American Mineralogist, Volume 76, pages 1722-1727, I99I

Brushite, hydroxylapatite, and taranakite from Apulian caves(southern ltaly): New mineralogical data

S,q,vnRroFronr Istituto di Ricerca sulle Argille, C.N.R., Via Nazario Sauro, 85, 85100 Potenza,Italy Rocco L.lvrl'No Dipartimento Geomineralogico,Universiti degli Studi di Bari, Campus Universitario, 70124 Bai,ltaly

Ansrn-lcr Samples of brushite, hydroxylapatite, and taranakite collected from seven karst caves of Apulia (southern Italy) have been investigated by means of chemical analysis and by DTA and XRD techniques. Brushite is found to crystallize in very pure form, whereas hydroxylapatite and taranakite show severalisomorphous substitutions.A high Zncontent was detected in hydroxylapatite. The TG curves for one sample of hydroxylapatite show a weight loss greater than the total HrO content, probably becauseof the loss of S. DTA and XRD analysesof taranakite indicate that the formation of AIPO4 at 1200 "C is not a reversible reaction. In accord with previously published parageneticsequences, the were probably formed by interaction betweenphosphatic solutions derived from bat and the sub- stratum minerals, specifically calcite and clay minerals. However, the data also suggest precipitation from solution for brushite and hydroxylapatite.

IxrnooucrroN careoussubstratum. In some caves,other phosphatessuch as ardealite' francoanellite' ' ' and vivi- Little has been reported (Murray and Dietrich , 1956; anite are present' as shown by X-ray powder diffraction Balenzano etal.,lg74;Sakae and Sudo, 1975) "";;;* p"IiTl- the mineralogical properties of brushite, hydroxylapatite, The samples for analysis were separatedand purified and taranakite associatedwith bat guano depol;tT;;; un{e1 binocular microscope by hand picking' and the cavesbecause rhey occur rarely and sampling ir alffi-;;i. grains were1 then cleanedby a weak ultrasonic treatment' In this study, several samplescollected fto- ta""" ep"- Refractive indices were determined in calibrated oils (es- lian karst caves (southern Italy) are analyzed. timated elror : +0.002). X-ray analyseswere performed Moon oF occuRRENcE AND using a Philips powder diftactometer (PW 1710) with EXpERTMENTALMETH9D5 Ni-filtered CuKa radiation and 0.01" 20 step scans. was 30 s per step' and a NaF internal The sampled caves are in a cretaceous limestone that 9:lltttt ]*e emploved' X-rav powder diffraction pat- constitutes the Apulian Platform (southern It;il;. il; :9di:1:-"t terns ofpreviously heatedsamples were recordedat room caveswere selectedbecause they were uccettiUf"-inO uf- cell parameters were refined by the least- preciable amounts of bat guano were present. rn" r"*tb" 9-p"*1-:1: soulles m11hod' chemical analyseswere obtained using of the caves, their description, and io"tttm"utioo "oo" x-31 spectrometer (Philips PW l4l0) number are reported by orofino (1965). All A;-;t forT major-elementslyorescence and atomic absorption spectropho- have high humidity and a large amount orout"pLiguu- tometry (Perkin Elmer 5100) for minor and trace ele- no; only cave PU l2ll presents a comparatl""iv tiiJr ments. Hro was measured using the Penfield method. mass of guano. otl1i9 curves were obtained using a Netzsch si- Brushite (hereafter BR) occurs as a continuous paste- To apparatus (srA 409) under the fol- like layer up to 20 cm thick betweenguano and ilJr;;; loynqllt-t:l:.'hermal conditions: thermocouples' Pt-Pt/Rh; or as very small pulverulenr massesirregulary di;;;;;;; -1ofi"c heating rate' 5 .C per min; DTA sensi- rhroughout the guano. Hydroxylapatite (hereaft"eiilPj fry.*t::lAt,O,; TG sensitivitv' 25-50 mg; sample weight' is frequently found as small grains (0.5-1.5 -;; ;il; lXi?;^t9!jv;50-100 mg; staticairatmosphere' the brushite layer and rarely as thin crusts oi'"ut"iti" concretions that underlie the guano. BR is always more Rnsur_rs abundant than HAP, except in cave PU l2l I where we Brushite have found the inverse relation. Taranakite (hereafter TR) forms irregularaggregates within the insolubleresiduum Brushite, CaHPOo'2IJ.O, forms cryptocrystalline of limestone ("terra rossa") and also occurs within the pastelike aggregatesvaryrng in color from white-ivory to guano as fine layers (2-3 mm thick) parallel to the cal- yellow-ivory; in cave PU 32 only it occurs as colorless 0003-004x/91/0910-1722$'02.00 1722 FIORE AND LAVIANO: BRUSHITE. HYDROXYLAPATITE. AND TARANAKITE 1723 platey crystals 0.64.8 mm in size. The is opti- Tracings of the DTA and TG curves of HAP are given cally biaxial with refractive indices and 2V values that in Figure l. The DTA curve shows only a broad exo- fall within narrow ranges.In the several samplesexam- thermic effect between ll00 and 1200 oC, whereas the ined,values of a : 1.540,B: 1.546,t: 1.552,and2V TG curve gradually declines.On the basis of powder dif- = 90ohave been frequently observed.This homogeneity fraction patterns ofthe products heatedat 1000 and 1400 is related to the extreme chemical purity of the mineral. "C, we suggestthe following reactions: Six samplescollected from different caveshave given the 2[Ca,(PO"),OH] following mean values (wto/ot o): CaO : 33.12 t 0.98, - PrO,: 40.53+ 0.87,HrO :26.19 + 0.31;only Fe, Zn, 210-Ca'(Poo),1() + CaoPrOn and S are present as trace elements(< 100 ppm). + lH,O (1000 "c) The X-ray powder diffraction patterns, indexed on a B-Car(POo),- a-Ca.(POo),(1400 "C)' monoclinic cell with spacegroup (PDF C2/c 9-77), dem- A greaterweight loss has been recorded for sample PU onstrate that the samples are well crystallized and have l2ll/2, probably becauseofreleased SO.. very similar d-values. The differencesbetween the unit- cell parametersof the sevensamples investigated are very slight, with mean values:a : 6.361(3),b : l5.l9l(4), c Taranakite : 5.814(2),0: tt8.4s(4). At present, the structure of TR is unknown, even though In accord with previous authors (Murray and Dietrich, severalstudies have been carried out on both natural and 1956;Balenzano et al., 1974),theDTA curvesshows two syntheticproducts (Haseman et al., 1950, l95l; Murray 'C endothermicpeaks (195 and 410 "C) from the reac- and Dietrich, 1956; Smith and Brown, 1959; Balenzano tions CaHPO'.ZH,O - CaHPOo(monetite) + l2HrO and et al., 19741-Sakae and Sudo, 1975; Filipov, 1978a). 2CaHPOo - CarP,O, (amorphous) + 1HrO, with weight The chemical formula most frequently adopted is lossesof 210loand25o/o, respectively; the exothermic peak H6K3AI5(PO")8.18H2O,as given by Smith and Brown (at 495 'C) results from the transformation CarPrO, (1959) who also proposed a hexagonalsix-layer structure (amorphous) - CarP,O, (crystalline). Only slight differ- with a : 8.71 A, c : 96.1 A, and Z: 6. McConnell encesin temperature (a 15 "C) and peak sharpnesswere (1976) suggestedthat TR could be considered to be a recorded among the different samples. phyllophosphate rather than orthophosphate; he pro- posed a hexagonalstructure, with a: 8.71 A and c: 96.1 A, composedof six doublelayers of 4[(X.O10XOH)r], Hydroxylapatite where X: P, Al, or (Hr). McConnell assignedthe chem- Hydroxylapatite [Car(POo)rOH]always appearsas small ical formula I("[Al, ,(H3),](OH)r[AI,P.-, ,(H3),O,0]to TR yellowish or brown grains ranging from 0.5 to 1.5 mm in in analogy to that of illite, which is one of the reactants size.Although thesegrains are frequently included within from which TR can be formed. However, Fleischer(1987) the BR masses,an intimate associationbetween the two reports the formula KAI3(PO4)3OH'9HrO but assigns minerals never has been observed.The mineral generally francoanellite (Balenzano et al., 1976), a mineral struc- forms yellow-brown lumps or occasionally pale yellow turally relatedto TR, the formula H6KAI5(PO.)8'l3HrO. plates that are uniaxial negative and display weak ple- Becausethere is no well-establishedstructure of the min- ochroism. The refractive index values, measuredonly on eral and in consideration ofour data, we have decided to two samplesbecause the crystalsare too small to provide adopt the suggestionof Smith and Brown (1959) so as reliabledata, are c,r: 1.651and e : 1.645for samplePU not to create further confusion and to preservethe anal- 502/3,a: 1.638and e : 1.628for samplePU l2ll/2. ogiesbetween the formulae of TR and francoanellite. X-ray powder diffraction patterns of the samples(Table The aggregatesof TR from Apulian caves form as yel- l) exhibit very different d-values.Indeed, the crystallinity Iowish white, claylike, soft nodular masses.Under the index (CI), calculated as suggestedby Simpson (1964), microscope,the mineral appearsas a fine-grainedpowder indicatesa continuousvariation from low (CI:0.10 for and rarely as pseudohexagonal,thin colorlessplates that PU 7l/5) to very high crystallinity (CI : 0.03 for PU are uniaxially negative. Several determinations of refrac- l2ll/2) (Table l). Although the aand,c valuesvary from one specimento another, the c/q ratio (Table l) is almost constant and close to those ofhydroxyl-bearing varieties bQ2 of apatite(Deer et al., 1965). o The Apulian HAP shows limited substitution of Mg, Zn, Na, and K for Ca and S for P. Moreover, all the (25-134 ppm), (217-575 ppm), Ro samplescontain Sr Cu = Mn (346-867 ppm), and Cl (35-54 ppm) as trace ele- 2 ments. It is interesting to note the relatively high Zn con- tent (>0.550/oZnO), which, in 200 400 600 800 1000 1200 accordancewith Balenzano Temperature,C et al. (1974), seems a typical feature of samples from Apulian caves. Fig. l. DTA and TG curvesofhydroxylapatite. 1724 FIORE AND LAVIANO: BRUSHITE, HYDROXYLAPATITE, AND TARANAKITE

TABLE1. Hydroxylapatite:Data from X-raypowder diftraction patterns, cell parameters, and formulas PU71t5 PU 394t2 PU502/3 PU1211t2 ilh nh ilh ilh 100 8.19 14 8.13 9 8.15b 10 8.18 15 101 5.28 c 5.26 6 5.25 2 110 4.73 4 4.73 2 200 4.05 5 4.07b 2 4.08 7 111 3.90 10 3.89 17 4.02b 2 3.88 I 201 3.50b I 3.50 4 002 3.43 51 3.43 43 3.44 43 3.44 44 102 3.17b 10 3.16b 17 3.16b 4 3.16 I 210 3.09 16 3.07 32 3.08b 13 3.09 16 211 2.821 100 2.812 100 2.825 100 2.81',l 100 112 2.787 93 2.778 76 2.789 100 2.779 68 300 2.727 50 2.724 55 2.736 63 2.717 60 202 2.635 24 2.629 22 2.633 21 2.628 26 301 2.541 6 2.529b 6 2.531 I 2't2 2.300b 12 2.295 12 2.302b 9 2.292 7 310 2.272 24 2.266 20 2.268b 17 2.261 18 221 2.234 o 2.224 2 2.231 3 311 2.157 4 2.146 7 2.148 I 2.'145 6 302 2.137 4 2.129 3 113 2.066 4 2.061 o 2.058 5 400 2.043 4 2.042 2 2.042 1 203 1.996 5 2.000 6 1.993 4 222 1.948 20 1.945 25 1.948 22 1.941 24 312 1.891b I 1.891 12 1.893 10 1.887 10 320 1.871 6 1.872 3 1.877 o 1.868 4 213 1.839 33 1.841 30 1.845 27 1.839 30 321 1.807 22 1.809 6 1.803 12 410 1.786b I 1.781 7 1.782 6 1.778 10 402 1.753b 8 1.755 6 1.759 I 1.753 10 004 1.718 19 1.721 25 1.723 17 1.719 13 104 1.686 4 1.681 3 322 1.646 o 1.644 4 a (A) 9.458(6) 9.419(5) 9.,148(6) 9.410(2) c(A) 6.86s(4) 6.88i'(4) 6.891(7) 6.876(2) cla o.726 0.731 0.729 0.731 cl 0.10 0.08 0.09 0.03

PU7115 (Cas 5oMgoBZno @Nao6Ko 6XPs eso ,4)Or4,lOH), s PU 39412 (Cae ToMgoqzno 2,Naor|G o,[P" *)O." *(OH), * "rSo PU 502i3 (Cae sMgo qzno o,Nfu qKo 01XP5ezSo @)Or. dOH), m PU 121112 (Ca" rrMgo o,Znoo,Nao *l$ *[P" *So,s)OB efOH)2 @ Note.'The formulas have been calculatedon the basis of 26 (O,OH).Cl: crystallinityindex. tive indices for different sampleshave shown values very patterns ofthe products heat-treatedat 140 and 300'C closeto <,r: 1.508,e : 1.502. indicate that the former product is francoanellite,where- X-ray powder patterns of TR (Table 2) show sharp as the latter is noncrystalline. Between 300 and 500 "C, diffractions and a very strong (006) basal reflection. The the product gradually loses more weight and is wholly values of c increaseslightly with HrO content (Table 2), dehydrated according to the reaction HuKrAlr(POo)r - and ifone also takesinto accountthe published data from K3Al5P8O2e+ i3HrO. The exothermic peak (565-595 "C) Apulian TR (Balenzano et al., 1974), this tendency is resulted from the crystallization of two new phasesac- more evident (r : 0.809 and n: 7, where r is the cor- cording to the reaction KAl5PO8Ore * 2AIPO+ + relation coefficient and n is the number of samples). 3KAlPrO?. In agreementwith Balenzanoet al. (1974) rhe However, other values (Sakaeand Sudo, 1975; Filipov, presenceof KAlPrOr, a compound not reported in the 1978a), do not fit this trend, and more data are needed JCPDS file, was assumedbecause the powder pattern of to verify the relationship. The Apulian TR shows, as do the product heat-treatedat 620'C contains the reflections almost all natural samples, very limited substitution of of a compound similar to KAlorFeorPrOr.Also, Sakae Ca and Na for K, Fe for Al, and S for P. Other elements, and Sudo (1975) reported a similar but uninterpreted such as Rb, Cu, Mn, Sr, N, are present at < 100 ppm. powder pattern. A broad deflection is presenton the DTA The thermal data (Fig. 2) suggestthat the two endo- curves between 750 and 980'C; it is more pronounced thermic events at 80-140 "C and 140-300 "C (see also for samplesPU 7l/4 and PU 502/2,which also show a Balenzanoetal.,1974; Sakaeand Sudo, 1975)result from very slight weight loss. Such an endothermic effect was the reactions H6K3AI5(PO.)8.lSHrO - H6KrAlr(POo)s. also recognized by Murray and Dietrich (1956) and by l3HrO + l5HrO and H6K3AI5(PO.)8.l3HrO Sakaeand Sudo (1975). The latter authors observed the H6K3AI5(PO,)8+ ll3Hro, respectively.X-ray diffraction disappearanceof the three main reflections (5.72 4,2.94 FIORE AND LAVIANO: BRUSHITE. HYDROXYLAPATITE. AND TARANAKITE 1725

TABLE2. Taranakite: Data from X-ray powder diffraction patterns, cell parameters, and formulas

PU38/1 PU71t3 PU71t4 PU7116 PU 502t2 ilh nh ilh vh ,th 006 15.82 100 15.57 100 15.89 t00 15.60 100 16.01 100 0012 7.92 18 7.84 17 7.92 36 7.85 't7 7.92 44 101 7.60 8 7.56b 6 7.61 26 7.61 7 7.59 12 012 7.47 28 7.39 19 7.45 30 7.38 13 7.43 32 104 7.18 5 7.12b 6 7.19 8 7.16 4 7.23 10 1010 5.91 17 5.87 7 5.91 13 5.86 I 5.92 24 0114 5.07 3 5.02 2 s.03 4 5.09 3 5.06 4 1016 4.67 4 4.63 3 4.67 4.63 3 4.68 5 110 4.35b 12 4.33 10 4.35 12 4.32 7 4.36b 18 113 4.32 18 4.29 12 4.32 17 4.31 8 4.33 20 1019 4.20 4 4.15b 6 4.17b 4 4.19 1 4.20 c 119 4.02b 7 4.00b 4 4.02 J 4.00 3 4.03 10 1112 3.82 40 3.80 12 3.82 24 3.81 17 3.81 35 202 3.75 11 3.73 I 3.75 10 3.73 5 3.75 13 208 3.59 22 3.57 8 3.59 12 3.57 I 3.60 22 0213 3.36 29 3.34 I 3.36 10 3.33 14 3.37 14 2014 3.30 14 3.29 7 3.30 o 3.28 5 3.30 14 0216 3.19 15 3.17 3.16 13 3.16 5 3.18 14 2017 3.14 31 3.13 12 3.14 22 3.13 15 3.14 35 0129 3.03b 2 3.03 c 3.05b 4 3.06 1 3.O7 2 1124 2.954 7 2.941 5 2.951 6 2.937 3 2.955 10 0033 2.926 7 2.885b 3 2.932 o 2.905 <1 2.936 7 122 2.842 13 2.833 7 2.843 13 2.828 8 2.845 13 125 2.819b 17 2.810 12 2.814b 17 2.806 9 2.819b 19 217 2.792 7 2.781b 6 2.785 8 2.777 3 2.795 6 2110 2.734 10 2.731 c 2.732 10 2.723 4 2.760 12 1211 2.710 3 2.703 4 2.704b 3 2.69s 1 2.711 o 2026 2.639 5 2.633 o 2.642b I 2.629 s 2.639 14 12't4 2.632 15 2.622 I 2.627 9 2.615 7 2.634 19 q 2116 2.568 2.s60 c 2.572 3 2.557 3 2.571 7 0228 2.538 4 2.533b 2 2.534 2 2.529 1 2.546 b 306 2.487 4 2.475b 2 2.477b 2 2.470 <1 2.486b 3 0039 0138 2.396 13 2.389 4 2.394 o 2.389 4 2.402 11 2032 2.349 3 2.343b 3 2.349 3 2.336 1 2.3s0 4 o234 2.261 6 2.262b 3 2.266 4 2.257 1 2.270 3 223 2.178 3 2.165b 2 2170b 2 2.169b <1 2.176 4 a (A) 8.716(2) 8.676(6) 8.688(10) 8.660(7) 8.710(6) c (A) 96.08(4) 95.s6{8) 96.2(1) 95.8(1) s6.4(1)

PU38/1 H6oo(G?oNao3oXAl46rFeo3s)[email protected] PU7113 H5s(K, *Nao @Cao1oXAl4 75Feo s)P6 osoe'18.13 HrO PU7114 H556(K2s6Naoo7cao*XAlr66Feo1s,[email protected] PU7116 Hs 16(K2sNao 6oao rTXAll75Feo sXPT erso z)Os. 1 8.69HrO PU 502t2 H55r(K2 67Nao 15oao roxAl4 61 Feo 12XP7 *So,JO* 1 8.89HrO Nofe.'The formulas have been calculatedon the basis of 32 O atoms and 16 cations

A, 2.90 A;-ttrat we have referred to as KAlP,O,-at and limestone. This mechanism explains the origin of the 1000 "C. They concluded that the reversibility of the Apulian BR, present as a soft layer under the guano, where modifications was the result of a rapid phase transfor- the guano is in contact with the limestone. However, the mation. Our results do not confirm these observations. small pulverulent aggregatesof BR within bat guano not X-ray diffraction patterns of the samplesheat treated at in direct contact with the limestone neither have fallen 1200'C and then quenched to room temperature show from the roof of the caves nor are the result of water that AIPO. is the predominant or the only phasepresent. transport ofcalcitic grains becauseoftheir irregular form Such discrepanciesmight be related to the partial loss of and distribution and the total absenceof calcite in the P. In any case,when it occurs, the transformation is not nodules. On the basis of this depositional evidence, we reversible. suggestthat BR could also be formed by a chemical re- action between a phosphatic solution and bicarbonatic H'O. DrscussroN AND coNclusroNs The origin of HAP is also not known, although a trans- The data obtained in this study provide further infor- formation from BR through octacalcium has mation on the mineralogical properties of BR, HAP, and been hypothesized for human stones. A review of this TR, which occur together in cave deposits.It is generally topic is given by Ohta and Tsutsumi (1982). Such a hy- acceptedthat BR is formed by interaction betweenphos- pothesis cannot explain the origin of HAP in caves be- phatic solutions, derived from the breakdown ofguano, causethe high chemical purity of BR documented herein t726 FIORE AND LAVIANO: BRUSHITE, HYDROXYLAPATITE, AND TARANAKITE

in a neutral or a slightly alkaline environment, in which HAP is a more stablephase than BR (Posneret al., 1984). All the elements that occur in the phases we analyzed are the constituents of substratum minerals or guano; the latter contains P and also Na, K, Ca, Mg, Al, Fe, and S, as reported by Martelli (1925) and $o Hutchinson (1950). Only the Zn in HAP presentsa prob- e20 lem becauseZn occursin small amounts both in the lime- _c stone and in the guano. It is unlikely that limestone is the g primary source ofZn becauseit is necessaryto hypoth- very great quantities limestone; in 0 esizedissolution of of fact Moresi (1975)reports a mean value of 148 ppm for Apulian "terra rossa." Therefore, although we have de- tected only 500 ppm of Zn in guano, the guano remains the only identifiable source. In summary, the chemical composition and physical 2oo * looo 12oo properties of BR from Apulian cavesare consistentwith t".nf"oiu*,3* the tendency of this mineral to crystallize in very pure form. In contrast, HAP and TR show several substitu- Fig.2. DTA and TG curvesof taranakite. tions. In HAP, Ca may be replaced by Zn, Mg, Na, and K, whereasP is replaced by S. In taranakite, Na and Ca may for K, Fe for Al, and S for P. The HrO does not reflect the presenceof foreign elements in the substitute variability HAP structure. Therefore, it is likely that the precipita- content of TR shows slight and seemsto in- tion of BR and HAP crystals occurs at the same time creasewith increasing c. The causesof the variations of directly from solution as will be discussedlater. the chemical composition of TR and HAP remain un- The mode of occurrenceof TR agreeswith the hypoth- known. product esis that this mineral is a of a chemical reaction Acxxowr-nocMENTs betweenclay minerals and a phosphatic solution derived from guano. Indeed, we have observedlower contents of We are gra.teful to L. Dell'Anna (Bari University) for the review of an early version of the manuscript. We are also indebted to J.M. Hughes illite and kaolinite in "terra rossa" at the contact with (Miami University) and H.C.W. Skinner (Yale University), whose com- TR. Theseclay minerals are also characterizedby smaller ments and suggestions have improved the paper. This study was finan- crystallinity indices than where TR is absent. The occur- cially supported by Consiglio Nazionale delle Ricerche. rence of TR as fine layers in the guano may be because the circulation of HrO transports significant quantities of RnrnnnNcps crrED clay minerals from "terra rossa." Balenzano,F., Dell'Anna, L., and Di Pierro, M. (1974) Ricercheminera- The coexistenceof the minerals studied gives some in- logiche su alcuni fosfati rinvenuti nelle Grotte di Castellana (Bari): idrossiapatite. formation about cave genetic environments. BR and HAP Strengite alluminifera, vivianite, taranakite, brushite e Rendiconti della SocietA Italiana di Mineralogia e Petrologia, 30, 543- form in solutions having acidic and alkaline pH, respec- 573. tively, and the boundary between their fields of nucle- - ( 1976)Francoanellite, H6K3AI5(PO4)8 I 3HrO, a new mineral from ation lies in the pH range6.2-6.8 (Simpson,1964). Both the caves of Castellana, Puglia, southern ltaly. Neues Jahrbuch fiir the modes of occurrence and the larger amount of BR Mineralogie Monatschefte, 2, 49-57 . Boistelle, R., I-6pez-Valero,I. (1986) Nucleation and growth ofcal- pH and compared to HAP might result from a more favorable cium and magresium . In R. Rodriguez-Clemente and Y. to the nucleation of BR. In fact, the small amounts of Tardy, Eds., Proceedings International Meeting: Geochemistry ofthe HAP cannot be related to the presenceof a high Mg con- Earth Surfaceand ProcessesofMineral Formation, p. 8ll-827. CSIC, tent, which can inhibit HAP nucleation (Boistelle and Madrid. (1965) L6pez-Valero, 1986),because no magnesium phosphates Deer, W.A., Howie, R.A., and Zussmann,J. Rock-forming min- erals, vol. 5, p. 323-338. Longmans, London. were found. An acidic pH is also indicated by the pres- Elliot, J.S., Sharp, R.F., and kwis, L. (1959) The effect of molar Ca./P enceofTR becauseit precipitatesand is stablein a highly ratio upon the crystallization of brushite and apatite. Joumal of Phys- acidicenvironment (Haseman et al., 1950, l95l; Filipov, ical Chemistry, 63, 725-7 26. 1978a, 1978b). Therefore, it is likely that the minerals Filipov, A. (1978a) Taranakite from two deposits in Bulgaria. Ann,,rir€ de I'Universite de Sofra, 70, 287-298 (in Bulgarian). studied were formed at low pH, and HAP precipitated -(1978b) Thermodlnamic analysis of the genesisof taranakite in successivelyonly when the CalP molar ratio was favor- the system(K"NHo)rO-CaO-AlrO3-PrOr-HrO. Annuaire de I'Universite able (Elliot et al., 1959;see Flicoteaux and Lucas, 1984, de Sofia, 70,299-307 (in Bttlgarian). p. 292). Such a hypothesis might not be applicable for Fleischer, M. (1987) Glossary of minerals species(5th edition), 234 p. Mineralogical Record, Tucson, Arizona. cave PU l2l I becauseTR is absent.Furthermore, taking Flicoteaux, R., and Lucas, J. (1984) Weathering of phosphateminerals. into account that HAP is more abundant than BR and is In J.O. Nriagu and P.B. Moore, Eds.,Phosphates minerals, p. 292-317. more crystalline, it is likely that in this cave it crystallized Springer-Verlag, Berlin. FIORE AND LAVIANO: BRUSHITE, HYDROXYLAPATITE, AND TARANAKITE 1727

Haseman, J.F., Brown, E.H., and whitt, C.D. (1950) Some reaction of phology of brushite crystal glown rapidly in silica gel and its structure- phosphate with clay and hydrous oxides of and aluminum. Soil Journal of Crystal Growth, 56, 652458. Science,70, 257-271. Orofino, F. ( I 965) Elenco delle grotte pugliesi catastate frno al 3 I gennaio Haseman,J.F., I*hr, J.R., and Smith, J.P. (1951) Mineralogical character 1965. RassegnaSpeleologica Italiana, 17, 59-81. ofsome iron and aluminum phosphatescontaining potassiumand am- Posner,A.S., Blumenthal, N.C., and Betts, F. (1984)Chemistry and struc- monium. Soil ScienceSociety of America Proceedings,15,7G84. ture of precipitated . In J.O. Nriagu and P'B' Moore' Hutchinson, G.E. (1950) The biogeochemistryof vertebrate excretion. Eds., Phosphatesminerals, p. 330-350. Springer-Verlag,Berlin. Bulletin Arnerican Museum of Natural History, 96, 275 (not seen;ex- Sakae,T., and Sudo, T. (1975) Taranakite frorn the Onino-Iwaya lime- tracted from American Mineralogist, 37,905, 1952). stone cave at Hiroshima Prefecture, Japan: A new occurrence. Amer- Martelli, D. (1925) Guani. Nuova enciclopedia di chimica scientifica, ican Mineralogist, 60, 331-334. tecnologicae industriale, vol. 13, p. 353-359. UTET, Turin, Italy. Simpson, D.R (1964) The nature ofalkali carbonateapatites' American McConnell, D. (1976) Hypothetical phyllophosphate structure for tarana- Mineralogist, 49, 363-376. kite. American Mineralogist, 61, 329-331. Smith, J.P., and Brown, W.E. (1959) X-ray studiesof aluminum and iron Moresi, M. (1975) Elementi in tracce di "terre rosse" pugliesi: II tenori phosphates containing or ammonium. American Mineral- di Zn. Periodico di Mineralogia, 44,69-102. ogsst,44, 138-142. Murray, J.W., and Dietrich, R.V. (1956) Brushite and taranakite from Pig Hole Cave, Giles County, Virginia. American Mineralogist, 41, 270-280. M.c.NuscRrprREcErwD Aucusr 7, 1990 Ohta, M., and Tsutsumi, M. (1982) The relationship betweenthe mor- M,c.NuscnrprAccEF.rED Mnv 17, 1991