K. The resulting boehmite was well crystallized, as concern that the entropy and heat-capacity data com- shown by X-ray diffraction and SEM analysis. Aggregates 0003-004X/9 l/030,1-0445$02.00 445
Transcript

American Mineralogist, Volume 76, pages 445-457, 1991

Revised yalues for the thermodynamic properties of boehmite, AIO(OH),and related species and phases in the system AI-H-O

Bnucn S. Hr'urNcw,l,v, Rrcn.cno A. RonrnU.S. Geological Survey, Reston, Virginia 22092,U.5.A.

JonN A. AppsLawrence Berkeley I-aboratory, Berkeley, California 94720, U.S.A.

Ansrnlcr

Heat capacity measurements are reported for a well-characterized boehmite that differsignificantly from results reported earlier by Shomate and Cook (1946) for a monohydrateof alumina. It is suggested that the earlier measurements were made on a sample that wasa mixture of phases and that use of that heat-capacity and derived thermodynamic databe discontinued. The entropy of boehmite derived in this study is 37 .19 + 0. l0 J/(mol .K) at 298.15 K.

A review of recent 27Al solution NMR data and other experiments has shown that themethod of preparation of Al-bearing solutions can significantly affect the concentration ofmonomeric Al species in solution. Because the procedures by which Al solution concen-trations are measured in solubility studies determine the quantity of so-called monomericspecies present, apparent differences in calculated gibbsite stability most likely arise fromdiferences in experimental procedure rather than from differences in crystallinity, as oftensuggested. A review ofpublished solubility data for gibbsite suggests that the best valuesthat can be currently estimated from that data for the Gibbs free energies of formation ofAl(OHh and Al3+ are - 1305.0 + 1.3 and -489.8 t 4.0 kJ/mol, respectively (Hemingwayet al., 1978).

Based on our value for the entropy and accepting the recommended Gibbs free energyfor AI(OH);, we have calculated the Gibbs free energy and enthalpy of formation ofboehmite to be -918.4 + 2.1 and -996.4 + 2.2kl/mol respectively, from solubility datafor boehmite. The Gibbs energy for boehmite is unchanged from that given by Hemingwayet al. (1978).

fNrnonucrroN monly attributed to boehmite (Shomate and Cook, 1946)Minerals with compositions in the chemical system were based on heat-capacity data for a phase described

AlrO3-SiOr-HrO are ft"settt in a wide variety of sedi- as a_ monohydrate that produced an X-ray pattern similarmentary and metamorphic rocks and in soils. Metasta- to that of bayerite' Kelley and King (1961) were the first

bility of phases in this system has resulted in disparate to describe the phase as boehmite. Subsequent tabula-

experimental data and cons€quent disparate interpreta- tions of thermochemical data have followed Kelley andtion of the phase relationships of some of the phaies in King ( I 96 I )' Recently, Apps et al. ( I 988) concluded thatthe system. Boehmite, AIO(OH), is an example of such u the entropJ @I 298.15 K and I bar) of boehmite was toophase. large and Berman et al. (1985) concluded the heat capac-

Diaspore, AIO(OH), is generally accepted to be the sta- ities reported for boehmite were too large. We report here

ble aluminum hydroxide or oxyhydroxide phase at earth heat-capacity measurements for a well-characterized

surface temperatures (e.g., Parks, 1972; perkins et al., sampleofboehmitethatvalidatetheconcernsexpressed1979; Hemingway, 1982). However, conflicting conclu- by Parks (1972), Hemingway et al. (1978), and Berman

sions have been reached regarding the relative stabilities et al. (1985) and confirm the prediction of Apps et al.of other aluminum hydroxides and oxyhydroxides (e.g., (1988).

Sanjuan and Michard, 1987, have calculated a Gibbs en-ergy for bayerite that would make it slightly more stable Samplethan diaspore). Inconsistencies and errors in the ther- Two samples of boehmite were synthesized by K. We-modynamic data utilized by the several groups may be fers ofAlcoa. The first sample was prepared from gibbsiteresponsible for some of the disparate interpretations. Parks during a 20-h hydrothermal experient at approximately(1972) and Hemingway et al. (1978) noted and expressed 473 K. The resulting boehmite was well crystallized, asconcern that the entropy and heat-capacity data com- shown by X-ray diffraction and SEM analysis. Aggregates

0003-004X/9 l/030,1-0445$02.00 445

446 HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

TABLE 1. Chemical analysis for synthetic boehmite

Abundance'o/"

Trau 2. Experimental low-temperature molar heat capacitiesfor boehmite

Temper- Heat Temper- Heat Temper-ature capacity ature capacity ature

K J/(mol K) K J/(mol.K) KAI

MgNaK

AuBaCaCeCoCrCsDyFeGaHfLaMgMnMoNiRbScSiSmTaUYbZn

HeatcapacityJ/(mol.K)

ppm

- NAA analyses by Helen V. Michel.-'lCP analyses by Andrew W. Yee.

of crystals average l5 pm in diameter with the individualcrystallites 0.2-2 pm in size and displaying well-definedcrystal outlines. No evidence of amorphous material wasfound; however, two additional phases [diaspore,AIO(OH), and akdalaite, 4Al,O3 .H,Ol were found in smallquantity. This sample was used for all calorimetric stud-ies reported here.

Following identification of diaspore in the first sample,a second sample was synthesized from the same startingmaterial, at the same temperature, but for a shorter pe-riod of time. The resulting sample was well-crystallizedboehmite in the form of aggregates of very thin plateletsof individual crystallites. No evidence of amorphous ma-terial or other phases was found. However, the very smallcrystallites made this sample impractical as a sample forcalorimetry because of the low effective packing densityand the possibility that the He exchange gas would beadsorbed on the fine crystallites at temperatures less thanapproximately 15 K.

A chemical analysis of the boehmite sample is listed inTable l. Combining the results of SEM, X-ray, andchemical analyses, we estimate that the sample contains< l0lo combined diaspore, AIO(OH), and akdalaite,4Alr03.HrO.

Portions of the first sample were used for both low-temperature and differential scanning calorimetric (DSC)measurements. The sample used for low temperature heat-capacity measurements was 27.3540 g. Superambient heat

43.4(22146.2(23)*o.12('t2)0.814(8)

<0.06

<0.0019<5.3<0.73<0.29<0.016

2.96(14)<0.017<0.04071(s)44.8(67)<0.029<0.14r 1 9(1 2s)-'

7.43(35)s.8(s)

15.501)<0.57<0.003536(2s)'-0.0031 1 (43)0.0045(6)0.036(5)

<0.009219.0(10)

Sedes 1305.67 55.51309.93 55.93314.72 56.67319.55 57.45324.36 58.14329.17 58.89333.96 59.36338.74 60.12343.55 60.77348.38 61.56

Series 28.38 0.03859.16 0.0292

10.3s 0.029311.70 0.0594'12.98 0.100414.52 0.143116.29 0.160418.22 0.161620.30 0.178322.53 0.209325.01 0.264827.79 0.350630.88 0.480634.36 0.672638.27 0.857942.63 1.25747.47 1 .65252.93 2.19458.92 2.907

Series 355.12 2.45660.56 3.134

Series 365.03 3.74770.07 4.52875.96 s.53081.63 6.57187.27 7.72192.89 8.95198.49 10.24

103.98 1 1 .53109.38 12.83'114.70 14.151 19.96 15.47125.14 16.78130.27 18.10135.35 19.40140.38 20.68145.36 21.97150.30 23.24155.21 24.46160.08 25.721U.92 26.92169.73 28.10174.51 29.30

Series 4179.13 30.42184.15 31.65188.85 32.79193.54 34.28198.1 0 35.06202.67 36.11207.22 37.07

Series 4225.65 41.01230.38 42.04235.18 43.04240.03 43.96244.90 44.88249.79 45.81

Series 5260.21 47.85265.83 49.07277.39 50.81243.23 51.82289.00 52.80294.72 53.70300.44 54.73306.14 55.7131 1.82 56.33317.48 57.23

Series 6267.57 49.06269.12 49.27270.68 49.49272.29 49.83273.89 50.11275.48 50.44277.08 50.71

Series 7316.98 57.05322.s6 57.82328.18 58.66333.86 59.44

211.77 38.22 339.52 60.22216.35 39.05 345.19 61.03220.97 40.01

Nofe.'Molar mass : 59.989 g.

capacities reported in this study were determined frommeasurements on portions that were approximately 25mg. The 1975 values for the atomic weights (Commissionon Atomic Weights, 1976) were used. The molar mass ofboehmite is 59.989 g.

Apparatus and procedures

The low-temperature adiabatic calorimeter and datahandling procedures are described elsewhere (Robie andHemingway, 1972; Robie et al., 1976; Hemingway et al.,1984). Low-temperature heat capacities were measuredusing the intermittent heating method under quasi-adia-batic conditions. The sample was sealed in the calorim-eter under a small pressure of pure He gas (approximately5 kPa).

High-temperature heat capacities were determined byDSC following the procedures outlined by Hemingway etal. (198 l). The samples were placed in unsealed Au pans.

Low-rnupnRATURE HEAT cApAcrrrEs ANDTHERMODYNAMIC FUNCTIONS

Experimental heat capacities for boehmite are listed inTable 2 in the chronological order of the measurements.The results are corrected for curvature (e.g., Robie and

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE 447

Hemingway, 1972),bu;t not for the heat capacity contri-butions from the diaspore and akdalaite contaminantphases that are discussed below. The observed heat ca-pacities were smoothed using cubic spline smoothing rou-tines and were graphically extrapolated as Cr/T vs. Z2 to0 K using the experimental and smoothed values for tem-peratures below 30 K. Smoothed values of the heat ca-pacities and derived thermodynamic functions are listedin Table 3.

The heat capacities of diaspore (Perkins et al., 1979)are smaller than those of the impure boehmite sample atall temperatures below 300 K. Therefore, corrections tothe observed heat capacities of the impure boehmite sam-ple for the contribution of diaspore would result in anincrease in the calculated heat capacities and entropies.A correction of approximately +0.02 J/(mol.K) wouldbe required in the entropy of boehmite at 298.15 K foreach lolo of diaspore.

Low-temperature heat capacities have not been re-ported for akdalaite; however, Mukaibo et al. (1969) havereported heat-capacity measurements for tohdite at su-perambient temperatures that may be used to estimatethe magnitude of the correction for the akdalaite impu-rity. (Note that akdalaite and tohdite are considered tobe the same phase, e.g., Fleischer, 197 l.) The specific heatof akdalaite is approximately 4o/o greater than that of co-rundum at 298.15 K. This difference is expected to in-crease at lower temperatures. Based on a comparison withboehmite and diaspore, we estimate the average differ-ence in the specific heat ofakdalaite between 0 and 298. I 5K to be l0o/0. Using the entropy of corundum from Robieetal. (1979), we estimate the entropy of akdalaite (4AlrO3.H,O) to be 234 J/(mol.K) at 298.15 K. Based on thisestimate for the entropy of akdalaite, we estimate that acorrection of +0.04 J/(mol.K) would be required in theentropy of boehmite at298.15 K for each l0lo of akdalaitermpunty.

We estimate the total correction to the entropy calcu-lated for the impure boehmite sample a|298.15 K (Table3) to be less than 0.06 J/(mol.K). Because the boehmitesample contained a small quantity of very small crystalsthat would have a slightly higher specific heat than coars-er material of the same composition and because we es-timate the uncertainty in the calculated boehmite entropyat 298.15 K to be +0.I J/(mol.K), we have not applieda correction for the observed impurity phases (diasporeand akdalaite).

The measurements reported by Shomate and Cook(1946) for a monohydrate of Al and variously attributedto the phase boehmite (e.9., Kelley and King, l96l) maybe compared with the heat capacities reported herein. Atall temperatures ropresented by their measurements, theresults of Shomate and Cook (1946) are significantly larg-er than those reported in Table 2. At 100 K the heatcapacities differ by approximately 44o/o and by approxi-mately 2lo/o at 298.15 K. Shomate and Cook (1946) alsoreported heat capacity data for gibbsite that may be com-pared with the data for gibbsite reported by Hemingway

TABLe 3. Molar thermodynamic properties of boehmite

Heatcapacity

cgEntropyS"r - 58

GibbsEnthalpy energyfunction function(t/f, - -(G9 -Hlllr r$lr

TemperatureK J/(mol.K)

51 01 52025303540455060708090

1001 1 01201301401501601701801902002102202302402502602702802903003 1 0320330340273.15298.15

0.0030.0270.1 530.1760.2640.4400.6961.0051.4441.8973.0524.5156.2748.311

10.5812.9815.4818.O220.5923.1425.6828.1930.6833.1335.4937.7339.8841.9543.9545.8847.7349.5251.2552.9354.53s6.0657.5158.9360.3350.0754.24

0.0010.0080.0410.0880.1350.197o.2840.3950.5390.7141.1571-7332.4483.3014.2935.4136.6497.9889.418

10.9312.5014.1315.8117.5419.3021.0922.8924.7126.5428.3730.2132.O433.8735.7037.5239.3441.1442.9344.7132.6237.19

0.001 0.0000.006 0.0020.033 0.0080.065 0.0230.094 0.0410.136 0.0610.198 0.0870.277 0.1180.382 0.1560.511 0.2030.833 0.3231.251 0.4821.766 0.6822.377 0.9243.082 1 .2103.872 1.5404.735 1.9145.659 2.3296.634 2.7847.649 3.2768.697 3.8039.770 4.362

10.86 4.95211.97 5.56813.09 6.21114.21 6.87715.33 7.56316.44 8.26917.54 8.99218.64 9.73119.72 10.4820.79 11.2521.85 12.0222.89 12.8123.92 13.6024.93 14.4025.93 15.2126.91 16.0227.87 16.8421.13 11 .4923.73 13.45

Note-'Molar mass : 59.989 g

et al. (1971), who used a calorimeter that was similar tothe one used in this study. The data of Shomate and Cook(1946) were higher at all temperatures, but the maximumdifference was approximately 7o/o aI 52.8 K and that dif-ference decreased to l.5olo at 298.15 K. The foregoingcomparison suggests that the differences between the heatcapacity values reported by Shomate and Cook (1946)and the values reported in this study arise primarily fromdifferences in the sample, not in the equipment or pro-cedures for data processing.

ffrcrr-rnupnRATuRE HEAT cApAcrrrEs ANDTHERMODYNAMIC FUNCTIONS

Experimental superambient heat capacities for boehm-ite are listed in Table 4. The results represent measure-ments based on several samples, with new samples pre-pared and used following any partial dehydration of asample. Although the majority of HrO was lost in the

448 HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

Series 1338.9 59.77348.9 60.94358.8 61.93368.8 63.02378.7 64.05388.6 65.00398.6 66.02408.5 66.99418.5 67.95428.4 68.85438.3 69.81448.3 70.48458.2 71.50468.1 72.24478.1 73.03488.0 73.90497.0 74.66

Series 2338.9 59.82348.9 60.97358.8 62.03368.8 63.08378.7 64.09388.6 65.07398.6 66.07

Series 2408.5 67.07418.5 67.97428.4 68.91438.3 69.83448.3 70.57458.2 71.61468.1 72.37478.1 73.18488.0 74.00497.0 74.85

Series 3607.3 79.86616.2 81.35

Series 4468.1 72. ' t1478.1 72.75

Series 5597.3 79.53607.3 79.88616.2 81 .79

Series 6537.7 77.39547 .6 77.76557.6 78.24567.5 78.53

Selies 6577.5 78.93587.4 79.05597.3 79.40607.3 79.81617.2 80.31627.2 80.91637.1 81 .14647.0 81 .73657.0 82.36666.9 83.21

Series 7537.7 77.40547.6 77.77557.6 78.20567.5 78.63577.5 78.97587.4 79.32597.3 79.61607.3 80.30617.2 80.55627.2 81.11637.1 81.44647.0 81.69657.0 82.56666.9 83.54

TABLE 4, Experimental superambient molar heat capacities forboehmite

Tem- Heat Tem- Heat Tem- Heatperature capacity perature capacity perature capacity

K J/(mol.K) K J/(mol.K) K J/(mot.K)

Between 300 and approximately 420 K, the heat capac-ities derived from the data of Shomate and Cook (1946)are larger than those reported here. Above approximately420 K. they are smaller.

As noted earlier, the most reasonable explanation forthe difference between the data presented by Shomateand Cook (19a6) for a monohydrate of Al and the resultspresented here is that the monohydrate studied by Sho-mate and Cook (1946) was not pure boehmite. Bayeritewas identified by X-ray analysis to be a constituent of thesample studied by Shomate and Cook (1946). Mukaiboet al. ( I 969) g1ve 47 3 K as the temperature of dehydrationof bayerite, consistent with the observation of Shomateand Cook (1946) thar significant HrO loss occurred attemperatures at or below 520 K. In addition, the densityof the sample used by Shomate and Cook (1946) wasdetermined and found to be 2.83 g/cm3, considerablylower than the theoretical value of 3.07 g/cm3 for boehm-ite; however, the value of 2.45 g/cm3 reported in the samestudy for gibbsite is identical with the theoretical valuefor gibbsite, suggesting that analytical error was not afactor.

The procedure reported by Shomate and Cook (1946)for the preparation of their monohydrate alumina samplewas followed as closely as possible. Dehydration ofgibbs-ite yielded a fine-grained, poorly crystallized mixture ofboehmite and akdalaite. The presence of akdalaite in oursynthesis may indicate a possible cause of the nonstoichi-ometry (low HrO content) of the sample prepared by Sho-mate and Cook (1946). Also, the addition of H,O to oursample with subsequent heating at 80 "C (following theprocedure of Shomate and Cook, 1946) yielded strongX-ray peaks for bayerite and significant loss of boehmiteX-ray peak intensities. Thus, several lines of evidencesuggest that the monohydrate sample studied by Shomateand Cook (1946) was a mixture of phases.

Fnrn nNpncv oF BoEHMTTE

The entropy derived in this study cannot be combinedwith the enthalpy of formation of boehmite to estimatedirectly the Gibbs energy of formation, as the enthalpyof formation of boehmite has not been established inde-pendently. Estimates of the Gibbs free energy of boehm-ite may be developed from phase equilibria and solubilitydata.

Ervin and Osborn (1951) developed a phase diagramfor the system AlrO3-HrO based upon a study of unre-versed synthesis experiments. Fields of stability are shownfor gibbsite, boehmite, diaspore, and corundum based onthe products of crystallization of AlrO, gel and 7-alumi-nia. Equilibrium was considered to be proved if bothstarting materials (i.e., structurally different materials)yielded the same product. Boehmite commonly formedand then slowly recrystallized to form diaspore or corun-dum in the P-Iregions designated as the stability regionsfor those phases. Therefore, the synthesis experimentsprovide only a limit as to the minimum or maximumfree energy that a phase may have. Realistically, the ex-

Note.'Molar mass : 59.989 g.

temperature interval of 680-710 K, some loss occurredas low as 450-600 K and resulted in the calculated heatcapacities for that interval having a U-shaped curvature.Weight loss calculated after such experiments was gen-erally ofthe order ofa few hundredths ofa percent ofthesample weight and probably represented loss of adsorbedHrO from some of the very small crystals. Each scanpresented in Table 4 represents one continuous set ofmeasurements or the average of several continuous setsof measurements.

Smoothed values of the heat capacities and thermo-dynamic functions are listed in Table 5. Smoothed valuesof the low-temperature heat capacities between 298.15and 350 K were combined with the high-temperature heatcapacities and the combined data set was fit by leastsquares to a 4-term polynomial with the constraint thatthe equation exactly fit the smoothed heat capacity at298.15 K. The resultant Equation l,

C" :205.721 - 0.0349217" - 2635.27T o5

+ 1.02666 x 106Z ' ( l )

fits the data with an average deviation of +0.30/o and isvalid from 298 to 600 K.

The measurements presented here may be comparedwith the data of Shomate and Cook (1946) who measuredthe heat contenr of their sample from 321 to 520 K. Sho-mate and Cook (1946) terminated their measurements atapproximately 520 K because HrO evolved irreversiblyfrom the sample and condensed in the sample capsule.

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

TABLE 5. Molar thermodynamic properties for boehmite

AIO(OH): Crystds 298.15 to 600 K.

449

Heat capacityFormation from elements

Temperature q .+ tf, - l+*"YT -(@, - l4""llr Enthatpy Gibbs free energy

298.1 5

Gibbs energyEntropy Enthalpy function function

300350400450500550600

54.24+0.1054.5161.0266.4170.8574.5177.5480.04r0.32

37.190+0.10037.52646.43354.94463.03070.69177.93984.797+0.320

0.000

0.3358.553

15.4592'1.37526.51131 .01735.001

37.190

37.1 9137.88039.48541.65644.18046.92249.796

-996.389+2.100

-996.415-996.846-997.269-997.367-997.365-997.058-996.990

+2.300

-918.400+2.200

-917.916-904.720-891.595-878.351-865.153-851.917-838.740

+2.400

ft"" - l4 7.07s + o.o2o kJTransitions in phase

Molar volume 1.9535 + 0.001 J/barTransitions in reference state elementsAr M.P. 933.45 K

Equationsq:205.721 - 0.034921 r- 2635.277-05 + 1026660r+

(Valid range: 298.15 to 600 K; Average absolute percent deviation: 0.50)

Note.'Molar mass : 59.989 g.

perimentally determined phase boundaries may representonly those regions in which the kinetics of recrystalliza-tion are rapid enough to be observed during the durationof the experiments performed by Ervin and Osborn ( I 9 5 I ),and boehmite may not be stable under any of the P-7conditions studied by Ervin and Osborn (1951).

We may use the experimental reaction boundaries giv-en by Ervin and Osborn (1951) to calculate minimumand maximum free energy values for boehmite. Trans-formation of gibbsite to boehmite is estimated to occurat approximately 400 K and 3 bars. Thus the minimumfree energy of boehmite at 400 K is estimated to be - 883.0kJlmol at 400 K and -909.8 kJ/mol at 298.15 K (usingancillary data from Robie et al., 1979). Transformationof boehmite to corundum is estimated to occur at ap-proximately 658 K and 136 bars. From these results andancillary data from Robie et al. (1979), the maximumfree energy for boehmite is estimated to be -922.1 kJ/mol at 298.15 K. Similarly, the maximum free energy fordiaspore is estimated to be -922.7 kJ/mol from the re-action boundary for the reaction 2 diaspore : corundum+ HrO given by Ervin and Osborn (1951). Within theforegoing calculation and in subsequent calculations in-volving gibbsite and corundum, the thermodynamicproperties listed in Robie et al. (1979\ are assumed ro begood estimates of the true values for these phases and,therefore, are used as fixed values. Results and evalua-tions presented by Haas et al. (1981), Hemingway et al.(1978), Hemingway (1982), and Apps er al. (1988) pro-vide support for this assumption.

Boehmite was considered by Kittrick (1969) to be morestable than gibbsite, based on a comparison of thermo-chemical data (see also Parks, 1972; and, Hemingway etal., 1978). However, Kittrick (1969) selected free energyof formation values for gibbsite and boehmite that were

based on different values for the free energy offormationof Al3+. Correcting the data for gibbsite to the same Alreference value internally consistent with the boehmitedata set reverses the relative stability ofthe two phasesat 298.15 K (assuming the activity of HrO is unity), butthe difference is less than the experimental uncertainty.The contrary conclusion ofChesworth (1972), that gibbs-ite and HrO at unit activity are stable with respect toboehmite at near-surface conditions, must also be con-sidered suspect because it is based on the highly uncertainestimate for the Gibbs energy of boehmite given by Ros-sini et al. (1952).

Parks (1972) and later Hemingway et al. (1978) select-ed the solubility data of Russell et al. (1955) for boehmiteas the best data set from which to determine the freeenergy of formation of boehmite. The result of the cal-culation is - 9 I 8.4 + 2. I kJ/mol for the Gibbs free energyof formation of boehmite at 298.15 K, and it suggeststhat boehmite * H,O is more stable than gibbsite. In thiscase, the assumptions made are (l) that the solubilityproduct at 298.15 K can be calculated from the extrap-olation of solubility data at higher temperatures, (2) thatthe solution species involved in the higher temperaturesolubility experiments is the same species as that com-monly referenced at lower temperatures, and (3) that thefree energy offormation is accepted for that species.

The estimate of the Gibbs free energy of formation ofboehmite given by Hemingway et al. (1978) is consistentwith two recent studies. Hovey et al. (1988) calculated arevised value for the Gibbs free energy of formation ofAI(OH); from solubility data for boehmite and a valuefor the Gibbs free energy of formation of boehmite ob-tained from Apps et al. (1988, then in preparation). Theresult, ArGln, : -1305.6 kJ/mol, is in agreement withthe value derived by Hemingway et al. (1978) but was

450 HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

derived somewhat circularly. Apps et al. (1988) deriveda value for the Gibbs free energy of formation of boehm-ite from the thermodynamic properties of gibbsite (Hem-ingway and Robie, 1977a) and from solubility data forgibbsite and boehmite. Apps et al. (1988) reported valuesof ArGln, of -917.5 and - 1304.8 kJ/mol, respectively,for boehmite and AI(OH);. From these results, we mayonly conclude that a Gibbs free energy of formation forboehmite of - 918 kJ/mol is consistent with the solubilitydata for gibbsite and boehmite and with a value of - 1305kJlmol for the Gibbs free energy of formation ofAI(OH);. However, since the value for the Gibbs freeenergy of boehmite determined by Apps et al. (1988) isreferenced to solubility data interpreted to representgibbsite solubility, the results and interpretations are sub-ject to the same questions posed for the studies of Hem-ingway et al. (1978) and Hemingway (1982) that are dis-cussed in the next section.

DrscussroN

Recent work by May et al. (1979), Couturier et al.(l 984), and Sanjuan and Michard (l 987) have questionedthe free energy of formation of AI(OH); derived by Hem-ingway et al. (1978) and Hemingway (1982). Specificallyat issue is the question ofthe phase or phases that controlthe solubility of Al in solutions with pH > 6 at 298.15K and at higher temperatures. Of more general concernis the question of the relative stabilities of the aluminumhydroxide and oxyhydroxide phases and the mechanismsby which precipitation ofthese phases are controlled. Inthe discussion that follows, we provide a detailed reviewof research that establishes the relative stability of theAI(OH)3 polymorphs (bayerite, nordstrandite, and gibbs-ite). We select a set of solubility data to represent con-ditions of metastable equilibrium between gibbsite andAl solution species and calculate the Gibbs free energiesof formation of Al3+ and AI(OH);. Finally, we providean explanation for the apparent variation ofgibbsite sol-ubility that has variously been attributed to grain size,acid pretreatment, or sample crystallinity.

Historical perspective

Hemingway and Robie (1977a) identified an error inthe calorimetric procedure upon which the enthalpy andfree energy of formation of gibbsite were based and re-ported a revised set of thermodynamic values for gibbs-ite. In subsequent work, Hemingway and Robie (1977b)and Hemingway et al. (1978) reported revised values forthe free energy of formation of AF* and AI(OH); basedon solubility studies for gibbsite by Kittrick (1966) andSingh (1974) and reported the revised value for the freeenergy of gibbsite. Implicit in the work of Hemingway etal. (1978) is the assumption that gibbsite controlled theAl solubility observed by Kittrick (1966). The validity ofthis assumption was first questioned in the study of Mayet al. (1979).

May et al. (1979) determined solubilities for a naturaland a synthetic gibbsite at several pH values between pH

4 and9 using several organic pH buffers. May et al. (1979)obtained two subparallel curves, one for each sample,that displayed offsets toward lower solubility at approx-imately pH:7 . May et al. (1979) concluded that, in basicsolutions, the solubility of Al was controlled by a phasemore stable than gibbsite (when synthetic gibbsite wasused as the starting material). May et al. (1979) tenta-tively identified the phase as boehmite, although no ev-idence for a phase other than gibbsite was found. May etal. (1979) utilized the solubilities determined from mea-surements obtained from the natural gibbsite sample insolutions with pH > 7 to calculate a revised value for thefree energy of formation of AI(OH); and, subsequently,to calculate the free energy of boehmite as -920.9 kJ/mol. May et al. (1979) concluded that the difference insolubilities observed between the natural and syntheticgibbsites in acid solutions was a consequence ofa differ-ence in crystallinity of the two samples; however, theyalso argued that, in basic solutions, the similar differencein observed solubilities resulted from control of solubilityby two phases. The apparent inconsistencies in interpre-tation of the solubility data led Hemingway (1982) toquestion the interpretations.

Hemingway (1982) combined the results of May et al.(1979) with other solubility studies from the literature toprovide an alternative explanation for the features ob-served in the solubility data of May et al. (1979). As Alhydrolysis progresses, the nature and characteristics ofthe aqueous Al species change. The observed offset insolubility curves corresponds with the change from thedominance of species traditionally considered to be of theform Al(OH)t3-v)+ to the form AI(OH); (e.g', Baes andMesmer, l98l). Hemingway concluded that a change inthe mechanism of precipitation accompanied the changein species structure. Hemingway (1982) reasoned that thischange in precipitation mechanism allowed bayerite toprecipitate only in solutions with a pH greater than ap-proximately 6. Hemingway (1982) concluded that in theexperiments performed by May et al. (1979) supersatu-ration with respect to phases other than gibbsite occurredand that the subparallel solubility curyes resulted fromcontrol of solution concentration of Al by precipitationof AI(OH), phases, nordstrandite and gibbsite in the acidregion, and bayerite and nordstrandite in the basic region.Hemingway (1982) further concluded that bayerite wasthe least stable of the three A(OH)3 phases and thatgibbsite was the most stable.

Sanjuan and Michard (1987) measured the solubilityof gibbsite at 323 K using procedures similar to those ofMay et al. (1979), obtained a similar offset in the ob-served solubility curve, and concluded that both the in-terpretations of May et al. (1979) and Hemingway (1982)were incorrect. Their interpretation was based on evi-dence that bayerite is either found in alkaline solutionsor replaces gibbsite in alkaline solutions (Verdes and Gout,1987; Schoen and Roberson,1970) and that AI(OH); hasa stability constant similar to that reported by May et al.(1979) (Couturier et al., 1984). Although both Sanjuan

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE 451

and Michard (1987) and Hemingway (1982) questionedthe phase or phases controlling Al solubility in the alka-line region, they reached different conclusions with re-spect to the relative stability ofbayerite and gibbsite, andwith respect to the free energy assigned to AI(OH);. Thelatter difference, in particular, must be resolved beforethe free energy of boehmite can be calculated.

Bis-Tris, an organic pH buffer used by May et al. (1979)in their solubility studies ofgibbsite, has a strong tenden-cy to form a complex with the aluminate ion (Wesolowskiet al., 1990). The increase in total dissolved Al resultingfrom this process is thought by Wesolowski et al. (1990)to be the cause of the offset observed in the solubilitycuryes published by May et al. (1979). Tris, the organicpH buffer used by May er al. (1979) at high pH, does notshow a strong tendency to complex aluminate ion (We-solowski et al., 1990). Thus, the results of May et al.(1979) may not represent equilibrium between gibbsiteand only hydrolized Al solution species in the pH rangebuffered by Bis-Tris, and the presence of a phase morestable than gibbsite may not be needed to explain thesolubility curve for gibbsite published by May et al. (1979)and by Sanjuan and Michard (1987). The enhanced sol-ubility of Al is consistent with the arguments of Heming-way (1982) and with the observation of bayerite in suchsolutions by Verdes and Gout (1987).

Solubilities lower than those found by May et al. (1979),Singh (1974), and Kittrick (1966) were reported by Bloomand Weaver (1982) and Peryea and Kittrick (1988) forseveral gibbsite samples that had previously been studiedby others (Kittrick, I 966; Frink and Peech, 1962). Bloomand Weaver (1982) attributed the lower solubility theyobserved in acid solutions to the removal of fine crystalsof gibbsite or reactive surfaces by acid pretreatment ofeach sample, and they ascribed the downward shift ofsolubility seen in the results of May et al. (1979) to morerapid Ostwald ripening of gibbsite in basic solutions.

Bloom and Weaver (1982) have compared their solu-bility results with one of several sets of results reportedby Frink and Peech (1962) also for acid pretreated ma-terials. Bloom and Weaver (1982) did, in fact, find a low-er solubility than the set they chose, but other data pro-vided by Frink and Peech (1962) are equivalent to thesolubility reported by Bloom and Weaver (1982). Hem-ingway (1982) compared the data of Frink and Peech(1962) to the model he proposed (his Fig. 5). AlCl, so-lutions to which Frink and Peech (1962) added HCI andgibbsite, and that were aged one month, showed solubil-ities equivalent to those reported by Bloom and Weaver(1982). AlCl, solutions to which no HCI or gibbsite wereadded hydrolyzed and showed solubilities after 3 monthsthat were equivalent to those found by May et al. (1979)for their natural gibbsite. A similar solution to whichgibbsite was added showed solubilities for pH < 4 thatwere equivalent to the solubilities for the solution equil-ibrated without gibbsite, whereas those with solution pH> 4 showed solubilities equivalent to those given by Mayet al. (1979) for synthetic gibbsite. Finally, solutions that

contained no added AlCl., but contained gibbsite andwere acidified with HCl, showed the lowest solubilities.The results reported by Bloom and Weaver (1982) areequivalent to the results reported by Frink and Peech(1962) where the same experimental parameters weremaintained.

Bloom and Weaver (1982) observed a significant dif-ference in the solubility of two sized fractions of FisherACS A(OH).. The sample FC with the smaller size frac-tion showed the greater solubility. The sample FF withthe larger size fraction was pretreated with dilute acidwhereas sample FC was not. The solubility difference ob-serVed by Bloom and Weaver (1982) was ascribed to theacid pretreatment processes. However, the solution inwhich sample FC was suspended was 0.01 M KNO. andresulted in a somewhat different chemistry for the studiesof FC and FF. The effect of this difference is discussedbelow.

Bloom and Weaver (1982) have shown that acid pre-treated gibbsite samples FF, C-730, and C-33 yield iden-tical solubility products. Samples C-730 and C-33 werestudied previously by Kittrick (1966, solubility study) andHemingway et al. (1978, solution calorimetry). Heming-way et al. (1978) also measured the enthalpy of solutionof Fisher A(OH)3 similar to samples FC and FF. Bloomand Weaver (1982) and Hemingway et al. (1978) con-cluded from their studies that all of the gibbsite sampleshad equivalent free eneryies.

Peryea and Kittrick (1988) have used a similar proce-dure to that used by Bloom and Weaver (1982) to studythe solubility of corundum, gibbsite, boehmite, and dia-spore. Peryea and Kittrick (1988) found Al concentra-tions in apparent equilibrium with gibbsite sample C-730lower than that reported by Kittrick (1966), and in agree-ment with the results reported by Bloom and Weaver(1982). Peryea and Kittrick (1988) calculated the free en-ergy of formation of the four Al-bearing phases based onthe solubilities they measured using the value of the freeenergy of Al3+ given by Hemingway et al. (1978). Theresults of these calculations were free energy values thatwere considerably more negative than those listed in sev-eral recent tabulations. However, there is an error in theprocedure followed by Peryea and Kittrick (1988) in thecalculation of the free energies of the phases (Hemingwayet al., 1989). The free energy of formation of Al3* re-ported by Hemingway and Robie (1977b, incorrectly cit-ed previously as Hemingway et al., 1978) is based on theassumption that gibbsite solubility was accurately deter-mined by Kittrick (1966). If the revised solubility forgibbsite is accepted, then the free energy of formation ofAl3* must be recalculated because the Gibbs free energyof formation of gibbsite has been determined by calori-metric methods and represents the reference value for Alin the calculation. The revised free energy of formationof Al3* would be -487 .5 kJ/mol and the corrected Gibbsfree energies of formation of corundum, boehmite, anddiaspore would be -1583.7, -919.1, and, -923.4 kJ/mol, respectively (Hemingway et al., 1989). These results

452

are approximately -1.5 kJ/mol more negative than re-sults reported in recent tabulations (e.g., Robie et al.,r979).

The free energy of formation of Al3* is subject to ad-justment, as discussed above, if acid pretreatment is shownto result in the best solubility data for gibbsite in acidsolutions. Assuming the other solubility results given byPeryea and Kittrick (1988) to also be the best solubilityvalues for corundum, diaspore, and boehmite in acid so-lutions, then the free energy of formation of Al3* can becalculated from these data and the free energies of for-mation of -1582.2 kJ/mol (CODATA, Cox, 1978),-922.9 kJ/mol and -918.4 kJlmol (Hemingway et al.,1978), respectively, for corundum, diaspore, and boehm-ite. These calculations yield -486.8, -487.0, and -486.8kJlmol, respectively, for the free energy of formation ofAl3+. The values are less negative than the value obtainedfrom the calculations based upon gibbsite solubility, butthe results agree within experimental error. If the solu-bility for gibbsite given by Bloom and Weaver (1982) isused in place of the data from Peryea and Kittrick (1988),one obtains -487.3 kJ/mol for the free energy of for-mation of Al3+. Whether or not the free energy for Al3*should be modified, these results demonstrate that thefree energies for the phases corundum, diaspore, boehm-ite, and gibbsite, as given above, are consistent. This, ofcourse, assumes that each phase has equilibrated with thesame Al solution species.

Rnr,,lrrvr srABrlrry oF THE A(OH)3POLYMORPHS

Before the Gibbs free energy of formation of boehmitecan be calculated, the Gibbs free energy of formation ofAl3* or AI(OH);, or both, must be established. To dothis, the relative stability of the AI(OH), polymorphs mustbe established.

If the evidence cited by Sanjuan and Michard (1987)is substantiated (discussion in an earlier section), then theconclusions they reached would directly follow and es-tablish the free energy of formation of AI(OH);. It isappropriate, therefore, to evaluate the supporting studies.The results of Couturier et al. (1984) and Schoen andRoberson (1970) are critical and are discussed below. Theresult of Verdes and Gout (1987) supports either view-point and thus is not definitive.

Couturier et al. (1984) have reported that they mea-sured the stability constants of hydroxocomplexes of Al3+and AI(OH); with oxalic acid. The authors chose thisprocedure because they believed that only dissolved spe-cies would be involved in their study, thus eliminatingthe problem of identifying the controlling AI(OH), phasethat is necessary in the application of solubility studies(e.g., May et al., 1979; Hemingway, 1982). Using ther-modynamic properties for Al3* (Hemingway and Robie,1977b), Couturier et al. (1984) calculated the free energyof formation of AI(OH); as -l3ll.3 kJ/mol, a valuesubstantially more negative than the value of - 1305 kJ/

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

mol, calculated by Hemingway et al. (1978) and Hem-ingway (1982).

Couturier et al. (1984) assumed that the strong com-plexes that occur between Al solution species and oxalicacid would prevent precipitation of AI(OH), phases. Cou-turier et al. (1984) also assumed that no mixed hydroxyl-oxalate complexes were formed. These assumptions ap-pear to be valid in acidic solutions (pH < 5), but may beinvalid in more basic solutions (Sjdberg and Ohman, 1985;Bilinski et al., 1986). Sjoberg and Ohman (1985) studiedthe equilibria between Al solution species, hydroxide, andoxalic acid from pH 0.2 to approximately 7. The upperpH limit in their study (coincidently the pH region inwhich May et al., 1979, observed an offset to lower sol-ubility) was set by the onset of precipitation as deter-mined by turbidity measurements. Violante and Violante(1980) studied the effect ofpH and chelating organic an-ions on the synthesis of aluminum hydroxides and oxy-hydroxides. Oxalic acid was found to not inhibit bayeriteprecipitation in alkaline solutions at low concentrations,but as the concentration (with respect to Al) was in-creased, bayerite precipitation was inhibited and nords-trandite or gibbsite precipitated. The studies of Sjdbergand 6hman (1985) and Violante and Violante (1980) showthat precipitation of A(OH)3 phases does occur in thepresence of oxalic acid in slightly basic solutions. Thus,the assumption made by Couturier et al. (1984) is invalidfor basic solutions and precipitation can be expected.Where precipitation does occur, the precipitation mech-anism will control the Al solution concentration and theequilibria with respect to oxalic acid will adjust accord-ingly. Therefore, the free energy of formation ofAI(OH); calculated by Couturier et al. (1984) must bequestioned.

The second critical study cited by Sanjuan and Mi-chard (1987) was that of Schoen and Roberson (1970)who reported that they had observed a gradual disap-pearance of gibbsite in solutions precipitating bayerite.Schoen and Roberson (1970) concluded that bayerite wasmore stable than gibbsite in basic solutions. However,examination of the data presented by Schoen and Rober-son (1970) suggests that nordstrandite was misidentifiedas gibbsite. Schoen and Roberson (1970) identified earlyformed solids on the basis of one or two X-ray diffractionpeaks (or calculated d-values) they considered definitive.Observed d-values were commonly from 4.7 to 4.9, andapproximately 4.4 and 2.2 A. Ttre d-values of 4.4 and2.2 were assigned to bayerite and the 4.7-4.9 A d-valueswere assigned to gibbsite. Although bayerite exhibits a d-value of 4.71 A, the early formed bayerite was consideredto have crystallized with poorly developed basal planes.Although nordstrandite was not considered by Schoenand Roberson (1970), it appears to be likely because Vio-lante and Violante (1980) assigned d-values of 4.72 Atobayerite, 4.79 A to nordstrandite, and 4.85 A to gibbsiteformed and examined under similar conditions.

The discussion given above strongly questions the in-terpretations and results of Couturier et al. (1984) and

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE 453

Sanjuan and Michard (1987) with respect to the relativestabilities of the A(OH). polymorphs and their calculatedvalue for the free energy of formation of Al(OH);. How-ever, the interpretations given above support, but do notprove, the interpretations of Hemingway (1982), Hoveyet al. (1988), and Apps et al. (1988).

Verdes and Gout (1988) provide evidence for the rel-ative stability of bayerite and gibbsite that is consistentwith the results of this study. On the basis of solubilitymeasurements, Verdes and Gout (1988) conclude thatgibbsite is more stable than bayerite and obtain a valuefor the Gibbs energy of formation of AI(OH); similar tothat proposed by Hemingway et al. (1978) and Heming-way (1982). Further, using the free energy of AI(OH);,Verdes and Gout (1988) have calculated -916 arrd -921kJlmol for the Gibbs free energy of formation of boehm-ite and diaspore, respectively, from a combination of sol-ubility measurements and from crystallization fromamorphous oxides.

The relative stability of the three common A(OH)3polymorphs also may be inferred from results presentedby Violante and Violante (1980) who studied the influ-ence ofchelating organic anions on the synthesis ofalu-minum hydroxides and oxyhydroxides. The authors foundthat, in alkaline solutions, increasing the ratio of the com-plexing organic anion to dissolved Al produced a changein the phase that precipitated, from bayerite to nord-strandite to gibbsite. This information is consistent withthe inference that bayerite is the least stable polymorphand gibbsite is the most stable.

Sor,unrr,rry oF GTBBSTTEVarious studies have concluded that a range of free

energies will be shown by gibbsite samples as a conse-quence of differences in crystallinity (e.g., Helgeson et al.,1978; May et al., 1979; Bloom and Weaver, 1982; San-juan and Michard, 1987). In acid solutions, there are threefairly consistent data sets that may be represented as threesubparallel curves ofAl concentration vs. pH in the acidregion (see Fig. 5 of Hemingway, 1982). Frink and peech(1962) used the same gibbsite sample, but report solubil-ities that fall along the three curves (one ofwhich is de-fined by some of their data). Because the same gibbsitesample was used in these experiments, the crystallinity ofthe gibbsite cannot be the cause ofthe observed solubilitydifferences. Also, it is unlikely that equivalent degrees ofcrystal imperfection could be obtained in different gibbs-ite samples (note agreement of Kittrick, 1966 Singh, 1974;May et al., 1979). Therefore, it would appear that vari-ations in sample crystallinity are not the major cause ofobserved differences in gibbsite solubility.

It does not appear that the acid pretreatment utilizedby Bloom and Weaver (1982) and Peryea and Kittrick(1988) is the cause of the lower solubility observed bythese authors. Kittrick (1966), Singh (1974), and May etal. (1979, synthetic eibbsite) used different gibbsite sam-ples but obtained nearly identical solubility values for Alin acid solutions. The solubility reported by Bloom and

Weaver (1982) for the treated sample studied by Kittrick(1966) is lower, but it is in agreement with results re-ported by Frink and Peech (1962) where the same generalexperimental approach was used. Of greatest importanceis the fact that Frink and Peech (1962), using the samegibbsite sample but different experimental parameters,found different solubilities, some of which agreed withthose of Kittrick (1966), Singh (197a), and May et al.(1979). Also, May et al. (1979) pretreated their gibbsiresamples by repeated suspension (seven times) in deion-ized HrO followed by centrifugation. Thus the sampleused by May et al. (1979) is as likely to have had fine-grained gibbsite particles removed and active surface de-fects modified as that sample studied by Bloom andWeaver (1982).

Several studies have shown that Cl- has an inhibitingeffect on the formation of crystalline AI(OH), (e.g.,Thomas and Whitehead, l93l; Hsu and Bates, 1964; Hsu,1967; Turner and Ross, 1970; Ross and Turner, l97l).In the latter two studies, the concentrations of mononu-clear and polynuclear Al ions were determined by theeight-quinolinolate extraction method (Turner, 1969) andthe amount of Al in the solid phase was calculated as thedifference between the initial total of the dissolved Al andthe sum of the mononuclear and polynuclear species. Ofcritical importance was the observation that after ap-proximately 12 d,, the concentration of mononuclear spe-cies remained essentially constant for periods of 100 d ormore and the solid phase showed no evidence of anA(OH)3 phase. The solid phase consisted of a basic alu-minum hydroxychloride that was X-ray amorphous ex-cept at higher chloride ion concentrations and times ofapproximately 300 d. The period in which the concen-tration of mononuclear species remained constant in-creased with increase in the concentration of Cl-. Follow-ing this period, gibbsite appeared in the crystalline phase,the concentration of polynuclear species decreased to 0,and the concentration ofmononuclear species increased(e.9., Fig. 28, Turner and Ross, 1970). The differences inthe solubilities ofgibbsite reported by Bloom and Weaver(1982) and Peryea and Kittrick (1988), as compared torhose of Kittrick (1966), Singh (1974), and May et al.(1979) are consistent with the results ofthese studies andsuggest an alternative explanation to that ofacid pretreat-ment. The extended period in which the concentration ofthe mononuclear species remains nearly constant and inwhich any aluminum hydroxychloride is X-ray amor-phous could easily be mistakenly interpreted as showingequilibrium between gibbsite suspended in such solutionsand the mononuclear species.

The most probable cause of the observed differences inAl solubility lies in differences in the experimental pro-cedures used in the various studies. Recent advances insolution nuclear magnetic resonance (NMR), in particu-lar the use ofthe Fourier-transform procedure beginningin the 1970s, has allowed extensive documentation of thebehavior of Al in solutions, and that information is ap-plicable to this study. A major contributor in this area is

454

Akitt who, with his coworkers (Akitt et al.,1972a, 1972b,l98l ; Aki t t and Far th ing, l98 la, l98 lb, l98 lc , l98 ld,l98le; Akitt and Elders, 1985), has provided much ofthe information that will be drawn upon in the followingdiscussion.

It has been known for many years that Al-bearing so-lutions behave differently when the solutions have some-what different chemistries (e.g., Hsu, 1967; Ross and Tur-ner,l97l1, Hemingway, 1982 and references therein; Tsaiand Hsu, 1984, 1985). Until recently, the nature of someofthese differences has been obscure. Akitt and Farthing(198 lb) used solution NMR and gel-permeation chro-matography to study the Al species present in two solu-tions prepared with different procedures. Both solutionswere prepared to have a ratio (m) of OH/AI of 2.5. Bothsolutions were prepared at approximately 100 'C. Thefirst solution was prepared by hydrolysis of AlCl, by therapid addition of NarCOr. In the second, aluminum met-al was dissolved in an AlCl. solution. The first solutionshowed one peak in the NMR spectrum which was as-signed to the species [AlO4Alrr(OH)ro(OHr),r]'* whichwill be described by the usual symbol Alli. The secondsolution spectrum was more complex and interpretationby Akitt and Farthing (l98lb) suggested at least four Alspecies, two of which were assigned to Al13* and[A(OH,)6]3* (designated as Al3+). Akitt and Farthing(l98lb) indicated that other hydrolysis methods yieldedsolutions showing spectra that differed from those de-scribed above, but exhibited the same general features,that is, varying ratios of the same Al species. Necessarily,these alternative hydrolysis methods involve changes inthe bulk chemistry of the solution [e.g., the use of AI(NO.),in place of AlCl.l as well as in the preparation procedures,but the work of Akitt and Farthing shows that the differ-ent procedures lead to diferences in speciation of Al. Ak-itt et al. (1972b, p. 605) have shown that the concentra-tion of monomer Al species is dependent on the procedurefollowed in the preparation of solutions with 0 < m <2.6.

Of importance to this study is the observation that forequivalent ratios of OH/AI, solutions to which hydro-chloric acid was either added or was an initial componenthad a lower concentration of monomer Al species thanthose prepared with solutions of AlCl, or AI(NO,), (Akittet al., 1972b). Peryea and Kittrick (1988), Bloom andWeaver (1982), and Frink and Peech (1962) used HCI aspart of the preparation of their experimental solutions.May et al. (1979) and Couturier et al. (1984) utilized ni-trate solutions. Singh (197 4) used AlCl, solutions. Kit-trick (1966) adjusted the pH of his samples with HCI;however, Kittrick did not pretreat his sample to removefine Al(OH), material, and evidence (see Bloom andWeaver, 1982; Kittrick, 1966) suggests that the fine ma-terial reacted rapidly to cause the solutions Kittrick stud-ied to quickly become oversaturated with respect togibbsite. Based on the results from "Al solution NMRand the procedures used to extract Al, one would antici-pate that Peryea and Kittrick (1988), Bloom and Weaver

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

(1982), and Frink and Peech (1962, HCI added) wouldobserve lower concentrations of Al in monomeric speciesthan would Frink and Peech (1962, HCI not added), Kit-trick (1966), Singh (1974), or May et al. (1979) simply asa consequence of the experimental approach followed inthe studies.

Akitt et al. (1972a) studied sulfato-aluminum com-plexes using 27Al solution NMR spectroscopy. The au-thors observed no change in the solution species uponaddition of small amounts of sulfuric acid, but observedthe destruction of the complex when hydrochloric acidwas added. Although sulfate is a strong complexing agentand may result in basic aluminum sulfate precipitation(e.g., Johansson, 1960, 1962; Johansson et al., 1960), thepresence of chloride results in reduction of the sulfato-aluminum complex. These results support earlier studies,described above (also see Barnhisel and Rich, 1965), thatsuggest that chloride appears to alter Al hydrolysis andto inhibit precipitation of aluminum hydroxides.

The precipitation mechanism for AIOOH and AI(OH).phases has not been determined as it has for the basicaluminum sulfates. The addition of sulfate ion to acidicAl solutions may result in the precipitation of two basicaluminum sulfate phases (Johansson, 1960, 1962). Thestructure of the Al within these phases is considered byJohansson and coworkers to reflect the structure of theAl polymers in the solutions from which the phases wereprecipitated. The Al polymers are the dimer and Alli.Bertsch et al. (1986a, 1986b) have shown that rapid neu-tralization of Al-bearing solutions to high m values re-sults in a lower production ofAlli and a greater produc-tion of pseudoboehmite. However, it is not clear whetherAllf is involved in the formation of the gelatinous pseu-doboehmite or whether competing reactions decrease theAl solution constituents required for the production ofAlT3*. Tsai and Hsu (1984 and references therein) haveshown that the Alli ion is lost through aging of Al solu-tions with the development of a more stable polymer(also see Akitt and Farthing, I 98 ld; Bertsch et al., I 986a,1986b). The structure of this polymer is unknown; how-ever, Tsai and Hsu (1984) have shown that developmentof the polymer results in a change in the morphology andstructure of basic aluminum sulfate that precipitates fromthe solution and may result in gibbsite precipitation. Tsaiand Hsu (1984, 1985) suggest that these polymers mayresemble fragments of crystalline Al(OH)r, as was sug-gested earlier by, for example, Smith and Hem (1972).Tsai and Hsu (1985) found that the negative logarithmof the solubility product (pK") of the initial solutionscontaining the metastable (by the definition of Tsai andHsu, 1984 and 1985) Allf was 32.32.This value is con-sistent with that given by Hem and Roberson (1967) andSmith and Hem (1972) for the solubility of what theycalled microcrystalline gibbsite. Tsai and Hsu (1985)found that the aged solutions that lost the All{ species,but retained the more stable polymer, had a pl(o of ap-proximately 33.4 to 33.5. This value is consistent withthe pK" calculated from the data of Frink and Peech

(1962,1963) and of May etal. (1979) for solubility of thenatural gibbsite sample and is consistent with the pI!,Hemingway ( I 9 82) postulated for nordstrandite.

Using ,7Al solution NMR, the structure of Al in themore stable polymer proposed by several authors (e.g.,Tsai and Hsu, 1985) cannot be determined, nor can it beproved that a polymer rather than a crystalline materialis present. However, work reported by Bottero et al. (1980)has shown that AI(OH); and Allr* can be present in so-lutions prepared at 20 "C with m as low as 0.5. Conse-quently, additional studies are necessary to determine thespecies actually involved in the precipitation of AIOOHand AI(OH), phases.

CoNcr,usroNs

The results discussed above provide only a glimpse ofthe complexity of the system under study. Chloride ionclearly interacts with Al and changes the mechanism ofhydrolysis in a manner that is different from that of thenitrate ion. 27Al solution NMR has not yet focused onthese processes and may not be able to resolve the struc-tural differences. However, the technique has establishedbeyond doubt that the method of preparation of Al so-lutions may result in differences in the type and amountof species present and, together with the preponderanceof other experimental data, supports the choice of thedata sets of Kinrick (1966), Singh (1974), and May et al.(1979) to determine the Gibbs free energies of Al3+(-489.8 + 4.0 kJ/mol) and AI(OH); (- 1305.0 + t.3kJ/mol). Based upon the measurements reported here, thisanalysis and the analyses of Hemingway et al. (1978),Hemingway (1982), Hovey et al. (1988), and Apps et al.(1988), the recommended values for the entropy, Gibbsfree energy, and enthalpy of formation of boehmite are37.19 + 0.1 J/(mol.K), -918.4 + 2.lkJ/mol,and, -996.4+ 2.2k1/mol, respectively. The recommended Gibbs freeenergy of formation of boehmite is intermediate betweenthe value of -917.5 kJlmol calculated by Apps et al.(1988) from solubility studies in basic solutions and thecorrected value from Peryea and Kittrick (1988) of -919.1kJ/mol based upon solubility studies in acidic solutions.

The recommended results are consistent with the anal-yses ofHovey et al. (1988) and Apps er al. (1988); how-ever, these analyses are not totally independent. Hoveyet al. (1988), using the Gibbs free energy of boehmite andset of solubility data recommended by Apps et al. (1988,then in preparation), calculated a value for the Gibbs freeenergy of formation of AI(OH); of -1305.6 + 0.2kJ/mol. Apps et al. (1988) used the same solubility data toestablish the Gibbs free energy of formation of boehmite(-917.5 kJlmol) and then calculated a consistent valuefor the Gibbs free energy of formation of Al(OH);(-1304.8 kJ/mol). Consequently, differences in the in-terpretation of models employed and model fit to exper-imental data result in small variations in the specific val-ues reported for either the Gibbs free energy of boehmiteor AI(OH);.

Both differences in sample crystallinity and acid pre-

455

treatment were examined as possible causes for observeddifferences in gibbsite solubility. Neither of these factorswere shown to be significant. Gibbsite would appear tocrystallize with an ordered and well-defined crystal struc-ture for which a single value of the Gibbs free energy isappropriate. However, as with any mineral, grinding ofgibbsite may result in distortion of the crystal surface andmay result in a surface energy contribution in some typesof measurements. Differences observed for the solubilityof gibbsite and commonly ascribed to differences ingibbsite crystallinity or acid pretreatment of gibbsite sam-ples more probably are caused by differences in the pro-cedures used in the experiments which results in a finalstate for the Al that is different from that assumed by theinvestigator.

ACKNowLEDGMENTS

We wish to thank K. Wefers of Alcoa for preparing the two boehmitesamples and Helen V. Michel and Andrew W. Yee of Iawrence BerkeleyI-aboratory for the chemical analysis. We thank our U.S. Geological Sur-vey colleagues Susan Russell-Robinson for help in the collection of DSCand X-ray data and John L. Haas, Jr. and J.J. Hemley for thoughtfulreviews of the manuscript. Constructive comments were prowided by R.O.Sack, J.D. Hem, and J.Y. Walther, and they are gratefully acknowledged.

RnrnnnNcns crrEDAkitt, J.W., and Elders, J.M. (1985) Aluminum-27 nuclear magnetic res-

onance studies of the hydrolysis of aluminum (III). Journal of theChemical Society, Faraday, 81, 1923-1930.

Akitt, J.W., and Farthing, AIan (l98la) Aluminum-27 nuclear magneticresonance studies of the hydrolysis of aluminum (III). Pan 2. Gel-p€rmeation chromatography. Joumal of the Chemical Society, Dalton,l 606- 1 608.

-(l98lb) Aluminum-27 nuclear magnetic resonance studies of thehydrolysis of aluminum (IID. Part 3. Stopped-flow kinetic studies.Joumal ofthe Chemical Society, Dalton, 1609-1614.

-(l98lc) Aluminum-27 nuclear magnetic resonance studies of thehydrolysis ofaluminum (II!. Part 4. Hydrolysis using sodium carbon-ate. Joumal ofthe Chemical Society, Dalton, 1617-1623.

-(l98ld) Aluminum-27 nuclear magnetic resonance studies of thehydrolysis ofaluminum (IID. Part 5. Slow hydrolysis using alurninummetal. Journal ofthe Chemical Society, Dalton, 1624-1628.

-(l98le) Hydrolysis ofhexa-aquo-aluminum (III) in organic media.Journal ofthe Chemical Society, Dalton, 1233-1234.

Akitt, J.W., Greenwood, N.N., and Khandelwal, B.L. (1972a) Alurninum-27 nuclear magnetic resonance studies of sulphato-complexes of thehexa-aquo aluminum ion. Journal of the Chemical Society, Dalton,1226-t229.

Akitt, J.W., Greenwood, N.N., Khandelwal, 8.L., and kster, G.D. (1972b)':7Al nuclear magnetic resonance studies of the hydrolysis and poly-merisation of the hexa-aquo-aluminum (III) cation. Journal of theChemical Society, Dalton, 604-6 10.

Akitt, J.W., Farnsworth, J.A., and Letellier, Piene (1981) Nuclear mag-netic resonance and molar-volume studies of the complex formed be-tween aluminum (III) and the sulphate anion. Journal ofthe ChemicalSociety, Faraday, 81, 193-205.

Apps, J.A., Neil, J.M., and Jun, C.-H. (1988) Thermochemical propertiesof gibbsite, bayerite, boehmite, diaspore, and the aluminate ion be-tween 0 and 350 "C. Iawrence Berkeley Laboratory-2l482, Berkeley,Califomia.

Baes, C.F-, Jr., and Mesmer, R.E. (1981) The fiermodynamics of cationhydrolysis. American Journal of Science, 281, 935-962.

Barnhisel, R.I-, and Rich, C.L (1965) Gibbsite, bayerite, and nordstrand-ite formation as affected by anions, pH and mineral surfaces. Soil Sci-ence Society of America Proceedings, 29, 53 1 -534.

Berman, R., Brown, T.H., and Greenwood, H.J. (1985) An internally

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

456

consistent database of the tlermodynamic properties of the systemNarO-KrO-CaO-MgO-FeO-FerOr-Alr03-SiOr-TiOr-HrO-CO,. Atom-ic Energy Canada Ltd. Technical Report 377.

Bertsch, P.M., Iayton, W.J., and Bamhisel, R.I. (1986a) Speciation ofhydroxy-aluminum solutions by wet chemistry and aluminum-27 NMRmethods. Soil Science Society ofAmerica Journal, 50,1449-1454.

Bertsch, P.M., Thomas, G.W., and Bamhisel, R.I. (1986b) Chtacteiza-tion of hydroxy-aluminum solutions by aluminum-27 nuclear magneticresonance spectroscopy. Soil Science Society of America Journal, 50,825-830.

Bilinski, Halka, Horvath, Iaszlo, Ingri, Nils, and Sjiiberg, Staffan (1986)

fuuilibrium aluminum hydroxo-oxalate phases during initial clay for-mation; H*-Al3+-oxalic acid-Na* system. Geochimica et Cosmochimi-ca Acta. 50. l9ll-1922.

Bloom, P.R., and Weaver, R.M. (1982) Efect of the removal of reactivesurface material on the solubility of synthetic gibbsite. Clays and ClayMinerals. 30. 281-286.

Bottero, J.Y., Cases, J.M., Fiessinger, F., and Poirier, J.E. (1980) Studiesof hydrolyzed aluminum chloride solutions. L Nature of aluminumspecies and composition of aqueous solutions. Joumal of Physicalchemistry, 84, 2933-2939.

Chesworth, Ward (1972) The stability of gibbsite and boehmite at thesurface ofthe earth. Clays and Clay Minerals, 20,369-374.

Commission on Atomic Weights (1976) Atomic weights of the elements.Pure and Applied Chemisrry, 47, 75-95.

Couturier, Yves, Michard, Gil, and Sarazin, G6rard (198a) Constantes deformation des complexes hydroxydes de I'aluminum en solution aqueusede 20 a 70 eC. Geochimica et Cosmochimica Acta, 48, 649-659.

Cox, J.D. (1978) CODATA recommended key values for thermodynam-ics. CODATA Bulletin 28, l-17.

Ervin, Guy, Jr., and Osbom, E.F. (1951) The system AlrO3-HrO. Journalof Geology, 59, 381-394.

Fleischer, Michael (1971) New mineral names. American Mineralogist,5 6 , 6 3 5 .

Frink, C.R., and Peech, M. (1,962) The solubility of gibbsite in aqueoussolutioDs and soil extracts. Soil Science Society of America Proceed-ings, 26, 346-347.

-(1963) Hydrolysis of the aluminum ion in dilute aqueous solu-tions. Inorganic Chemistry, 2, 473-478.

Haas, J.L., Jr., Robinson, G.R., Jr., and Hemingway, B.S. (1981) Ther-modynamic tabulations for selected phases in the system CaO-AlrOj-SiO,-H,O at 101.325 kPa (l atm) between 273.15 and,1800 K. JournalofPhysical and Chemical Reference Data, 10, 575-669.

Helgeson, H.C., Delany, J.M., Nesbitt, H.W., and Bird, D.K. (1978) Sum-mary and critique of the thermodynamic properties of rock-formingminerals. American Journal of Science, 2784, l-229.

Hem, J.D., and Roberson, C.E. (1967) Form and stability of aluminumhydroxide complexes in dilute solutions. U.S. Geological Survey WaterSupply Paper 1827-A.

Hemingrvay, B.S. (1982) Gibbs free energies of formation for bayerite,nordstrandite, AI(OH),*, and AI(OH)|, aluminum mobility, and theformation of bauxites and laterites. In K.S. Saxena, Ed., Thermody-namic methods and data, p. 285-314. Springer-Verlag, New York.

Hemingrvay, B.S., and Robie, R.A. (1977a) Enthalpies of formation oflow albite, NaAlSi3OB, gibbsite, (AI(OH),), and NaAlO,; revised valuesfor AHtr* and AGfr* of some aluminosilicate minerals. U.S. GeologicalSurvey Journal of Research, 5, 413-429.

Herningway, B.S., and Robie, R.A. (1977b) The entropy and Gibbs freeenergy of formation of the aluminum ion. Geochimica et Cosmochimi-ca Acta,41, l4O2-14O4.

Hemingway, B.S., Robie, R.A., Fisher, J.R., and Wilson, W.H. (1977)The heat capacities ofgibbsite, AI(OH)3, between 13 and 480 K andmagnesite, MgCOr, between I 3 and 380 K and their standard entropiesat 298.1 5 K and the heat capacities ofCalorimetry Conference benzoicacid between 12 and 316 K. U.S. Geological Survey Joumal of Re-search. 5. 797-806.

Hemingway, B.S., Robie, R.A., and Kittrick, J.A. (1978) Revised valuesfor the Gibbs free energy of formation of [AI(OH)*], diaspore, boehm-ite and bayerite at 298. I 5 K and I bar, the thermodynamic propertiesof kaolinite to 800 K and I bar, and the heats of solution of severalgibbsite samples. Geochimica et Cosmochimica Acta,42, 1533-1543.

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE

Hemingway, 8.S., Krupka, K.M., and Robie, R.A. (1981) Heat capacitiesofthe alkali feldspars between 350 and 1000 K from differential scan-ning calorimetry, the thermodynamic functions of the alkali feldsparsfrom 298.15 to 1400 K, and the reaction qtraftz + jadeite: analbite.American Mineralogist, 66, 1202-1215.

Hemingway, B.S., Robie, R.A., Kittrick, J.A., Grew, E.S., Nelen, J.A.,and I-ondon, D. (1984) The thermodynamic properties oftwo naturalchlorites to 500 K the heat capacities ofosumilite from 298.15 to 1000K, and the thermodynamic properties of petalite to 1800 K. ArnericanMineralogist, 69, 7Ol-7 10.

Hemingway, B.S., Kittrick, J.A., and Peryea, F.J. (1989) Relative solu-bilities of corundum, gibbsite, boehmite, and diaspore at standard stateconditions: An addendum. Clays and Clay Minerals, 37 , 566-567.

Hovey, J.K., Hepler, L.G., and Tremaine, P.R. (1988) Thermodynamicsof aqueous aluminate ions: Standard partial molar heat capacities andvolumes of AI(OHI (eQ from l0 to 55 lC. Journal of Physical Chem-istry,92, 1323-1332.

Hsu, P.H. (1967) Effect ofsalts on the formation ofbayerite versus pseu-

doboehmite. Soil Science, 103, 101-110.Hsu, P.H., and Bates, T.F. (1964) Formation of X-ray amorphous and

crystalline aluminum hydroxide. Mineralogical Magazine, 33, 7 49-7 68.Johansson, Georg (1960) On the crystal structures of some basic alumi-

num salts. Acta Chemica Scandinavica, 14,771-773.Johansson, G. (1962) The crystal structures of [AI'(OH),(FI'O)E](SO"),'

2H,O and [Al, (OH)'(H,O)8](SeO.),' 2H'O. Acta Chemica Scandinav-ica.16.403-420.

Johansson, G., Lundgren, Georg, Sill6n, L.G., and Sdderquist, Rune (1960)

On the crystal structure of a basic aluminum sulfate and the corre-sponding selenate. Acta Chemica Scandinavica, 14, 7 69-7 7 1.

Kelley, K.K., and King, E.G. (1961) Contributions to the data on theo-retical metallugy. XIV. Entropies of the elements and inorganic com-pounds. U.S. Bureau of Mines Bulletin 592, 34.

Kittrick, J.A. (1966) The free energy of formation of gibbsite andAl(OFI)r from solubility measurements. Soil Science Society of Amer-ican Proceedings, 30, 595-598.

-(1969) Soil minerals in the AlrOr-SiOr-HrO system and a theoryof their formation. Clays and Clay Minerals, 17 , 157 -167 .

May, H.M., Helmke, P.A., and Jackson, M.L. (1979) Gibbsite solubilityand thermodynamic properties of hydroxy-aluminum ions in aqueoussolutions at 25 qC. Geochimica et Cosmochimica Acta, 43, 861-868.

Mukaibo, T., Takahashi, Y., and Yamada, K. (1969) The heat capacityand the heat of dehydration of the hydrated aluminas. Proceedings,First International Conference on Calorimetry and Thermodynamics,Warsaw, 375-380.

Parks, G.A. (1972)Free energies offormation and aqueous solubilities ofafuminum hydroxides and oxide hydroxides ar 25 t. American Min-eralogist, 57, I 163-l 189.

Perkins, Dexter III, Essene, E.J., Westrum, E.F., Jr., and Wall, V.J. (1979)

New thermodynamic data for diaspore and their application to thesystsm AlrO3-SiO,-HrO. American Mineralogist, 64, 1080-1090.

Peryea, F.J., and Kittrick, J.A. (1988) Relative solubilities of corundum,gibbsite, boehmite, and diaspore at standard state conditions. Claysand Clay Minerals, 36,391-396.

Robie, R.A., and Hemingway, B.S. (1972) Calorimeters for heat of solu-tion and low-temperature heat capacity measurements. U.S. GeologicalSurvey Professional Paper 755,32 p.

Robie, R.A., Hemingway, B.S., and Wilson, W.H. (1976) The heat ca-pacities of Calorimetry Conference copper and of muscoviteKAl, (AlSi3)Oro(OH)' , pyrophyl l i te Al ,Si .O'o(OH), , and i l l i teK3(MgAl,)(Si,oAlJOoo(OH)' between l5 and 375 K and their standardentropies at 298. I 5 K. U.S. Geological Survey Journal of Research, 4,631-644.

Robie, R.A., Hemingway, B.S., and Fisher, J.P. (1979) Thermodynamicproperties of minerals and related substances at 298.15 K (25 oC) andone bar (105 pascals) pressure and at higher temperatures (revised

1979). U.S. Geological Survey Bulletin 1452, 456 p. (originally printed

in 1978).Ross, G.J., and Tumer, R.C. (1971) Etrect of different anions on the crys-

tallization ofaluminum hydroxide in partially neutralized aqueous alu-minum salt systems. Soil Science Society of America Proceedings, 35,389-392.

Rossini, F.D., Wagman, D.D., Evans, W.H., Levine, S., Jaffe, I. (1952)Selected values of chemical thermodynamic properties. U.S. NationalBureau ofStandards Circular 500, 1268 p.

Russell, A.S., Edwards, J.D., and Taylor, C.S. (1955) Solubility of hy-drated aluminas in NaOH solutions. Journal of Metals, 203, ll23-r128.

Sanjuan, Bernard, and Michard, Gil (1987) Aluminum hydroxide solu-bility in aqueous solutions containing fluoride ions at 50.C. Geochimi-ca et Cosmochimica Acta, 51, 1823-1831.

Schoen, R., and Roberson, C.E. (l 970) Structures ofaluminum hydroxideand geochemical implications. American Mineralogist, 5 5, 43-7 7 .

Shomate, C.H., and Cook, O.A. (1946) Low-temperature heat capacitiesand high temperature heat contents of Al,O, 3HrO. American Chern-ical Society Journal, 68, 21 40-21 42.

Singh, S.S. (1974) The solubility product ofgibbsire at 15, 25, and 35 qC.

Soif Science Society of Arnerica Proceedings, 38, 415-417 .Sjiiberg, Staffan, and Ohman, L.-O. (1985) Equilibrium and srructural

studies ofsilicon (IV) and aluminum (III) in aqueous solution. Part 13.A potentiometric and ,?AI nuclear magnetic resonance study of speci-ation and equilibria in the aluminum (Ill)-oxalic acid-hydroxide sys-tem. Joumal Chernical Society, Dalton, 2665-2669.

Smith, R.W., and Hem, J.D. (1972) Etrect of aging on aluminum hydrox-ide complexes in dilute aqueous solutions. U.S. Geological Survey Wa-ter-Supply Paper, 1827-D, 51 p.

Thomas, A.W., and Whitehead, T.H. (1931) Ion exchanges in aluminumoxychloride hydrosols. Joumal of Physical Chemistry, 35, 27-47 .

Tsai, P.P., and Hsu, P.H. (1984) Studies of aged OH-AI solutions using

457

kinetics of Al-ferron reactions and sulfate precipitation. Soil ScienceSociety ofAmerica Journal, 48, 59-65.

-(1985) Aging of partially neutralized aluminum solutions of sodi-um hydroxide/aluminum molar ratio : 2.2. Soil Science Society ofAmerica Journal. 49. 1060-1065.

Turner, R.C. (1969) Three forms of aluminum in aqueous systems deter-mined by 8-quinolinolate extraction method. Canadian Journal ofChemistry, 47, 2521-2627.

Turner, R.C., and Ross, G.T. (1970) Conditions in solution during theformation ofg,ibbsite in dilute AI salt solutions. 4. Effect ofCl concen-tration and temperature and a proposed mechanism for gibbsite for-mation. Canadian Journal of Chemistry, 48, 7 23-7 29.

Verdes, G., and Gout, R. (1987) Rehydration d'hydroxydes d'aluminiumamorphes. Application i l'6tude de l'6quilibre boehmite-bay6rite. Bul-letin de Min6ralogie, ll0, 579-587.

-(1988) Determination of thermodynamical properties of alumi-nate ion and ofmain aluminum oxides and hydroxides (bayerite, gibbs-ite, boehmite, diaspore and corundum) (abs). TERRA cognita, 8, 180.

Violante, A., and Violante, P. (19E0) Influence ofpH, concentration, andchelating power of organic anions on the synthesis of aluminum hy-droxides and oxyhydroxides. Clays and Clay Minerals, 28,425-435.

Wesolowski, D.J., Palmer, D.A., and Begun, G.M. (1990) Cornplexationof aluminate anion by Bis-Tris in aqueous media at 25-50 lC. Joumalof Solution Chemistry, 19,159-173-

MnNuscnryr REcETvED Apnrl 19, 1990Mexuscnrrr AccEprED Jet.rueny I l, l99l

HEMINGWAY ET AL.: THERMODYNAMIC PROPERTIES OF BOEHMITE


Recommended