crystal chemistry of the basic iron phosphates paul b ... · the basic phosphates of the transition...

THE AMERICAN MINERALOGIST, VOL. 55, ]ANUARY-FEBRUARY, 1970

CRYSTAL CHEMISTRY OF THE BASIC IRON PHOSPHATES

Paul B. Moono, The Department of the Geophysical Sci,ences,The Un'ivers,ity of Chicago, Chicago, Ill,i,nois, 60637.

Assrnacr

A general structural principle is widespread among the basic phosphates of ferrous andferric iron. It is based on a highly stable polyatomic complex involving ferrous-ferric oxy-hydroxy octahedral face-sharing triplets. The orientation of the associated corner-linkednearest neighbor octahedra and tetrahedra is so similar in several mineral structures thata general hierarchy of structure types has been derived. Crystal structures of the variousmembers are dictated by the ways in which the fundamental poiyatomic octahedral three-cluster and its nearest neighborhood of octahedra link along a third "variable" axis.

The refined crystal structures of dufrenite and rockbridgeite are revealed. Structurecell and indexed powder data are presented for dufrenite, souzalite, rockbridgeite, berau-nite, laubmannite, and mineral A (a new species). Data in the literature are misleadingand largely in error for most of these compounds. Members of the basic ferrous-ferricphosphate family knorvn to possess the octahedral face-sharing three-cluster includebarbosalite, Iipscombite, dufrenite, souzalite, rockbridgeite, beraunite, laubmannite, andmineral A. Other compounds which may also involve the three-cluster include azovskite,cacoxenite, richellite, mitridatite, egudite, tinticite, borickyite, and foucherite.

All basic ferrous-ferric phosphates which possess the three-cluster are strongly pleo-chroic and greenish-black in color. Their completely oxidized equivalents are yellow to red-dish-brown and weakly pleochroic. The relative absorption of polarized white light appearsto be dictated in a complex way by the orientation of the three-cluster in the crystal, theextreme pleochroism resulting from directed charge transfet psz+fFer+-=:ps:+fpsz+.

Paragenetic schemes involving some of these minerals are proposed on the basis ofspecimen study and crystal chemistry. Typical sequences involve denser phases followedby higher hydrates. Typical steps in the hydrothermal reworking of the early phosphatesrnvolve laubwannite d.uJrmi.te beraunite; rockbri.d.geite-beraunite; rockbrid.geite'mineral,4. Color banding along the fibers in mammillary aggregates of some rockbridgeites resultsfrom the appearance of different phases. Typical species and their colors are: rockbridgeite(greenish-black), Iipscombite (bluish-green granular), mineral A (brown to greenish-yellow), and beraunite (orange).

INrnotuctroN

The basic phosphates of the transition metal ions (in particular thoseof Fe2+, Fe3+, Mn2*, and Mn3+) are among the most perplexing substancesin the mineral kingdom. Many scientists have contributed to the subjectyet many species still remain insufficiently characterized. The problemis complicated, since variable oxidation states of the, metals may occurfor the same species. Furthermore, their frequently fibrous habit allowsvariable water content and nonessential impurities; their chemical com-positions are often so similar that diagnostic features, such as hardness,density, color and optical properties overlap. The problem has even per-sisted throughout the course of early to recent X-ray diffraction studies,

135

136 PAUL B. MOORE

and for simple reasons: many species are fibrous so that suitable singlecrystals are dificult to obtain and perfect cleavage directions contributetoward preferred orientation in cylindrical and slide powder mounts.

A detailed paper by Frondel (1949) serves adequately to outline the

tumultuous history in this research. That paper also offers considerableinsight into the problem and the crystallochemical arguments thereinare quite noteworthy, especially in light of the paucity of crystal celldata available at that time. Unfortunately, the most important diagnos-tic properties-the X-ray powder patterns-are misleading and referenceto them should be abandoned.

Additional light was shed by a paper on Fe-Mn orthophosphate hy-

drate classification by Moore (1965a). This paper considered these andrelated compounds as coordination complexes and focused attention on

the importance of the octahedral clusters which were treated as poly-

atomic complexes in these structures. Advances in the crystal chemistry

of these compounds then proceeded in normal fashion, through the sys-

tematic analysis of their crystal structures. A preliminary account of thispresent study appeared in Moore (1969).

The list of basic iron phosphates (Table 1) is very extensive, compris-ing no less than 40 species. Single hand specimens may show as many

as a dozen species in close association, especially if the material was de-rived from hydrothermally reworked triphylite-lithiophilite. As many as30 species are known from the reworked triphylite pods at the Palermo

No. 1 Pegmatite, North Groton, New Hampshire; a remarkable collec-tion of thousands of Palermo specimens, assembled by the Iate Mr' G.Bjareby and presently owned by Prof . J. V. Smith, ofiers a unique oppor-tunity for detailed study of their relationships.

Two other noteworthy kinds of basic iron phosphate occurrences in-cludes druses, knobs and bands in gossans, "Iimonite" and novaculitebeds; and iron-rich nodules in marls, sands, and soils. The former are

derived from phosphatic solutions which reworked the ferric oxy-hydrox-ides, and the latter are usually associated with organic remains.

The species discussed in detail in this paper include dufrenite, souza-lite, rockbridgeite, laubmannite, beraunite, cacoxenite, mineral A,

Iipscombite, and barbosalite. Since the subject matter is rather in-volved and interdependent, the relevant species are briefly discussedindividually, but their relationship to the general family of basic ironphosphates is given especial emphasis under the general headings. Much

of this work was inspired by the formal solution of the atomic arrange-ments of dufrenite and rockbridgeite; these crystal structures are dis-

cussed in detail.

r37

U

BASIC IRON PHOSPHATES

F

s - 5

!

. 2 4

t o

q.,

" . " . 9 P q ' E E i l " . E , ? o o. ; q . : . ; . 5 g I . E I 9 H E s , . I E ' : , , 9 ' : . ;i iE E E E E H.#E H g E€ E:E ' !€ !#+EEg ! 3 € € H ;3 9 E 'q € : J E $ E E 3E E:

+

h cE oq o t

! a' 2 E - - a0 g u b :H F . , 9 9 f iU O i n M @ :o x ! d o , i , h

? . E 5 E G ? *

-N

^ i i ix A v v : 1

- - . O A : : ;. i : a X ^ - ^

, o a , q . \ / | : t ] t 1

3d*sd ig

€

i € . 9 E f . * ex b 3 H 5 E E€ H E E ! E To - e i H s u rd d d d c d d

.=o * q E ' - .

E q ! : ; ' = . :3 F . E E A F F. 3 : E " € E : i! A € S q : >

,q

a

z

E

O

^|l

./.za

a

r

H

FA

A2

ts

il(,

F

138 PAUL B. MOORE

ExprnrueNte,l Procrlunr

crysta] ce)tr Data. single crystals of most basic iron phosphates are very infrequently en-

countered, since the species usually occur as fibrous aggregates' The crystals, when they do

occur, ars small andiurved; larger crystals are usually bunched or crinkled, prohibiting

unambiguous interpretation of the complex single crystal photographs. The crystal cell

data in Table 2 were obtained by splitting fibers in acetone until an optically single plate-

let was found. It was observed that single individuals are clear and transparent but crystal

fragments, though seemingly single in outward appearance, always proved to be multiplets

whinever they were turbid in reflected iight, the turbidity arising from random rotations

about the fiber axis.

The dufrenite (univ. of chicago No. 1247) used in this study occurs as greenish-black

tabular crystals up to 1 mm across, superficially altered to brown oxy-hydroxides and

associated with pharmacosiderite in a fractured quartz vein from Cornwall, England.

The material apparently corresponds to the wheal Phoenix specimens described by Frondel

(1949). Rotation photographs about 10101 and weissenberg photographs of the 0-, 1-, and

2-levels were obtahed. This same crystal was used in the three-dimensional crystal struc-

ture analysis discussed further on-

The cell data of cornish dufrenite are similar to those reported by Mrose (1955);

indeed, reorienting to the 1-cell shows the relationship (Table 2). Unfortunateiy, the space

group criteria arein conflict, resuiting either from a misinterpretation in the earlier report

i, iro- the unlikeiy existence of a different but related phase. Three-dimensional crystal

structure analysis confirms the space group C2/c lor the dufrenite from cornwall.

The rockbridgeite crystal (univ. of chicago No. 799) was obtained as a splinter from

Irish Creek, Virginia material. It is greenish-black in color; although the crystal was small

(about 80 microns in mean dimension), it afforded superior photographs. This crystal rvas

also used for three-dimensional crystal structure analysis, discussed further on' Inspection

of an assortment of rotation and Weissenberg photographs led to the celi criteria in Table 2'

The rockbridgeite cell dimensions closely match those reported previously by Lindberg

(1949),butwedonotagreeontheinterpretat ionof thespacegrcup'Myf i lmsandcrystal

structure anaiysis support the space grotp Bbmtn'

Laubmannite was first named and described by Frondel (1949). Its type locality is

Buckeye Mountain, near Shady, Polk co., Arkansas. The minerai occurs as radial discoidal

aggregates of fibers, rich yellowish-green in color' During the course of this study' a new

ri"r6ty was discovered, the laubmannite occurring as bright yellow-green aggregates and

afiording a powder pattern virtually identical with the Arkansas material' The location is

Leveiiniemi in the Svappavaara mining district, Norrbotten Province, sweden and the

specimens were collected by Prof . F. E. Wickman in 1957. The specimens are preserved in

tle collection of the Swedish Natural History Museum, and occur in vugs in a gossan zone

surrounding magnetite ore.

Single crystals of Arkansas laubmannite were obtained from specimens I personally

collectecl and from specimens of a cotype collected and kindly donated by Mr. A. Kidwell.

The crystals are small (about 50 microns in greatest dimension) but yielded good rotation

and weissenberg photographs. The resulting data presented in Table 2 conflict with the

conclusion of x{rose 1to.5S; that laubmannite is isostructural with dufrenite' Frondel

(1949) states that Iaubmannite is isostructural with andrewsite on the basis of similar

molecular ratios and powder patterns, but claringbull and Hey (1958) ofier cell criteria for

andrewsite which indicate isotypy with rockbridgeite! The status of andrewsite remains

questionable and further study on type material will be required. I observed the appear-

anc" of a darker green zone, especially near the tips of the fibers on many laubmannite

specimens from Aikansas. The Jarker green mineral is conclusively dufre'ite on the basis

139

d g

a a ;

"< 15

x 9 '

. i o Iu (Q .5

d P o ! o e


* 1 " * + ^ $ N $ N N e

l F i !

9 - G9 : ? 9o | + | I h o IN i I - | t N O I- . i - ; o - ii o

€ o N

@ t s € $ N i H 6 @ bN 6 N $ 3 +

0 D A 4 0 N N T O O

o o ^v v +€ 6 v I ' ' -N O N D t t € O l

E 6 c j d d . j ;

" < GN . v N O d N. ' - o -+ € a o t s h o @ o1 - 1 o 1 1 : " . : A ob o o + o b r F N F

N H N

.9

6j

N A\ \. S B B ^ , i l g s S. : s s s i i n q l > < >

N 6 6 € N J $ N N €U F c a . \ ( J * \ q \ qa

E6

Fr

a

z

H

q

Fq

E.D

ts

A

(J

H

(,N

F{

" 9 c o d'n -=

3gH:ieEE**tEpE E E g g.HR 8EeS'EgS'5 'A.E5

t40 PAUL B. MOORE

of powder and single crystal studies, whereas the more yellow-green material is laubmann-

ite. this suggests that Mrose probably investigated a durfrenite occurring in close

association with the laubmannite.

Data for mineral A, certainly a new species, are also listed in TabIe2. The specimen

studied was collected by Mr. A. Kidwell from a prospect on Fodderstack Mountain,

Montgomery co., Arkansas. considerable effort was expended attempting to obtain-single

crystJ material, but the mineral occurs only in the finest fibers, yielding a 5.12 A fiber

length and nearly unintelligibie weissenberg photogaphs. Thus, the crystal cell data,

crystal chemistry of souzalite is discussed in more detail in the section on the dufrenite

stlucture,

sional single crystal data, the powder Iines were unambiguously indexed. This method of

matching powder intensity with single crystal intensity also demonstrated that preferred

orientation was eliminated during the course of sphere mount preparation. on the other

hand, powder patterns obtained from rolled cylinders or from glass slide smears showed

,".,r".e^pref"rr"d orientation efiects, no doubt contributing to the conflictingdata reported

in the literature. The indexed powder data for these species were then used for a least-

squares refinement of the cell parameters.

Indexed powder data for dufrenite, rockbridgeite, laubmannite and beraunite, and

unindexed data for mineral A are given in Table 3. It is clear that the low-angle reflections

the cell criteria obtained for Arkansas material. Since chemical analysis usually requires a

substantial quartity of material, it is probably based on true laubmannite, whereas the

powder study was performed on adjacent dufrenite.

The powder pattern for beraunite is similar to that previously reported by Frondel

(1949), except foi considerable differences in relative intensities. My beraunite is from the

iur"t-o No. 1 Pegmatite and these data, with the exception of a slightly expanded cell,

agree closely with those obtained from the more typical reddish-orange oxidized varieties


Teslr 3. X-nev Pomrn Plrrnnns or SoME Basrc Inon Pnosrne,rrs

DurnrNrre

r41

IlIo d(obs.) d(calc.) I/Io d(obs.) d.(calc.)

9 12 .002 6 . 7 8 33 6 .4072 5 .9949 5 .0022 + . 7 9 54 4 .3505 4 . 1 1 02 3 . 7 6 33 3 .630I 3 .5087 3 . 3 9 3

5 3 . 2 r +

12.0456 . 7 7 86.4266.0225.0144 . 7 9 5+.3ff i4 .1063 . 7 6 73.6393 . 5 1 13.3893.3893.2r33 . 2 W

3 . 151 3 . 1532.990 2 .9862.860 2 .8s82.795 2 .802

2 . 7 8 72 . 6 2 7 2 . 6 2 72.565 2 .56s2.487 2 .4862.427 2 .426

2.4262.4212.412

2.409 2 .4092 362 2.3672 .277 2 .2782.223 2.2202 .150 2 . r5 r

2m202N2400110T113 1 1Tr2+02511510404313OM8025137rr3r41t45127rr0204043-15221513802

1 0 . 0 . 0314222l i J

806r2.0 2I I . 1 . 1

,r41 2 . 0 . 0

822206622224

I 1 .469I1 1 .45311 1 .M621 | 4235| 1.4049| 1 .39461 1 .38611 1 .37431 1.36281 1 .34551- 1.33211 1 . 3 1 6 52 1 .2952t r.2697| 1.26011 1 .2497

L1

1.8767 1.874+ 8241 .8198 1 .8196 I4 .2 .2r .7645 r . 7618 11 .1 .2r .7s03 1.7483 917

r .74J i0 T3 .1 .51..7232 1 .721s 608

1 .7207 14 .0 .01.7004 r.6948 6261 .6772 1 .6769 7151.6520 r.&76 826

r .&73 I2 .2 .2|.6247 1.6239 718

1 .6224 T5 . r . 21.6222 531

1.6105 1.ff i64 0081.$44 T6.0.4

1 .57+9 1 . s765 m.2 .61.5712 5J3t .57r t 226

1 . 5534 1 .5537 13 .1 .2r .5226 1.5194 5321.4972 | .4970 t t . l .4

1 .4927 T4 .2 .2

I2

10

62

4213

5J

2

2J

I

2 .1502. IOI 2.0992.055 2.0532.008 2.007

2.003r.9897 1.98851.9389 r .93701.9061 1.9080

142 PAUL B. MOORD

T.+n:rn 3-(Continued')

Rocrgnmcorte

I/Io d(obs.) d(calc.) hkl I /Io d(obs.) d(calc') hhl

3 8.41 8.40 0204 6.87 6.89 2001 6.40 6.38 2105 4.842 4.8+2 101+ 4.630 4.653 1112 4.352 +.347 2303 4.180 4.201 040

4.195 r2 l2 3.676 3.663 1315 3.573 3 587 2404 3.433 3.435 3014 3 .3& 3 .365 311

10 3.196 3. 180 32r3 3.015 3.021 2502 2.934 2.935 430

2.928 3314 2 .754 2 .761 1512 2.663 2.6& 4403 2.584 2.586 0025 2.409 2.n2 3511 2.332 2.337 52r2 2.262 2.267 270| 2.214 2.216 620

2.202 M22 2.1+8 2.r5r r7 l3 2.103 2.r0r 080

2.098 2424 2 .052 2.053 4r22 2.Or5 2.016 &03 1.9624 t.9&5 2524 1.8333 7.8317 2622 1.7925 r.7936 4801 t .7r90 17174 f f i2

1.7139 8101.7081 t .7047 2721.680.s 1 .6805 0 .10 .01.6133 1.6141 3031 . 5 8 9 0 1 . 5 8 7 6 1 . 1 0 . 11 .5514 1 .5502 6801.5298 r .5293 6521 . 5 0 9 7 1 . 5 1 0 4 4 . 1 0 . 0

1.5067 343r .4778 r .4786 2921.45781.3927r.36971 .34181.29451.2544r . 2 3 1 2t.2184t . 2 l o r1 .18721 . 1 6 5 0r .14871 .08951.07601.06701.03621.02871 .00630.99s60.98800.98410. 98150.979s

2315z

I

22113I

I1II1z

2

Ll.unueNNrrn

:

I/Io d(obs.) d(calc.) hkl I/Io d(obs.) d'(calc.) hhl

9 1 5 . 3 04 7 . 6 54 6 . 8 62 6 . 3 2

1 0 5 . 1 5 0

15 .387 .696 8 06 . 3 55 . 1 6 65. 161

0200402r0220240001

14

A

I

4 . 7 7 9 4 . 7 8 r4 .638 + .6174 . 1 1 5 4 . 1 3 1

4.1113.990 4 .0053 . 7 1 6 3 . 7 1 8

1 1 112 l260211221270


' r^ rsr,r, 3 - (C o ntinu e d)

t43

t /In d(obs.) d(calc.) I /Io d(obs.) d(calc.)

7 3.M6

6 3 .273

/ J . l . ) l

s 3 . 0 1 2s 2 .8175 2 .7983 2 .5804 2 .428

3 2 .303| 2.26r3 2.143I 2 .1083 2 .070

3.464 4103.454 3013.301 4303.273 3313. 1.51 3413.012 3512 . 8 1 . 5 2 . 1 0 . 02.792 1912.583 4802.423 5212 . 4 2 1 t . l 1 . I2.305 0622.267 6302. t42 5712.114 6tl2.074 4022.069 4r22.068 5812.033 4321.9752 2921.9655 4521.9090 67r1 . 8 5 4 6 6 . 1 0 . 0

1 .8537 2 .16 .01 . 8387 2 .15 .Ot.7+32 800

2.027r.9703

1.90691 .8513

1.8364r.74551.7062t .6&71.6120I .59301 .5751r.54731 .5141r.4734r.4636t .M731.4106r.36191.32731 3007r .29121.26621.2376r.2206t .14251.0859

24Iz

I

223L

l l

4 LL t

1zII

I2I

33+

3+

BrneuNrrn

I /Io d(obs.) d(calc.) I /Io d(obs.) d(calc.)

10

5

1652J

J

l u - 5 /

9 . 5 87 . 2 2 95. 1844.8254.4184.091J - l + t

3.468

2 3.4171 3 .3264 3.1876 3.0822 2.8383 2 .7323 2.7053 2.582

2 2 .4882 2 .4222+ 2.3122 2 . 2 2 7| 2 . r5+3 2.109| 2.082

1 2.0583 2 . W| 1.97t83 r.9232r 1.8697I 1.83021 1 .81281 1.7838r 1 .7671| 1.7423

2.486 2212.419 5152.315 7 t42.227 4082.154 4232.106 3182.080 9122 . O 7 7 1 0 . 0 . 02.060 4242.007 6221.9696 518|.923t 624

10.389 .597.zf f i5 . t92+.8254.4114.07I3 .7503.4873.4613.4183 .3183.1873.0792.8352 .7322 . 7 t l2 . s78

200w2-n2

4001 1 1l t23113t2Ir4600J I J

602314314.s13604It6020

IM PAUL B. MOORE

Ter;n 3-(Conti'nued)

I /Io d(cbs.) d(calc..1 hkl. I /Io d(obs.) d(calc.)

2 t . 71751+ 1.6912L 1.66221 1.64053 r .62W1 1.59851 1.57002 1.5482 1.4594

2 1.43641 1 . 3 7 1 5I 1 .32761 1 . 3 1 5 12 1.29012 1 .2790| 1.2400t+ r .2r14

MrN-enA.r, A

I/Io d(obs.) d(calc.) I /Io d(obs.) d(calc.)

9 9.492 6.894 6.38I 5 . M5 5 . 1 33 4 .793 4 .542 4 .W4 4.O23 3.8263 3.6M8 3.425

10 3 .1894 2.0862 2.9593 2 .8512 2 .7883 2 .5782 2.4872 2.4182 2.400r 2.287| 2.L872 2.0672 2.0501 2.0031 1.9519

2 1.90892 1 .85581 1 . 8 1 7 02 1 .7373I r . 7 t r 61 1 .69181 1.66352 1.64ff i2 1.ffi233 1 .59341 1 .56342 1 .5485I r . 5 2 2 6r t .5r222 r.4883I 1 .4&61 1 .42742 1.39901 1 .38803 1 .32853 r.29202 1 .27041 L .2266L r.2lo92 L .17761 1.0864

BASIC IRON PEOSPHATIJS

T arl-n 3-(Continueil)

T,rpscol,tsnB (L) eno Mrxoner, A (A)

I/Io d(obs.) d(calc.) hht I/Io d(obs.) d(calc.) hht

t45

?L , AL

A 1 4 .560A 2 4 .M2A 1 3 .845A 2 3 . M 1L 10 3 .305

L , A 6 3 . 1 8 8A 1 3.096A 2 2 .812L 4 2 .586A | 2 .507A 2 2 .435L 2 2 .396

L 4 2 . 2 7 8

L 5 2 .039L 2 2 .017A | 1 .9M1A 2 r .8576L 2+ 1 .8328

4 9 .482 6 . 4 12 6 . 1 72+ 5 .176 5 .19s3 4.803 4.809

4.807

r+ r.743e| 1 . .7120 r .7r41 1175 1.6600 1.6709 3323 1.6036 1.6032 3335 1.5932 1.5907 422

1.5903 4042 r .M85 l . 4 r l 5101 1.4306 1.4307 317

1.4303 2r8| 1.3754 1.3799 425| 1 .3657 t.3&5 520

1.3&3 s131 1 .3071 1 .3115 3183 1,.2976 r.2989 44n

1.2988 523r.2983 4261.2975 219

2 1.2247 1.2247 6001+ r.2090 r.2079 5332 1.1604 1.1608 3291+ 1 .1372 1 .1416 2 .2 .102 r.0934 1.0945 3391 1.0650 1.0605 6332 1.0231 1.0247 3.3.102 1.0186 1.0187 6262 1.0095 1.0092 713

3.2833 . 1 8 1

2.598 220

2.+052.4032.2852.2832.0372.O11

1 8351

LL

1 1 0 L1 1 1 L102

LL

L1 1 3202

LL

301 L204 L302 L1 1 5 L313 L224 L

LL

206 r.

IM PAUL B. MOORE

DulruNtrn: Irs Cwsrer' S:rnucrunt

(priv. comm.).

ln"*l-A'r" data appear in Table 5.r

crystallochemical, Formul,a. Based on this crystal structure analysis, the

l To obtain a copy of Table 5, order NAPS Document fi00747 fuom ASIS National

Auxiliary Publications Service, c/o ccM Information Sciences, Tnc., 22 West 34th Stleet'

New York, New York 10001; remitting $1.00 for microfiche or $3.00 for photocopies pay-

able to ASIS-NAPS.

BASIC IRON PHOSPHATES I47

Tenr 4. DulnnNrrr auo Rocxnnmcurn: Fonwure unr:r Mur,rrplrcrrrEs, AtourcCoonorNetns, nNo Isomoprc Tnupnnarurr Facrons.

(Es:rnrernn Sr,c.NDano Ennons rN par_rNrnnses)

DUFRENITE

B(,4P)

Fe (1)

Fe (2)

Fe (3.1Fe (4)P (1)

P (2)Cao (1)o (2)o (3)o (4)o (s):oH-o (6)o (7)o (8):oH-o (e)o (10)o ( 1 1 ) : O H -o (12) :H,O

I12z

z

2

01

0.1s29 (1).1401 (2).218s (3).07e0 (3)

0.0887 (7).076e (8).01e3 (8).1224 (8).1727 (8).2134 (8).202e (8).t28e (e).1766 (8).2223 (8).076e (8).024s (e)

014

- 0 .01s0 (8)lDn (9\

.2612 (13)

.2808 (14)- .1474(28)

.0660 (34)

. s468 (39)

.22ss (s8)

.2896 (37)

.2183 (s7)

.0108 (36)- .s144 (38)- . 2 s 7 4 ( s e )

.2s2e (36)- .2034(36)

.143e (38)- .28s3 (41)

00

0 .1116 (3 ).3s4s (3).3312 (s).3e70 (s)

I1

.3293 (r3)

.3422 (rs)

.4030 (14)

. s0ss (14)

.0170 (14)

.3878 (14)

.3907 (1s)

.2046 (1s)

.2203 (r4)

.1670 (r4)

.064s (14)

.1119 (16)

1 . l s (8 )1 .2s (8)1 .01 (6)1 .18 (6 )r .20 (e)1.38 (10)r.78 (20)1.14 (26)r .77 (sr)1 .8s (30)r.60 (2e)1.58 (29)r.47 (28)2.r3 (s1)1 .8s (33)r .7s (2e)|.62 (28)2.07 (3o)2.2s (34)

0 . 5

c

2n

a

z

z

22

ROCKBRIDGEITE

Fe (1 )

Fe (2)Fe (3)

P (1)

P (2)

o (1)o (2)o (3):oHo (4)o (s):oHo (6):oHo (7)o (8)

00.0687 (s)

.s21+ (4)

.1420 (6)

.4806 (10)o-0477 (r9)

.0829 (10)

.3132 (31)

.3126 (22)

.217s (r8)

.4204 (18)

.4r71 (18).2276 (16)

00.1s74 (4)

.138s (3)

.0432 (s)1414

0.0605 (8)14

.03s7 (18)

.1720 \14)

.1071 (14)

.1763 (1s).1044 (13)

00

0. 238s (14)I2

00. 2s08 (66)

.26rs (s2)

.3876 (9s)00Iz012

0.10 (e)r.64 (e)0. 10 (8)0 . 0 8 ( 1 1 )0.76 (re)2.38 (43)0. s6 (20)0.81 (s7)1 . 8 6 ( s 1 )0.7s (37)0.31 (37)0. e l (37)

(0.10) (33)"

a Refined to B: -0.04.

unit cell of dufrenite contains X2Fe2aP6oe6Has, where X is essentialyCa2+, according to the analysis of Kinch in Kinch and Butler (1336)for cornish dufrenite crystals. The crystallochemical computation based

148 PAUL B. MOORE

Tenr-r 6. DurnrNrrn eNo RocrsRrDcurn: CnrurcAl ANALYSES

DUFRENITE

CaOFeOFezOaPzOrHrORem.

J J . O J

30.26r0.621 .48

3 . 2 r8 .23

45.7332.5r10.32

56. 1633.2810.56

I .2 .

99.49 100.00

Analysis of Kinch in Kinch and Butler (1886). Rem' is SiO? 0'53 and CuO 0'95'

CaFel+Fel03+(OH) rz(HgO)r(PO, eFeos+(OH)s(HzO)r(POr)r

ROCKBRIDGEITE

CaOFeOFezOaPzosHzORem.

t . t 26 . 1 4

50.843t .76

6 . J J

1 . 4 8

99.87

1 1 . 0 749.2032 .806.93

6 r . @32 .85

100.00 100.00

1. Analysisof Campbel l (881).Rem.is0.76MgO,0.40MnO,0'2 lAlzOr,0 ' l l insoluble '

2. Fe2+Fe13+(OII)bGOt,3. Feas+(O)(OH)n(POd'

on this analysis in Table 6 shows the good resolution between the crystal

structure analysis and wet chemical analysis. Based on the evidence

offered here, as well as discussion of valence states later on, the molecular

formula for dufrenite ranges from

x'*F. '*F"' t(oH)6(Hro)2(Poa)+ to r. ' t(on)u(nso)z(Po+)a.

Topology and, Topochem,istry. The octahedral framework in the dufrenite

structuie is shown in Figure 1a. A face-sharing cluster of three octahedra

is the underlying unit not only in the dufrenite structure' but also in

the other basic iron phosphate structures. These clusters are joined along

[roo] uy corner-sha;ing octahedral triplets and along [oot ] ly individual

members of this corner-sharing triplet. The phosphate tetrahedra, all of

which are associated around the face-sharing three-cluster, further knit

BASIC IRON PHOSPHATES 149

HroA

\Y"

H.oFro. 1. Octahedral frameworks of dufrenite and souzalite. (a) The dufrenite octahedral

structure down the b-axis. (b) Proposed model for the octahedral bridge in souzalite.

together this cluster and the corner-sharing triplets by corne-r-sharing.The (PO)a- tetrahedra secure the face-sharing triplets along [0101, thefiber axis. The 5.13 A axis, common to many basic ferrous-ferric phos-phates, is approximately the sum of the phosphate tetrahedral edge andthe octahedral edge belonging to the face-sharing cluster. Taken togeiher,these clusters, the tetrahedra, and portions of the corner-sharing ttipletsmake up a reasonably dense slab which is oriented parallel to 1100|;this explains the perfect { 100} cleavage observed for dufrenite.

This remarkable face-sharing three-cluster involves eight oxygen atomsfrom the (POtt- tetradentate ligands and four hydroxyl groups. Ac-cording to Moore (1965a), its formula can be written Fer(OH)a(Oo)a.

150 PAUL B. MOORE

Further citing that paper, I shall refer to this cluster as the Z-cluster,following the proposed code for octahedral clusters involving four orIess nuclei. In the same fashion, the corner-sharing triplet, having com-position Fes(OH)o(H:O)z(Oo)s, wil l be called the e-cluster. Thus, theoctahedral framework in dufrenite involves the condensation and subse-quent corner-sharing of Z-clusters and e-clusters.

Remaining in the dufrenite structure is a fairly open cavity along thetwo-fold rotor at 0 y I/a. This cavity is partly occupied in the Cornishdufrenite studied herein. It is believed to be occupied by calcium, sincethe analysis of Kinch indicates the presence of 1.5 percent CaO. A crudeestimate of the average electron density can be obtained by the refine-ment of the isotropic temperature factor. Applying the scattering curvefor half-occupied Ca2+ at this site, the isotropic temperature factor re-fined to 1.78 A', a not unreasonable value for a large cation in a cavity.The interatomic X-O distances are 2 X-O(I) 2.4t, 2 X-O(2) 2.49,2X-O(12) 2.32, and. a long 2 X-O(3) 2.76 h. The first six distancesare reasonable values for Ca-O; discounting the long pair of distances,Ca2+ is in approximate octahedral coordination. The formula for Corn-ish dufrenite is then approximately CaFerz(OH)tr (HrO)n(POa)r, yielding

a computed CaO of 3.1 percent by weight.Though present evidence is inconclusive, I would like to suggest that

dufrenite follows an oxidation trend analogous to the well-known tri-phylite-ferrisicklerite-heterosite series. Triphylite is commonly observedto follow a step-wise progressive oxidation, with retention of the essen-tial crystal structure, from LiFe2+(POa) to Fe3+(PO+), the ferric end-member being heterosite. The general formula expressing this series canbe written Li*Fe2+*!'s3*t-*(POa), where progressive oxidation also in-volves concomitant alkali-leaching. In such a process, the electrostaticvalence balance about the oxide anions is preserved. Quite analogous isdufrenite, since the electrostatic valence balance computation in Table7 shows that only one water molecule is present in the asymmetric unit,this being O(12), with 2, the electrostatic valence sum,:0.50 for theferric end-member. The hydroxyl oxygens, OH(5), OH(8) and OH(11)allhave ):1.00, a value too high for neutral water molecules. Thus, toachieve stability during oxidation of the dufrenite crystal, a commen-surate amount of large cation is leached out of the structure. It now re-mains to obtain the limiting composition for the ferrous end-member inthe dufrenite series, as discussed in the following section.

Interatomic Distances. The average Fe-O interatomic distances areFe( l ) -O 2.00, Fe(2)-O 2.06, Fe(3)-O 1.98, and Fe(a)-O 2.04 Lwith estimated standard errors of +0.02A. Those for Fe(1) and Fe(3)

BASIC I ITON PIIOSP HATES

Tenr,r 7. Emcrnosrerrc verrNcn Belerqcns (z) or,DurnrNrrE AND Rocxnnmcnrrn

Rockbridgeite

>s >b| 7 0 1 . 7 51 7 0 1 . 7 51 . 7 5 1 . 7 5L 7 5 1 . 7 51 . 3 3 1 5 02 . 1 4 2 2 52 . 1 4 2 . 2 50 9 4 1 . 0 01 . 7 5 1 . 7 51 . 7 5 I 7 51 0 0 1 . 0 00 5 0 0 . 5 0

I J I

o (1)o (2)o (3)o (4)o (s):oH-o (6)o (7)o (8):oH-o (e)o (10)O ( 1 1 ) = o H -o (12) :HrO

P(2) *Fe(a)P(2) +Fe(a)P(2) fFe(1)P(2) *Fe(s)Fe(2) *Fe(3) *Fe(a)P(1) +Fe(2) +Fe(4)P(1) +Fe(2) +Fe(4)Fe(3) *Fe(a)P(1) *Fe(3)P(1) *Fe(3)Fe(1) *Fe(3)Fe(1)

>a >b2 . 1 4 2 . 2 52 . 1 4 2 . 2 51 . 0 0 1 . 0 01 7 5 1 . 7 51 . 0 0 1 . 0 01 3 3 1 . 5 01 . 7 5 1 . ? 51 7 5 1 . 7 5

o (1 )o (2)o (3):oH-o (4)o (s ) :OH-o (6):oH-o (7 )o (8)

pr?) rFd / r \ rF - r rV

P(1) *Fe(1) *Fe(z )Fe(3) *Fe(3)'P(1) *Fe(3)Fe(1) +Fe(3)Fe(1) +Fe(2) +Fe(3)P(2) *Fe(3)P(1) +Fe(3)

3 For Z-cluster iron atoms averaging f2.67 (based on ! psr+fpsr+).b For ferric end members.

are normal distances for Fe3+-o; similar values (1.9g and 1.9gA) werefound for the two independent FeB+-(o, oH) octahedrar averages inlaueite (Moore, 1965b). The Fe(2) and Fe(4) octahedral distances asso-ciated with the /-cluster are indicative of the presence of some iron in theferrous state. The electrostatic valence balance calculation in Table Zshows that the three oversaturated oxide and hydroxyl anions-oH(S):OH; 0(6) , O(7) :Op-are associated wi th both Fe(2) and Fe(a) .By lowering the average positive charge for these cations, the degree ofover-saturation can be diminished. This, in combination with chemicalanalyses, suggests that the l imit for the ferrous end-member is Fe2+.1-e2+

grade wil l be discussed later under the general crystal chemistry of thesecompounds.

152 PAUL B. MOORE

to conform with its associated polyhedron. In general, a Schlegel diagram

can be found for any polyhedron of genus zero (Coxetet, 1947), that is,

the family of closed convex polyhedra with no holes. These diagrams

offer the .advantage over tables in that M-O, O-O' distances and

angles can be conveniently inserted, easing the difficulty of relating tabu-

lar data to figures. The O-O' distances are particularly interesting

and offer us further insight into the Z-cluster, as discussed later'

Souzalite: a probable d.uJrenite-tike structure. Souzalite is a hydrother-

dufrenite. The souzalite space group is eithet A2/m or A2 whereas

dufrenite is C2/c: b, c and p are similar for the two minerals, but o is

halved in souzalite relative to dufrenite. Further evidence for a relation-

ship is the observed similarity in intensities of the Z0 levels for the two

minerals.I propose a plausible atomic arrangement for souzalite' This proposed

structure has the same Z-cluster orientation as in dufrenite, except that

the cluster is centered at the cell origin. The Z-clusters are linked in the

same fashion along [0Ot] resulting in denSe slabs parallet to {tOO}, iaen-

tical to those found in dlfrenite and explaining the perfect { tOO} cleav-

age also observed for souzalite. The difference appears in the Iinking

unit along [1OO]. Instead of the e-cluster, this unit consists of two ter-

minal octahedra corner-linked to a central tetrahedron, as depicted in

octahedral sites and T for tetrahedral sites (excepting phosphorus), the

structure type formula for souzalite is MvroTrv(OH)6(POr)+'2HzO, with

M and T including Al, Mg, and Fe' More detailed crystal structure

analysis will be required before a sensible ordering scheme can be pro-

posed. The tentative positions of the water molecules in Figure lb were

Iocated on the basis of geometrical limitations combined with hydrogen

bond distances.


RocKeRrncrrrE : hs Cnvsra,r, Srnucrurc

Ilxperimental, proced.we. A nearly equant greenish-black crystal of 80 microns mean dimen-sion was selected from Irish Creek, Virginia material. Its small size necessitated long expo-sure (96 hours) Weissenberg photographs using Zr-filtered Mo radiation. Packets of threefilms were separated by aluminum foils and the intensity data were estimated by eye usinga spot scale of fifteen intervals based on the strong (642) reflection. 491 independent reflec-tions were obtained for lhe hkl to 2ft5 levels, of which 340 were non-zero. These data werenot corrected for differential absorption, since the crystal was of favorable size and shape.

Solution oJ the structure. Knowledge of the dufrenite structure led to rapid solution of therockbridgeite atomic arrangement. The &-cluster and its associated (POr)3- tetrahedrawere assumed present in the rockbridgeite structure and oriented by placing the cluster'ssymmetry center at the cell origin. The space group criteria were then applied and thesolution was immediately obvious. The derived atomic cbordinates for the metals werethen referred to P(ryz); all prominent vectors could be labelled and gave sumcientlyaccurate trial coordinates for refinement. The oxygen atoms were approximately locatedfrom a polyhedral model and their coordinates were adjusted on the basis of an ensuingthree-dimensional Fourier synthesis.

Two space groups, Bbmm and Bbnt2, conform to the systematic extinction criteriaobserved on f-lms. Only Bbm2 can satisfactorily accommodate the cell contents, assuminga crystal with ordered and fully occupied sites. Only two atoms-Fe(3) and O(3)-militateagainst the centrosymmetric space group, assuming complete site occupancy, Refinementbased on Bbm2 involved 15 independent atomic species, and converged to R7,a1:6.14 1o.observed reflections, but resulted in disappointingly high e.s.d's for Fe-O distances(+0.06 A). Computation of Fe-O and P-O interatomic distances showed that seriousdepartures from expected values occurred, e.g. with P-O ranging from 1.43 to 1.70 A.

It was then decided to refine the structure based on Bbmm, asstming that Fe(3) at *,

!,-l/4 oniy haH-occupied its site. FuII occupancy of this site would result in 28 ironatoms in the unit cell, instead of 20 as supported by chemical analysis (Table 6). Besides,fully occupied face-sharing columns of octahedra would run parallel to the c-axis. Severalother arguments, including the Patterson synthesis, supported half-occupied Fe(3) sitesand a centrosymmetric arrangement. Topologically and topochemically, there is no dif-ference between occupancy at r y l/4 and r y 3/4 since the neighboring oxygen atoms andthe other cations reside on the mirror planes at z:0 and l/2. Thus, there should be noenergetic differences between the two sites. Incomplete occupancy of equivalent Fe sites iswell-documented by Katz and Lipscomb (1951) in the crystal structure of lipscombite,Fe2+Fe3+z(OH):(POa)s, where full occupancy would result in infinite face-sharing octahedralcolumns. In lipscombite, these sites are 75 percent occupied on the average, and the frac-tional occupancy is probably even lower for the more oxidized synthetic varieties.

In the space gro,lp Bbmw, only 13 independent atomic species are present: two sets ofoxygen pairs become equivalent by reflection. These pairs were averaged, starting with thecoordinates obtained from the noncentros),'rnmetric refinement. O(3) was assumed half-occupied at x, l/4, z (with z-0.38); full occupancy requires that O(3) be situated at z:I/2,bnt this results in an unreasonably long Fe(3)-O(3) distance of 2.35 A. Indeed, non-centric refinement at z: l/2led to B: 5.6 Ar, the model with the half-occupied site at thegeneral z-position further supported by a convergence to B:0.8 ff. Thus, O(3) isdisordered in the same manner as Fe(3).

Full-matrix three-dimensional atomic coordinate and isotropic temperature factor re-finement converged to Rml:0.12 for the observed reflections and 0.16 for all reflections.The final parameters are listed in Table 4 and the lf'.0"1 -f.,t" data appear in Table 5 (see

l J 5

154 PAUL B. MOORE

footnote under dufrenite). The temperature factors are generally lower than those obtained

for dufrenite and for the beraunite studied by Fanfani and Zanazzi (1967), and this effect

may arise from the more close-packed nature of the rockbridgeite structure. These tempera-

ture factors for the rockbridgeite structure show anomalous difierences (compare P(1)

with P(2); Fe(1), Fe(3) with Fe(2); and O(8), which is negative), and several attempts

were ma.de to obtain more meaningful values by alternating coordinate iefinements with

scale factor refinements, etc. Similar anomalies were observed for the refinements in the

noncentrosymmetric space group and it is believed that a strong correlation exists between

the temperature factors and atomic coordinates for some atomic positions. For this reason,

these temperature factors probably have little physical meaning.

Crystallochemi,cal formulu The crystal structure analysis confirms the

generally accepted rockbridgeite f ormula of Lindberg (1949). The crys-

tallochemical computation in Table 6 based on the analysis of Campbell(1SS1) shows good agreement, except for a higher water content in the

analyzed material. Analyzed rockbridgeites consistently show a slightly

higher water content, probably reflecting the fibrous habit of the ma-

terial, which may permit the presence of some occluded water. Based on

available chemical analysis, it is proposed that a continuous series exists

between unoxidized rockbridgeite and fully oxidized material:

F.'+F"'+n1oH)b(po4)3 ro r.t*u10;iou)r(Pon)r.

The electrostatic valence balance calculations in Table 7 show that

O(6) is severely oversaturated with respect to an hydroxyl group but

undersaturated with respect to 02- for the oxidized end-member. OH(6)

is probably an average of 02- and OH- anions, the ferric end-member

more precisely written Fe3+s(Orrz, OHvr)r(OH)a(PO+)a'

Topot'ogy and topochemistry.The octahedral framework in the rockbridge-

ite structure is shown in Figure 2a. The centrosymmetric Z-cluster is

situated at the cell origin and joins to adjacent Z-clusters by shared ter-

minal edges, forming octahedral chains parallel to the 6-axis. The fusion

of the Z-cluster by shared edges results in a more efficiently packed struc-

ture than those found in dufrenite and beraunite. The phosphate tetra-

hedra are situated around the Z-cluster in a manner identical to that in

dufrenite and beraunite. The cross-linking corner-sharing octahedra are

doublets, analogous to the b-cluster of Moore (1965). These b-clusters

have composition Fe2(OH)u(Oo)r. Thus, the dense slabs found in the

dufrenite and beraunite structures are preserved oriented parallel to

{OtO} i" the rockbridgeite structure. Ilowever, the addition of edge-

sharing elements along the b-axis results in the preferred cleavage direc-

tion parallel to { 100 l, not purallel to the slabs but perpendicular to

them.Rockbridgeite can be derived from the dufrenite and beraunite struc-


Frc. 2. Octahedral frameworks of rockbridgeite and beraunite. (a) The rockbridgeiteoctahedral structure down the c-axis. (b) Reraunite structure. Coordinates are from Fan-tani and Zanazzi (19 67 ).

156 PAUL B, MOORE

tures by twinning through reflection parallel to the slabs. In this manner,

one octahedron in the dufrenite e-cluster is removed, resulting in a D-

cluster and in beraunite one octahedron is joined by corner-sharing to

the o-cluster making it a 6-cluster. As a consequence, the denser rock-

bridgeite appears to be a highly strained structure, possibly explaining

the displacement of OH(3) away from the mirror plane at z:l/2.

Interatomi,c ilistances. The average Fe-O interatomic distances are

Fe(1)-O2.07. E(2)-O2.tt and Fe(3)-O2.00 A with estimated

standard errors in individual distances +0.03 A. Thus, Fe(3) is probably

essentially ferric iron. Fe(1) and Fe(2) belong to the Z-cluster and these

distances are suggestive of mixed Fe2+ and Fesi over these sites. As with

dufrenite, the oversaturated oxide and hydroxyl anions-O(1)' O(2)

and O(6)-are associated with both Fe(1) and Fe(2) in the Z-cluster'

The P-O distances average P(1)-O1.52 and P(2)-O1'56 A' The

distances associated with O(a) are rather unusual: P(1)-O(4) 1'47

is unusually short, Fe(3)-O(4) 2.L2long, and 0(4)-0(6) (with Fe(3))

3.22 h long; but the remaining polyhedral distances for O(4) are quite

normal. since o(4) is associated with the b-cluster bridged by phosphate

tetrahedra, these distances may indicate a "forced" arrangement in

the crystal structure. All M-O, T-O and O-O' polyhedral distances

are presented in Figures 4 and 5b as the Schlegel diagrams for the poly-

hedra.

Comporison of the d.ufreni,te, rockbridgeite and berouni.te polyhedral dis-

tances. The composite Schlegel diagrams (Figs. 4, 5a and 5b) for the

polyhedra in the three related structures afford clear and concise means

.of illustrating distances for comparative purposes. The octahedra in the

Z-clusters for the three species are drawn so that a one-to-one correspond-

ence exists in position. O-O'distances for shared edges are consis-

tently shorter than for the free edges. The central octahedron ol the h-

cluster for all three structures shows remarkable similarity in correspond-

ing distances among these structures. This is because the central octa-

hedron has the same nearest neighbors in the structures. The outer

octahedra in the dufrenite and beraunite Z-clusters are similar for the

same reason, but in rockbridgeite differ because of the extra shared edge.

In rockbridgeite, compensation of O-O' distances is especially pro-

nounced (compare the long 0(1)r-0(6)rrr and O(1)r-O(2)rv with

analogous edges in the other two species). The distances are substan-

tially longer in the rockbridgeite Z-cluster than in the other two sepcies,

indicating more ferrous iron in that mineral. The remaining polyhedra

in the three species cannot be compared because their nearest neighbor

polyhedra are arranged diff erently.


Tse Z-Clusmn

The Z-cluster is known to exist in the crystal structures of dufrenite,rockbridgeite, beraunite, lipscombite, barbosalite, laubmannite, andpossibly mineral A. Beraunite, with the most open octahedral frameworkof these species, is closely related to dufrenite. The beraunite structuralparameters possess fairly low estimated standard errors (Me-O+0.01A) as a result of careful study by Fanfani and Zanazzi (1967). The octa-hedral topology is shown in Figure 2b, and the cell criteria are offeredin Table 2. Here, the Z-cluster is corner-linked by octahedral a- and b-clusters. In a manner similar to dufrenite, the octahedra and tetrahedraform reasonably dense slabs oriented parallel to {tOO}, explaining theperfect {tOO} cteavage plane observed for the mineral. The octahedralframework results in large open channels oriented parallel to the fiberaxis and non-ligand water molecules occupy these channels; the crystallo-chemical formula for beraunite involves the series Fe2+Fe3+s(OH)u(HrO)n(POn)n.2H2O to Fe3+6(OH)6(HzO)B(PO4)4.2HzO and the electrostaticvalence balance computations parallel those observed for dufrenite.

Lipscombite, Fe2+Fea+r(9H)2(POa)2, has an intriguing structure whichwas revealed by Katz and Lipscomb (1951), the polyhedral diagram ofwhich appears in Figure 3a and the cell criteria in Table 2. Here, equiva-Ient chains of face-sharing octahedra run parallel to the a-axes of thetetragonal cell. The octahedra are only partly occupied, with an averageoI 3/4 Fe per octahedron for the ratio Fe2+: Fe3+: 1:2. These chains areactually face-fused Z-clusters with a phosphate environment identicalto that of the other basic iron phosphates. Lipscombite probably repre-sents the densest arrangement of Z-clusters among the basic iron phos-phates, and can be considered a limiting structure for these compounds.

Gheith (1953) investigated the stability of the lipscombite structurethrough synthesis, using various starting materials. He showed that thelipscombite structure could be synthesized as the pure ferric end-member,with 5 t/3 psa+ per cell instead of. 2 Fe2+*4 Fe3+ for the l:2 compound,but the analogous ferrous end-member (requiring completely occupiedoctahedral sites) could not be prepared.

It is tempting to add that an ordered derivative of the lipscombitestructure could possibly exist, where one set of face-sharing chains iswholly intact and fully occupied by cations. To satisfy the lipscombite1:2 stoichiometry, the other chain would reduce to the insular o-clusterswhich join to the phosphate tetrahedra and bridge the chains by corner-sharing. Because of the short Fe-Fe distances (about 2.7 A) along thechain, such an arrangement may have unusual electrical properties.Presumably, the o-cluster would be occupied by a cationic species ofdifferent crystal radius, thus increasing the possibility of such an order-

157

PAUL B. MOORE

: \ h

: Xo i 3bD

",H U )

? E6 o ,

Y ap o

. O +z 2

3 EI . .

o d

; d

# ah @ ^

c E ; 6

Y X

159BASIC IRON PHOSPHAT:ES

UJ

Z =

?5E.uJco

UJFzLdE.

lo

6= Yosy

o

N

UJF

UJoE.co:<

E.

o

ooN

DUFRENITE

o(3F{ t .96

o i l t ) r , MrF e ( l ) ' - - " '

0.5r) P(r) 0.46) P(a

Fro. 5. Schlegel diagrams of the remaining nonequivalent octahedra and tetrahedra in(a) dufrenite and (b) rockbridgeite.

000) ot4)(.53) il.sr)

p t 2 \


ing scheme. The resulting structure would be orthorhombic and, ifcentrosymmetric, have the space group Pbmn Otherwise, the cell cri-teria would be similar to those for lipscombite.

Closely related is the barbosalite dimorph, isotypic to lazulite. Barbo-salite, Iike the manganoan-lipscombite of Lindberg (1962), occurs as aIate-stage hydrothermally reworked product of triphylite in pegmatites.Barbosalite is an ordered equivalent of the Iipscombite structure, havingisolated Z-clusters instead of disordered face-sharing chains. It is notimmediately obvious that the cell criteria for barbosalite in Table 2 arerelated to those of tetragonal lipscombite. If the triclinic pseudo-mono-clinic cell with or:a*-u2c^;b1:a^-b^; ct:a^16* is chosen, its projec-tion down cr shows the relationship (Fig. 3b). The same dense slabs arefound in this structure as in the others, but these slabs are linked laterallyby the (POn)3- tetrahedra instead of bridging corner-sharing octahedralclusters. The coordinates of isotypic lazulite, determined by Lindbergand Christ (1959).were used for constructing this diagram.

The lipscombite-barbosalite dimorphism is further complicated by theallied mineral, manganoan-lipscombite. According to Lindberg (1962),

the space group is not body-centered but primitive pseudobody-centered,P422 and the o-axis is related to that of lipscombite by the factor t/2(Table 2). Comparison of these data suggests that a family of orderedsequences may exist, ranging from the completely disordered lipscom-

bite to the completely ordered barbosalite. The stability relationshipsof these sequences are unknown, and Gheith's (1953) study on thelipscombite-barbosalite pair is inconclusive for lack of detailed singlecrystal studies. Further work is necessary, before these compounds-par-ticularly manganoan-lipscombite-can be precisely defined.

A yet larger polyatomic complex can be chosen common to the basicferrous-ferric phosphate structures discussed above, if the nearest neigh-bor polyhedra around the Z-cluster are considered. This includes the Z-

cluster and the four corner-linked nearest neighbor octahedra and their

associated tetrahedra. The resulting packet of polyhedra, illustrated inFigure 6, is a convenient key to all the structures. As Table 2 shows,

these compounds have one axis oI ca. !3 A length which, l ike the 5.1 Afiber axis, appears to be invariant for these compounds.l The thirdcrystal axis is variable and extends along the direction of the four short

arrows in Figure 6. This figure also shows that the 13 and 5.1 A axes l ie

in the plane of the dense slabs rnentioned earlier and that the illustratedpacket includes all the necessary ingredients for constructing these slabs.Alternatively stated, the variable axis is that axis along which the slabs

are bridged together to form a three-dimensional arrangement.

1 Choosing one-haif the obtuse diagonal of a and c in beraunite yields 13 O A fnis

orientation conforms to the ca. 13 A repeat in the other minerals.

161

162 PAUL B. MOORE

Frc. 6. The /z-cluster and its immediate polyhedral neighborhood in the basic ironphosphates. The (PO,)a- tetrahedra, shown as P-O spokes, share corners with the octa-hedra and are above and below the cluster. The axis normal to the plan of the diagram is the5.t A fiber axis, and the 13.8 A axis is the remaining "invariant" in these structures, thesetwo directions in the plane of the polyhedral slabs. The remaining crystal axis, in the direc-tion of the arrows, is variable.

Where do laubmannite and mineral A stand in relation to these ar-rangements? The ceII criteria for these compounds leave little doubtthat similar, if not identical principles also apply. Indeed, preliminaryPatterson synthesis, P(xy), for laubmannite conclusively confirms thepresence of the Z-cluster. Ilowever, the large number of atoms in theasymmetric unit, many of which lie outside the /z-cluster and its immedi-ate environment, poses a difficult problem for the structure analyst.Mineral A so far does not afford suitable single crystals for a parallelstudy, but it is apparently a variant of the laubmannite structure. De-spite these impediments, I have derived plausible structures for these


minerals, but for lack of convincing experimen.tal evidence, will not offerthem here.

Pleocltroism, Color and the h-cluster. One noteworthy feature of the basiciron phosphates in general is the dependence of color on valence statesof the iron. Table 1 lists the iron phosphate hydrates and their typicalcolors. The ferrous compounds are pale-colored, in the greens and pinksand only weakly pleochroic. The ferric compounds are variously yellow,brown, red and pink and also only weakly pleochroic. However, thecompounds involving both ferrous and ferric iron are extremely pleo-chroic and deep greenish-black in the gross sample. The colors in thesecompounds must result from efiects distinct from the additive propertyof component colors of the pure ferrous and ferric salts. The oxidizedferric end-members of the ferrous-ferric phosphates are either orange,y€llow or brown and only weakly pleochroic. Gheith documented theeffect of valence on the color of lipscombite. Ferrous-ferric lipscombite isdeep greenish-black and opaque except in the thinnest slices, whereas theferric end-member is olive-yellow in color.

It is well-known that mixed valence states of a particular atomic spe-cies in the same crystal often contribute to dark colors. The probleminvolving mixed iron valences in minerals often falls into the realm ofintervalence transfer absorption of the type involving symmetrical elec-tron transfer between homonuclear cations (for an excellent review, seeAIIen and Hush, 1967). For such iron compounds, this can be writtenpsz+f Fe3+-psa+f psr+. Absorption studies conducted by Hush (1967)

on vivianite have led to the assignment of the band maximum of 15.1X 103cm-1 to Fe2+-Fe3+ intervalence transfer. This contribution is polarizedparallel to the D-axis of the crystal which is the direction parallel to theaxis including the two iron atoms separated by 2.95 A in the octahedraldoublet Fer(HzO)n(Oo)6. This doublet corresponds to the edge-sharingc-cluster of Moore (1965). The mixed valences arise from partial oxi-dation of iron in the c-cluster, aiz. Fd+Fel+(OH)(HrO)B(On)6, resultingin the deep-blue color of the mineral corresponding to the vibration direc-tion parallel to the 6-axis. According to Hush, the ratio o1 psa+/Fe2+ isabout 0.05 for this crystal. In the basic iron phosphates, where such ratiosare often much higher, the crystals can be very strongly absorbing tothe extent of being nearly opaque except in the thinnest pieces. Since theprobability of electron transfer and hence of absorption decreases withincreasinq interatomic distances, the short interatomic distance of co.2.7 A be;een the metals in the Z-cluster also renders electron transferhighly likely. The problem is not as simple as in the vivianite structurewhere the octahedral doublets are discrete and not interconnected bybridging Fe-centered oxygen octahedra. In the basic iron phosphates,

164 PAUL B. MOORE

the Z-cluster is further linked laterally by octahedra with Fe-Fe distancesof ca. 3.5 A. In rockbridgeite and lipscombite, the Z-clusters are furtherfused by edge- and face-sharing respectively, forming chains with 2.7 Aand 3.2 A Fe-Fe distances in the former, and partly occupied chains withFe-Fe distances of 2.7 A in the latter.

Small single crystals of greenish-black dufrenite, beraunite and rock-bridgeite from the localities mentioned earlier were mounted on aspindle stage and oriented by X-rays so that spindle axis correspondedto the 5.1 A fiber translation. These crystals were then examined on apolarizing petrographic microscope. Vibrations parallel to the fiber axisare least absorbed by the crystals and the pleochroism is pale orange-yellow with a slight greenish tinge. Vibrations perpendicular to the fiberaxis are very strongly absorbed and the colors range from dark greenish-orange or red to deep blue or green. Figure 7 depicts the relationships ofthe pleochroic colors, the axial positions and the orientation of. the h-cluster, projected down the fiber axis for dufrenite, beraunite and rock-bridgeite.

It was first believed that the vibration direction with deep blue togreen color would lie parallel to the axis of the Z-cluster, analogous to thevibration direction of the blue pleochroic color in vivianite, correspond-ing to the axis of the double group in that mineral. The results show thatthe problem is far more complicated: in dufrenite and beraunite, thevibration direction with deep green color is oriented about 40 to 45oaway from the axis of the Z-cluster. In rockbridgeite, where the mirrorplanes yield two nonparallel Z-cluster axes, the vibration directionscoplanar to these axes range from bluish-green to dark bluish-green incolor. These results indicate that the general problem of electron trans-fer absorption in crystals involving the Z-cluster is very interesting, butwill require much more elaborate analyses than the qualitative resultsoffered here.

PenecBNBsrs

Three important modes of occurrence are documented for the basiciron phosphates and include "limonite" and bog ore beds; hydrother-mally reworked products in pegmatitesl and replacements and cements inclays, sands and bone material.

The occurrence in "l imonite" and bog ore beds is quite characteristic.Here, the minerals occur as fibrous encrustations and mammillary ag-gregates upon goethite, hematite, etc. Rockbridgeite from its tvpe local-ity, for example, occurs as large blocks of fi.brous masses associated with

"limonite" beds. The mineral is also quite abundant in novaculite whereferric oxy-hydroxides have accumulated. Quite noteworthy are the manyoccurrences in prospect pits in the Ouachita Mountains, Arkansas. The


BERAUNITE DUFRENITE

y'h-c lusfer oxis

- -Eoo)- - t

oronge-rcd

(roo)--\ Perfecl cleovoge

- ! lLool (oot)

greenish - oronge

ROCKBRIDGEITE

, r h-c lusler oxis

-- ' Zlo(OlO) b lu ish green

h-clusfer oxis

i loo)dork blu ish-greenperfect c leovoge

Frc. 7. Gnomograms showing the cleavage planes, crystal axes, Z-cluster axes and absorp-tion maxima for beraunite, dufrenite and rockbridgeite. The fiber axes are polar.

phosphates cap and replace botryoidal aggregates of goethite, and thereis little doubt that the phosphates post-date the ferric oxy-hydroxidesand probably represent products of ferric oxy-hydroxide reworking byphosphatic waters. The characteristic gum-like and botryoidal featuresof the surfaces and the appearance of shrinkage cracks indicate a gelorigin or source for these minerals. Indeed, glassy patches of resinousdeep-red material still remain, often showing partial to complete re-placement by the iron phosphates. The powder patterns of some gelsresemble goethite, and with broadened lines.

One locality, "Avant's Claim" near Shady, Polk Co., Arkansas, is asmall nearly hidden pit showing bedded masses of porous goethite geland dense botryoidal seams of the phosphates in brecciated novaculite.

165

166 PAUL B. MOORE

The botryoids, when broken, show the characteristic radial aggregates of

fibrils. The earliest phosphate to appear is greenish-black rockbridgeite,

but it frequently grades outwardly into beraunite.Another locality on Buckeye Mountain (near Shady) is the type

Iocality for laubmannite. The mineral occurs as bright olive-green

botryoidal and mammillary aggregates fiIling crevices in novaculite. In

some instances, it replaces goethite. These laubmannites, when crystal-

lized in open cavities, frequently terminate as brilliant deep emeraldgreen crystals, which proved to be dufrenite on the basis of single crystal

studies.Both these Iocalit ies clearly show less dense, more open hydrated

structures replacing or following the earlier phases. This type of para-

genesis indicates falling temperature to successive regions where the

appearance of structures with increasing amounts of water molecules as

ligands and as cavity fillers are rendered more stable' The general se-

quence is thus,

'dense' hydroxylated structures

lower T

+HrOt open hydrated-hydroxylated structures'

A remarkable occurrence at Fodderstack Mountain' Montgomery Co.,

Arkansas, shows fibrous masses of rockbridgeite with mammillary sur-

faces and color banding. Rockbridgeites frequently show color bands at

millimeter intervals, involving the colors black, yellow, orange and

blue. In open cavities, the rockbridgeite is frequently postdated by a

vivid yellowish-green material. Through careful dissection of fibers

showing color banding, sphere mount powder patterns were prepared.

The color bands proved to result from the appearance of different phases,

with black : rockbridgeite, yellow-orange : yellowish-green : mineral A,

and blue:Iipscombite. The powder data for l ipscombite in these speci-

mens appear in Table 3. X-ray study of a large rockbridgeite from Rock-

bridge Co., Virginia, capped by the yellow pulverulent product described

by Frondel (1949) shows that the latter material is a mixture of mineral

Af goethite. Finally, I would like to point out the similarity of Frondel's

powder pattern of an unknown phosphate from Waldgirmes, Saxony

with mineral A (Table 3). Mineral A appears to be a f airly typical prod-

uct of rockbridgeite leaching. The rhythmic color banding in the Fod-

derstack materials indicates a mode of origin somewhat different than

the earlier mentioned parageneses. Perhaps these phases crystallized

from a gel with rhythmic alternations of FeO:FezOa:PzOolH2O ratios or

through rhythmic fluctuations in pH.Rockbridgeite is a rather abundant hydrothermally reworked product


of triphylite in granite pegmatites and specimen study shows the mineralin a range of associations and sequences, usually appearing earlier thanthe other minerals. Only a few specimens could be located which showedrockbridgeite partly replaced by another mineral. Several specimens fromPalermo 11 in the Bjareby collection show rockbridgeite fibers replacednear the surface by orange beraunite. Another specimen shows rock-bridgeite fibers tipped with olive-green dufrenite. These observationsparallel the settings observed in the Arkansas materials. The counterexample-showing the more open structures replaced by denser ones-has not been observed.

It is proposed that the typical sequence of replacement and/or para-genetic succession in the basic iron phosphates is

mineral A. ' , v

rocKDrlogerte I| ____)

Iaubmannite j\ ,'

dufrenite

--+ beraunite.

FunrnBn RBnanrs

Despite the increase in the number of investigations on the basicphosphates of iron, our knowledge of the group is still incomplete. Of thelist of 40 species in Table 1, the atomic arrangements are known for only15 of them. If the aluminum- and calcium-rich iron phosphates areadded to the list, the number of species increases considerably, for whichonly a few atomic arrangements are known.

Some of the more incompletely understood iron phosphates may proveto be related to the family involving the Z-cluster described in thispaper. Cacoxenite, for example, is an intriguing substance. A superiorsingle crystal of this mineral from Shady, Arkansas, was secured fromthe Bjareby collection and yielded the data in Table 2. Fot crystal-chemical calculations, I used the analysis of Church (1895), which isprobably more reliable than the other analyses reported for the mineral.Assuming the water removed by desiccation over sulfuric acid is zeolitic,the formula is close to Feg(OH)rr(HzO)g(POr)+.15H2O. Halving the a-and c-axes yield 13.80 and 5.22 A respectively, similar to the "invariants"for the structures discussed herein. The cavities in this structure areprobably quite large and the arrangement may be similar to that of thezemannite structure, discussed by Matzat (1966). Zemannite is a hexa-gonal tellurite mineral with large 8 A in diameter channels filled by watermolecules and alkalis. Indeed, guided by the cell criteria, Z-clusters canbe linked to form an open honeycomb structure, and this model ispresently being tested.

168 PAUL B, MOORE

Another group includes the gel-like basic calcian ferric phosphates

egudite, richellite, azovskite (in part), mitridatite (in part), borickyite

and foucherite. Many of these are chemically closely related if not iden-

tical to each other. Their compositions are reminiscent of the dufrenite

formula. Do these gels contain disordered Z-chrsters? Interesting in this

aspect is the experiment of McConnell (1963)' where thermal recrystal-

lization of richellite was shown to produce a calcium variant of the

lipscombite structure. If this recrystallization is assumed to follow a

mechanism akin to the recrystallization of metamict materials, then the

Z-clusters were probably present in the gel state, suggesting that these

amorphous minerals formed in an environment similar to the crystalline

basic ferrous-ferric phosphate minerals. These gel materials may be

remnants of the early-formed liquors from which the crystalline ferrous-

ferric phosphates grow.AcrNowlmcuButs

Specimens of basic iron phosphates, many of which were sacrificed for study, were

donated by Messrs. J. K. Nelson and A. Kidwell, and Prof. J. V. Smith. Dr' J' J. Fahey

arranged for me to obtain crystals of type souzalite, housed in the u.S. National Museum

and donated by Dr. G. Switzer. Mr. S. J. Louisnathan assisted in the computation of the

powder data and Mr. c. Stern in the intensity collection for dufrenite. Prof. D. R. Peacor

kindly read the manuscript and offered helpful suggestions as to its improvement.

This work was made possible through the NSF grant GA-907 and an ARPA grant ad-

ministered to the Department of the Geophysical Sciences.

RnrnnrNcrs

Ar-r-nx, G. C., nto N. S. Husrr (1967) Intervalence-transfer absorption. t. Prog. Inorg.

Chem. 8,357-386.

Cauernr,r,, J. L. (1881) On dufrenite from Rockbridge County, Ya. Amer- f ' Sci.2Zr 65-67 .

Cnuncu, A. H. (1895) A chemical study of some native arsenates and phosphates. 6.

Cacoxenite. M iner al. M ag. I lr 8-10.

Cr.mrNoaur-r., G. F., exo M. H. Hrv (1958) A re-examination of andrewsite. Mi.neral'. Soc.

Notdce lOO, l.

Coxerrt, H. S. M. (1947) Regulor Polylopes. Pitman Publishing Corp., New York, p. 10'

FrxraNr, L. aro P. F . ZllNlrzzr (1967) The crystal structure of beraunite. Acta Crystrallogr.

22, t73-181.Fnonorr, C. (1949) The dufrenite problem. Amer. Mineral.34,513-540.

Gnnrrrr, M. A. (1953) Lipscombite, a new synthetic "iron lazulite." Amer. MinaoJ.38,

612- 627.Husn, N. S. (1967) Intervalence-transfer absorption.2. Prog. Inorg. Chem'8,401-40'5'

Kntz,L. el{o W. N. Lrpscons (1951) The crystal structure of iron lazulite, a synthetic

mineral related to la iite. Acta Cr.yslallo gr, 4, 345-348.

KrNcn, E. aNo F. H. Burr-un (1886) On a new variety of mineral from cornwall. Mineral,.

Mag.7,65-70.LrNonono, M. L. (1949) Frondelite and the frondelite-rockbridgeite series. Ama. Mi.neral.

34, 541-549.- (1962) Manganoan lipscombite from the Sapucaia Pegmatite Mine, Minas Gerais,

Braail. Amer. Mineral.4T 353-359.

BASIC IRON PHOSPEATES

-, AND C. L. Csnrsr (1959) Crystal structures of the isostructural minerals lanilite,scorzalite and barbosalite . Acta Crystallogr. 12, 695-696.

Mntzr't, E. (1966) Die Kristailstruktur eines unbekannter zeolithartigen Teilurit-min-er als. T s c h er wak s M in er al. P etr o gr op h. M itt. lZ, 108-1 17 .

McCorwrr,r,, D. (1963) Thermocrystallization of richellite to produce alazulite structure(calcium lipscombite). Atner. M,ineroL 48, 300-307.

Moonn, P. B. (1965a) A structural classification of Fe-Mn orthophosphate hydrates. Amer.M iner al. 50, 2052-2062.

-- (1965b) The crystal structure of laueite. Amer. MineraL S0, 1884-1892.- (1969) The basic ferric phosphates: a crystallochemical principle. Science, 164,

1063-1064.Mnosn, M. E. (1955) Problems of the iron-manganese phosphates (abstr.). Gwtr. Soc.

Ama.80il1.66, 166O.Pr.r.Acsn, C., H. BrnmaN lxo C. Fnoxorr. (1951) The System of Minua!,ogy . . . oJ Dana.

Vol,. 2. JohnWiley and Sons, New York, p. 960.Pncon.L, W. T. aro J. J. Fanev (1949) The Corrego Frio Pegmatite, Minas Gerais: scorza-

lite and souzalite, two new phosphate minerals. Amer. Mi.neratr. g4r 8-93.

Monuscri.ptreceioed, Moy 29, 1969; accepted.Jor publicati,on September 20, 1969

169

crystal chemistry of the basic iron phosphates paul b ... · the basic phosphates of the transition...

Documents