American Mineralooistvol. 57, pp. 13s3-1i44 (1972)

NATROPHILITE, NaMn(pOa), HAS ORDERED CATIONS

Peur, Bnreu Moonn, Department ol the Geophgsical Sciences,U niu er sity o t C hi ca g o, C hica g o, I ll;inois 6 06 37

AgsrnAcr

Natrophilite, 4NaMn(POn), a 10.522(5), b 4.gtj7(2), c 6.812(ilA, space sroupPn'am, has ordered cations, isostructurar with triphylite, with Na --M(r) andMn - M(2) ieferring to the octahedral site populations of the olivine structuretype. R(hhl) - 0'055 based on atomic coordinate and isotropic thermal vibrationparameter refinement utilizing 78g independent reflections- Average polyhedralMe-O distances are Na-O 2.824, Mn-O 2211, and.p_O 1.545 A.

n11ber of sha.red polyhedral edges, renders prediction of cation ordering basedsolely on electrostatic arguments uncertain.

INrnolucrtoN

Recent interest in the crystal chemistry of transition metal phos-phates of the alkalies necessitates detailed study of their atomic ar-rangements and cation distributions. Determination of the alluauditestructure by Moore (1gzl) assisted in interpreting its paragenesisand range of cation substitutions. Natrophilite, NaMn (pOa) , is re-lated in composition but possesses a simple structure which is evidentlyrelated to olivine.

Natrophilite was first described by Brush and.Dana (1g90) asa rare species associated with coarse primary lithiophilite crystals andtheir products of hydrothermal reaction from the famous Branchvillepegmatite, Fairfield co., connecticut. The species replaces lithiophiliteand, like lithiophilite, is followed by species of hydrothermal rework-ing such as hureaulite, reddingite, fairfieldite, dickinsonite, etc. Theauthors suggested that natrophilite formed from lithiophilite in amanner akin to the replacement of adjacent spodumene by ((cymato_lite" (- albite * muscovite). rn other words, natrophilite is a metaso-matic exchange product of lithiophilite, involving the addition ofsodium and the removal of lithium. A paraller observation was noted

1333

1334 PAUL BRIAN MOORE

by Moore (1971) where alluaudite, co. Na2Mn'*Fe'*Fe'*(PO+)g usuallyreplaces triphylite, Li (Fe,Mn) (PO+) .

The atomic arrangement of natrophilite was investigated by By-striim (1943) who concluded on the basis of trial structure factor

calculations for disordered and ordered octahedral cations that the

structure involves disordered (Nao.rMno.s) populations over the M(L)

and, M (2) sites of the olivine structure type' Triphylite, on the otherhand, has been conclusively shown to possess ordered Li : M(1) and(Fe,Mn) = M (2) cations as a result of accurate three-dimensional ,atomic parameter refinement, by Finger and Rapp (1970), despitethe nearly identical crystal radii for Li- and Fe'* in octahedral co-

ordination by oxygen. Furthermore, the ordered nature of I'i and Mnatoms in lithiophilite was established by Geller and Durand (1960)'

That Na. and Mn'*, with crystal radii which differ by ca. 2O percent,

would occur disordered is surprising.I submit evidence that the Na* and Mn2* cations are ordered in

natrophilite, over the M(1) and M(2) sites respectively, and that

the ordering scheme is consistent with simple electrostatic argumentsinvolving cation-cation repulsion effects although o, priori prediction

of the correct ordering scheme is not obvious.

Expnnrrvrowrel

A single crystal was obtained from the sample investigated by Fisher (1965)

which possesses the number ,,Yale, Brush No. 2363". A cleavage fragment, nearly a

cube in shape, 90 microns on an edge, afforded superior single crystal data obtained

from a PaILRED automated diffractometer with molybdenum radiation and

seconds on each side of the intlividual peaks and 1.6o half-scans at a, rate of 2.5o/min

were applied. Symmetry equivalent reflections were averaged, leading to 789 non-

zero independ.ot lr(obr)|, 279 ,,zero,, reflections (including systematic absences)

and 106 rejected reflections (background differences in excess of 30 percent of the

and Bystriim (1943).

Rnnrvnnnur eNn Octerruoner, OnoonrNc Scnpuo

Determination of the cationic distributions for'natrophilite proceeded from

the scattering tables of MacGillavry and Rieck (1962). Full-matrix least-squares

NATROPHILITE

TABLE I. MTROPITILITE. PCIVDER DATA.

(Fe,/Mn radiqtion, II4.6 m camera diameter, calibrated fi lm)

l3it5

!/!9 d(e!e)

3 5 . 2 r + I

6 4 . 4 9 8

6 4 . 0 4 5

3 3 . 9 0 4

5 J . O 5 b

I 3 . 6 1 0

s 3 . I s 4

8 2 . 8 6 3

2 2 . 7 0 2

10 2 .6011

1 0 2 , 5 8 3

3 2 . q 8 7

4 2 .tt2t1

u 2 . 3 7 7

4 2 . 3 2 2

LI 2 .26L

2 2 .L20

2 r . 9 5 6

3 1 . 9 0 q

2 1 . 8 5 1

7 1 . 8 3 I

2 1 . 8 0 6

4 1 . 7 3 8

s 1 . 6 9 4

i 1 . 6 8 9

3 1 .642

t l_ .606

2 I . 5 8 6

3 1 . 5 7 8

3 1 . 5 3 3

r l . so t

3 I . 4 6 0

L/!9 d(oLt

2 l. t+38

2 I . q16

2 1 . 4 0 1

2 1 . 3 9 8

2 r . 3 8 r +

2 L . 3 7 2

1 r - 3 5 6

3 1 . 3 r + 8

2 r . 3 3 4

2 1 . 3 0 6

l _ r . 2 q 6

2 r . 2 2 3

2 r . 2 0 6

2 r . 1 9 0

2 L l 8 8

1 L . 1 6 9

3 r . t s 6

3 1 . 1 5 2

1 r . 1 3 8

2 L-L33

1 r . 1 2 0

I t . r r 6

I 1 . 1 0 9

2 1 . 0 8 9

4 r . 0 7 2

2 1 . 0 6 5

2 1 . 0 5 8

2 1 , 0 5 3

2 r .o37

r { 1 .026

dG-efd b\l

1 . 4 3 9 7 L O

t,f+16 232

r .403 7 t t

r . 3 9 9 6 2 I

L 383 31r+

r .372 423

L , 3 7 2 4 3 r

r .356 332

L.347 603

1.333 02 t+

1 . 3 0 4 0 3 3

r .247 040

r . 2 2 3 3 3 3

1.2u6 532

t . I92 722

1.189 4211

l . t88 7 13

1.185 623

I . 1 6 9 q 3 3

1 . 1 5 5 3 1 5

1.153 142

1 . 1 3 8 9 t 0

I . I 3 2 2 \ 2

t . l l8 234

1. r l5 730

1,109 r.r4l

1 . 0 8 8 3 3 4

1 . 0 7 1 9 l ?

I . 0 6 3 7 1 4

1.058 5r+1

1 . 0 5 8 5 1 s

1.bs2 ro 'o 'o

r.035 '{25

1 . 0 2 5 6 0 5

dGarg)

5 . 2 6 2

q . 5 0 7

4 .042

3 . 9 1 3

3 . 6 6 8

3 . 6 1 9

3 . I s 6

2 . 8 6 9

2 . 7 0 6

2 . 6 L 2

2 1 5 8 5

2. r+9q

2 -t+28

2 . 4 2 6

2 . 3 7 4

2.327

2 . I 2 2

1 . 9 5 7

1 . 9 0 6

1 . 8 5 4

r . 8 3 4

1 . 8 1 0

].7r+0

1 . 6 9 7

1 . 6 8 9

1 . 6 5 2

1 . 6 4 3

1 . 6 0 8

1 . 5 8 9

l . s 7 8

1 , 5 3 3

1 . 5 0 2

t . 116I

hkt

2 0 0

IIO

201

0r1

1t1

2r0

002

3r0

202

311

lr2

0 2 0

40t

r20

2L2

r{10

12r

22r

022

1r3

s l I

2 2 2

q 2 0

tl21

3r3

60r

512

4 0 3

03I

l_3r

00tl

602

3 3 0

33r

l3i]6 PAUL BRIAN MOORE

TABLE 2. NATROPHILITE. CRYSTAL CELL DATA,

I

a ro . s23 (s ) I

b r r . 987 (2 )

c 6 .312 (3)

space group 9l9T

f omula l ta (uno . n,Feo . o7)

(Po4)

r 0 .53

s . 0 0

6 ' O

Pnam

3

r0.s2 (r)

q .97 (1 )

6 .32 (r )

Pnam

Z

spec i f i c g rav i ty

densitv Gglg:)

r+

3 , q 0 , 3 . q 2

, 31 . C / g r y C f r

I T h i s s t u d y . T h e l l n : F e r a t i o a n d s p e c i f i c g r a v i t i e s i v e r e o b t a i n e d f r o m

B r u s h a n d D o n a ( f 8 9 0 )

2 f i s h e r ( 1 9 6 5 ) .

3 B y s t r b n , ( l ! t l 3 ) , t ' i . t h b - a n c l c - a x e s i n t e r c h a n g e a l .

refinement of atomic multiplicities, coordinates, and isotropic temperature factors

utilized a losal modification of the familiar ORFLS program for IBM 7@4 com-puter of Busing, Martin, and Levy (1962). The initial scattering curves applied

included (Mn* * Na.)/2 - M(D - M(2); P** = P and O- - O(1) - O(2) -

O(3). The following cycles in proper order were: scale factor; atomic co-

ordinq,tes; scale factor, atomic coordinates; M(1), M(2) multiplicities; scale

facLor, M(l), M(2) multiplicities and atomic coordina,tes; and scale factor,

M(l), M(2) multiplieities, atomic coordinates, and isotropic thermal vibra-

t ion parameters. In the f irst sequence, R(hhI) : > l l f ' (obs)l - lF'(calc)l l /> lF(obs)l

- 0.058, resulting in the M(1), M(2) multiplicities, atomic parameters

and isotropic temperature factor convergences in Table 3. From the disparatemultiplicities, it was obvious that a parallel refinement was necessary with M(1): Na* and M(2) - Mn* substituted for the (Na* a Mn*)/2 table. Convergencewas assured at R(hkt),- 0.055 and the results in Table.3 clearly show that

natrophilite is essentially an ordered crystal with M(L) = Na and M(2) -

(Nn,Fe).Several features are noteworthy in this study. First, convergence by atom

multiplicity refinement quickly leads to an approximate estimate of site popula-

tions, provided that the individuat curves and the combination of scattering

curves are reasonably congruent over sin a/tr. Second, both convergences in atom

coordinates and isotropic thermal vibration parameters are close to the limits

of error, adding confidence to the physical meaning of the therrral vibrationparameters, provided that a rather complete lf'(obs)l data set is available. The

isotropic temperature factors Na,* - 1.18, Mn4:0.55, Ptu - 047' and O(1) :

NATROPHILITE

M(I ) = Nao.sMno.s

M(2) = Nao.5Mn0.5

h1r1 = l l"r.0

h { r ) = " r o . n , t " o . o ,

q . 0 0 ( 2 ) 0 . 2 8 4 6 e ( 8 ) - 0 . 0 0 6 s ( 2 )

4 . 0 0 . 1 0 6 8 ( 1 ) 0 . ' 1 4 0 0 ( 3 )

4 . 0 0 . I I 3 I ( r l ) 0 . 7 r + 7 1 ( 9 )

r + . 0 0 . 4 6 8 7 ( 4 ) 0 . I 6 0 s ( 9 )

r337

TABLE 3. MTROPHILITE. CELL SITE MULTIPLICITIES, ATOMIC COORDIMTES AND

ISoTRoPIC TEMPEMTURE IACTORS. (Errors in parentheses refer to the last diei t . )

f i x y

2 . 3 0 ( q ) 0 0

s . 8 0 ( 3 ) 0 . 2 8 4 7 s ( e ) - 0 . 0 0 6 s ( 2 )

3 .93 (ri) 0 0

z 4R21

0 1 .u (3 )

vtt 0.r+6 (2)

0 r , r8 (q)

v4 o.ss (1)

vt+ 0 .47 (2)

v\ o.7q (s)

v4 o .7q (6 )

0 (r)

o(2)

0 (3 ) 8 . 0 0 . u s s ( 3 ) 0 . 3 [ 4 ( 6 ) 0 . 0 s 6 8 ( s ) 0 . 7 q ( r + )

_ Correct s i te prefeTence scheme of octahedral cat ions,

O(2) = O(3) = 0.74 A' are typical values for these ions in a dense-packed ar-rangement. since thermal vibration parameters are sensitive to the scatteringpowers applied when atomic site multiplicities are fixed, they add confidence tothe conclusion that, natrophilite has ordered cations.

we inquire why Bystriim (1g43) was unable to derive the correct orderingscheme for natrophilite. several explanations are possible. First, Bystriim notes"From the dimensions and space group of natrophilite, it could be assumed thatthe structure should be of the olivine type. It seemed thus as if the distributionwas the same as lithiophilite with the Li-atoms substituted by Na-atoms. Thiscould not, however, explain the difference in powder photographs and a pre-liminary calculation of the intensities showed the impossibility of this scheme,'.unfortunately, Bystriim was apparently not aware of the very delicate balance

His data set was too incomplete for accurate refinement. This leaves us withthe sobering thought thai, all older site preference schemes invorving handcalculations with limited data must be treated with some zuspicion.

The final lF(obs) l-F(calc) data appear in Table 41.

'To obtain a copy of Table 4, order NApS Document 0lg5g from ASISNational Auxiliary Publications service, c/o ccM rnformation corporation, g66Third Avenue, New York, N.Y. 10022; remitting $2.00 for microfiche or $b.00 forphotocopies payable to CCMIC-NAPS.

PAUL BRIAN MOORE1338

s9

UF

=

-J

F

- i

= =3 oE

z

E <

5 <

6 .

NATNOPHILITE 1&39

Inrnnemrrrc Drsrercns

Figure 1 presents Schlegel diagrams of the coordination polyhedra innatrophilite. Such diagrams have the advantage of direct comparisonof polyhedral interatomic distances between different crystals; thediagrams have been discussed by Moore (1970). Included are theinteratomic distances for lithiophilite, refined by Geller and Durand(1969), and for triphylite by Finger and Rapp (1970). In all threestructures, the P-O distances are essentially the same and withinthe limits of experimental error. For natrophilite, the averages areP-O 1.545, Na-2.374, and Mn-O 2.2t1 L.

A hierarchy of polyhedral distortions is evident. For all three struc-tures, the corresponding O-O'tetrahedral distances differ only within3o of each other and the tetrahedron exerts the greatest influence onthe other polyhedra. The M (2) polyhedral distances for lithiophiliteare the dilated equivalents of triphSrlite on account of the larger crystalradius of Mn2*. The M (2) polyhedron in natrophilite departs from thegeometrical similarity evident in the other two structures on account ofthe large Na* cation in M(l). Although O(t)-O(B), the shared edgebetween Jlf(l) and M(2), is shorter than O(t)-O(B,s) in the otherstructures, the reverse is true in natrophilite. In all three structures,the edge opposing the P : O : M(2) shared edge evinces extremecompensation and is the largest edge for the M (2\ polyhedra. This ispariicularly noticeable in natrophilite where the P = O : M(2) sharedO(3,m)-O(3) 2.435 A is compensated by O(3,m,s)-O(B,s) 8.824 A.

The M(1) polyhedron, on account of the cation of low charge, ismost dependent on the geometry of the M (2) and P polyhedra. Par-ticularly noticeable are the O(2,i)-O(B) edge distances on the equatorwhich includes the pair of opposing P : O : M (l) shared edges. Itsdistances are 3.13, 3.49, and 4.07 A for lithiophilite, triphylite, andnatrophilite respectively. The dilation from lithiophilite to triphyliteindicates that the Coulombic repulsion between M(l) and F is verysensitive to M(1)-P distances sinee this distance is shorter in triphy-lite. In natrophilite, the geometrical and Coulombie repulsion effectsare additive, since the O(2)-O(3) shared edges remain essentiallyinvariant for all three structures and the large Naoo octahedron resultsin extreme dilation of O(2,i)-O(3). This confirms the third rule ofBaur (1970), that the a,uerege value of the distances within a coordina-tion polyhedron is approximately constant. Thus, the compensatoryeffects between shared and opposing unshared polyhedral edges striveto approximate this invarianee for a siven coordination nolvhedronand its ionic species.

1340 PAUL BRIAN MOONE

Ononnrnc on M(1) aNo M(2) In N.trnopnrr,rro

For the classical ionic model involving non-equivalent sites of thesame oxygen coordination, several qualitative predictions can be madeas regards possible ordering of cations. The model introdueed here isoverly simplistic but, within the limitations of compounds belongingto the same structure type, should reveal at least trends when differentcharged species are considered. For cations of different charge butsimilar crystal radius in an identical anionic matrix, the high-chargedcation will favor the site with the minimum overall nearest neighborcation-cation repulsion. This can be stated in the following manner:define the coordination number of the nearest neighbor cations (K'),

the charge of the nearest neighbor cations (Z), the charge of thecentral cation (2"), its coordination number K, and the interatomicdistance between the central cation and its e-th nearest neighbor cation,(do). As a normalizing factor, the bond strength of the i-th nearestneighbor cation, (2"'/K')0, is introduced. The Coulombic repulsion is

Q : Z"/K E (Z"',/K',)o dn-',

To assess this term for the non-equivalent sites of a given structuretype, consider Zo' and. Z" of lhe same radius. The values of Q arethen computed. Aecordingly, the site with highest 0 will accommodatethe cation of lowest charge in the actual crystal, thus minimizingcation-cation repulsion eff ect.

In the olivine structure type, the M(1) octahedron shares two edgeswith ? (the tetrahedron), two edges with other M(1) and two edgeswith M (2) ; but M (2) shares only one edge with 7 and two edges withM(l). Thus, M(1), with the higher Q-value, should accommodate thecation of lower eharge, prouidinE the ffystal radii of the catinns imquestion, are sint ilar.

Now consider the case for two cations of like charge but differentcrystal radius. Where will the larger cation go? Intuitively, it shouldgo into the site with highest Q-value since the d;2 terms will be less,minimizing cation-cation repulsion effects. However, empirical evi-dence greatly favors the opposite effect: the smaller cation tends topreferentially occupy the site of maximum Q-value. My unpublishedresults on absolute volumes f.or 44 independent [AIOo1 octahedra inwell-refined oxide and silicate structures leveal that octahedral volumedeereases with increase in the value of Q. The dependence of volume(I/) upon Q is quite substantial: for example, in chrysoberyl, Al2BeOa,

Q = 3.2 and Iz : 8.85 A for M(1) and 1.5 and 9.48 A3 for M(2), avolume differenee of 7 percent. An extreme case of the preferential

NATROPHILITE 1341

occupancy of a small highly charged cation in positions with maximumshared edges occurs for CdSbzOe, which is based on hexagonal close-packed oxygen atoms. In this structure, the Sbs. cations occur in thecenters of octahedra which define a sheet of the Al(OH)a type, witheach octahedron sharing three edges. The Cdr* cation resides in anoctahedron between sheets above the sheet voids and shares onlycorners with other polyhedra (Magn6li, 1941). In sapphirine, whereeight non-equivalent octahedra occur in a cubic close-packed oxygenarray (Moore, 1969), the Al-rich populations occur in the sites withmost shared edges, whereas the Mg-rich populations appear at thepositions with least shared edges. In the olivine structures, Mg2. residesin M (I) and Ca2* in M (2) for monticellite (Onken, 1g6b).

Thus, there are two competing effects which determine octahedralsite preferences in the olivines: the electrostatic effect where cationsof low charge tend toward the site of maximum Q-value, and the sizeor geometrical effect where the smaller cation tends toward the siteof maximum Q-value. Table 5 lists olivine structures where structuralrefinements have afforded ordering schemes. For the ordered olivinestructures with cations of same charge (monticellite and glaucochroiie) ,fhe M (l) position accommodates the smaller cation. For the cationsof different charge, the smaller charged cation occurs aL M (l) (lithio-philite, triphylite, and natrophilite) with the exception of sinhalite.For sinhalite, the geometrical effect evidently is more important butthere is accumulating evidence in this laboratory that Mgr. and AIB*are partly disordered in this crystal. In the forsterite-fayalite series

TABLE 5 . COMPARISON OF IONIC MDI I OF GT]ONS- IN M( I ) AND M(2) S ITES

toR o l l v lM s rRut ru i l s " ' .

Compound

monticellite

Fomula

F a

C a M s , S i o , , l M c " '- a

sinhalite uearpo{ rr3*

h2*glaucochroite cdnlsio4

IithiophiLite lu"[eo{ Li*

tliphylite rt"fro{ r.i*

forster i te-fayal i te 1ug,re)r[s;of n"""rv stat ist ical

natrophi Li te r"v" l eoJ N,+L ' J

M(I) M(2) ( [ r x I00) t - f f i

c.2+ + 28' /6 0.00

M g 2 * + 2 6 - . 5 0

c . 2 * + 1 8 o . o o

I , h 2 * + t o + , 5 0

r . 2 + + t t + . 5 0

d i s t r . + 6 0 . 0 0

M r 2 + - r g + . 5 0

I ^ - , ion ic rad ius (M(21 - Mt r l l : The e f fec t i ve ion ic rad i i (based on r V Io2= = f . ' JO 8)

ion ic rad ius M(2) a re f rcm shannon and prs i t t (1969) .

2 z@(r ) ) n 'a z@(2) ) a re the iodc charees on M(1) and M(2) respec t ive lv '

1342 PAAL BRIAN MOORE

where both charge and ionic radius are similar, (Mg,Fe) disorder isextensive.

Natrophilite is of particular interest since it is the only compoundof an ordered olivine structure type where the M(1) cation is muchlarger lhan M (2), with radius ratio ra(l):ra(2\ = I.24.It points outa "patadox": possible solid solution may occur only when the twoeffects "cross", that is, when the larger cation of low charge is po-

tentially more stable in a solid solution with a smaller cation ratherthan another cation of identical radius. Solution would appear whenthe ratio in their ionic size is erce,eded by some amount considerablygreater than 1.0. In other words, M(2) is predicted to potentially ac-commodate the larger cation of lower charge as well, as observed forsinhalite. Below this value, the electrostatic effect predominates, andat 1.0 and below the geometrical effect as well, the last two casesrepresenting ordered olivine' structures. This tempts the question:given the composition, say, NaMg(PO+) (providing it crystallizes withthe olivine structure type), rNs* lrMg2* - I.41, is the crystal orderedwith Na : M (l) or with Na = M(2), or is it disordered?

Tno NernoPHrLrrE PemcnNnsrs

Natrophilite has been reported only from the Branchville pegmatite

where it occurred replacing lithiophilite nodules associated with alteredspodumene in a cleavelandite replacement unit. The altered spodumeneconsists of t'B-spodumene", resulting from extensive replacement ofNa for Li leading to a mixture of eucryptite and albite, and "cymato-lite" which is a mixture of albite and muscovite. According to Brushand Dana (1890), natrophilite occurs in a paragenesis which parallelsthat of the metasomatically altered spodumene' suggesting that natro-philiie is a Na-metasomatic exchange product of lithiophilite.

The Custer Mountain Lode, also called the Skookum Feldspar Mine,I l/2 miles east-southeast (SW l/4, SE l/4, sec.30, T.3s) of Custer,South Dakota, reveals a replacement unit which in many respectsis similar to the Branchville pegmatite. According to Fisher (1945) thelithiophilite, which is compositionally near the Branchville material,crystallized in the early hypothermal stage followed by spodumenewith which it is closely associated. cleavelandite in turn postdates bothminerals. I have personally exarnined the main pit and the lithiophiliteand spodumene occurring in place. As noted by Fisher, the spodumeneassociated with the lithiophilite is fresh and reveals no "cymatolite"alteration. Examination of several hundred lithiophilite specimenscollected by myself and Fisher failed to reveal any natrophilite. In-stead, the alkali-leached oxidized products sicklerite and purpurite

NATROPHILITE I.3/'}

were found bordering fractures and rimming the lithiophilite. The deepgreen mineral noted by Fisher (1945) which occasionally replacesthe lithiophilite in a paragenesis similar to natrophilite at Branchvilleis definitely a member of the alluaudite series on the basis of myX-ray studies.

Alluaudite is a very abundant and typical metasomatic reactionproduct of triphylite-lithiophilite at many pegmatites in the BlackHills. Too numerous to mention, the primary and secondary phos-phate-bearing pegmatites are being presently stddied in detail. Ac-cording to Moore (1971), the alluaudite series at the iron-rich sidecan be written NaCaMnr-Fe2r-(POa)3-NaNaMn2*Fer.FeB'(pO+)s-Mn2'Fe23*(POa)3, and all examples studied so far involve some ironin the ferric state.'Why

doesn't natrophilite occur at the Custer Mountain Lode? Itappears that Na-metasomatic exchange of the lithiophilite and spodu-mene was not nearly as extensive as at Branchville. But more signifi-cant, all examples of Black Hills Na-metasomatism of triphylite-liihiophilite involve partial oxidation of Fe2* which favors the stabilityof the alluaudite structure. since no alluaudite has evidently been re-corded from the Branchville pegmatite, it can be inferred that Na-metasomatic exchange occurred under eonditions which did not leadto oxidation of the subordinate iron in solid solution. Postdating thisexchange reaction at lower temperature were the oxidative alkati-leaching reactions leading to the formation of sicklerite and purpurite.

Since Mn'* is more resistant than Fe2* to oxidation in crystals andin solution, potential natrophilite occurrences should be found onlywhere the lithiophilite member prevails. such localities are few and theproducts are at least partially oxidized. At Varutrii,sk, Sweden, analluaudite replaced the lithiophilite (Quensel, 1g56) and at pala,california, where lithiophilite nodules have suffered much hydrother-mal attack, the secondary phosphates contain ferric iron.

Since primary pegmatite phosphates are highly susceptible to chemi-cal exchange and reaetion, they are indicative of the delicate reactionswhich postdate primary giant crystal formation and can reveal muchinformation easily overlooked in studies on the far more resistantsilicates. These observations on phosphates have not been appreciated:survey reports on the mineralogy of pegmatites are notoriously inac-curate in interpretation of the phosphate phases present and in caseswhere gross Na-metasomatic exchange results in alluaudite replacingtriphylite-lithiophilite, a jig in the pegmatiLe p.uzzle has been over.looked. This part of the przzle is necessary to our understanding ofreplacement units in pegmatites and when repeated from pegmatite

LU4 PAUL BRIAN MOORE

to pegmatite, as in the Black Hills, suggests that such phenomena arenot unusual or very local but more general than previously suspected'

Acrwowl,nncunNrs

This study was zupported by a Dreyfus Foundation Award and an Advanced

Research Projects Agency grant awarded to The University of Chicago' I thank

Dr. Gordon E. Brown for rewarding discussions and comments which con-

tributed to the improvement of the paper.

Rnrnnexcos

Beun, W. H. (1970) Bond length variation and distorted coordination polyhedra

in inorganic crystals. Trans. Amer. Crgstallogr. lssoc' 6, t29-155'

Bnusn, G. J., aNo E. S. Da}re (1890) on the mineral locality at Branchville,

Connecticut : Fifth paper. Amer. J. Sci. 39, 201-216.BusrNc, w. R.. K. o. Menrrn, eNn H' A' Lnw (1962) ORFLS' a Fortran crysfal-

Iographic least-squares program. u.s. oak Rid,ge Nat. Lab. (us. cleari,ng-

house Fed. Sci. Tech. Inlo.) Rep. ORNL-TM-305.Brsrndrt, A. (1S43) The structure of natrophilite, NaMnPOr. Arlt' Kemi' M'ineral'

GeoI .77B,14.FrNcnn, L. W., eNo G. R. R-tpp, Jn. (1970) Refinement of the crystal structure of

triphylite. Ann. C arne g ie I rts t. W ash. Y ear B o ok, 68, m0-m1

Frsson, D. J. (1945) Preliminary report on the mineralogy of some pegmatites

near Custer. S. D ahota GeoI. Suru. Rep. Inuest. 50' 37-56'- (1965) Natrophilite. Amer. Mineral. 50, 1096-1097

Gur,r,nn, S., eNn J. L. DunaNr (1960) Refinement of the strrtcture of LiMnPO*'

Acta C ry stallo gr. 73, 325-331.MacGrr,r,evny. C. H., a}[r G. D. Rrncr< (1962) Internati'onaL Tables Jor X-rag

Crgstallographa, VoI. 3, the K1'noch Press, Birmingham, England'

MecNr6r,r, A. (1941) The crystal stmcture of Iead metantimona,te and isomorphous

compounds. Arh. Kemi. M'i,neral. GeoI. l5B,l-6.Moonn, P. B. (1969) The crystal structure of sapphirine. Amer. M'i'neral- 54,3149'- (1971) Crystal chemistry of the alluaudite stmcture type: contribution to

the paragenesis of pegmatite phosphate giant crystals. Amer. M'ineral' 56,

1955-1975.oNrnN, H. (1965) Verfeinerung der Kristallstruktur von Monticellit. Tschermalc's

Mineral Petrogr. Mitt. 70,3+-44.

QuoNsnr,, P. (1956) The paragenesis of the varutrhsk pegmatite including a

review of its mineral assemblage. Arlt. Mi,neral. Geol. 2,9-83.

Manuscri'pt receiued,, October 19, 1971; accepted, Jor publication, Mag 26, 1972'