American Mineralogist, Volume 73, pages 798-808, 1988 Effects of compositional variation on the crystal structuresof pyroxmangite and rhodonite LrNoa R. hncxxny Corning Glass Works, SullivanPark FR-51, Coming, New York 14831, U.S.A. Cn.q.RLrs W. BunNrr.qlr Department of Earth and Planetary Sciences, Harvard University, Cambridge,Massachusetts 02138, U.S.A. Ansrnlcr As adjacent members of the pyroxene-pyroxenoid polysomatic series,constructed of fragments of the wollastonite (W) and pyroxene (P) structures,pyroxmangite (WPP) and rhodonite (WP) possess very similar structures with comparablesite distortions and cation ordering patterns. The two structuresrespond quite similarly to cation substitutions. Oc- tahedral and tetrahedral distortions generally lessenand the silicate chains straighten as larger cations substitute for smaller. Both structuresexhibit limited stepwiseordering of cations over the octahedral sites,with large cations preferentially entering the sites on the edges of the octahedral bands.Detailed structural responses to cation substitution generally parallel those observedin pyroxenes. The characteristicsof the inner octahedral sites strongly influence several structural parameters, including the sizesand configurationsofthe outer polyhedra. There is, how- ever, no well-defined mean cation size limit that differentiates rhodonite from pyroxman- gite from pyroxene.Structural parameters for both rhodonite and pyroxmangite structures change smoothly with composition and produce only minor structural adjustments.These adjustments,however, produce localized higher-energy structural configurationsthat clus- ter at the boundariesbetweenthe W and P modules of the structures. Such configurations include a strongly kinked tetrahedral chain, short Si-Si distances, and a highly distorted octahedron. In contrast, the P-P boundary in pyroxmangite is virtually distortion-free. Concentrationsof strain energyat W-P boundarieslikely play a major role in controlling phasetransformations in this system. INrnooucrroN structed by appropriately stackingthesewollastonite (W) Pyroxenesand anhydrous pyroxenoids have the gen- and pyroxene (P) modules, thereby deriving wollastonite eral chemical formula MSiO3, where M most commonly (W), rhodonite (WP), pyroxmangite (WPP), ferrosilite III is Ca, Mg, Fe, and Mn. Their structuresconsist of single (WPPP), and pyroxene (P). The structures of rhodonite chains of silicate tetrahedraarrangedin layers parallel to (n: 5) and pyroxmangite (n : 7) areillustrated in Figure (100) that alternate with layers containing bands of di- l. Bustamite, another three-repeatpyroxenoid, is based valent cation octahedra; oxygenatoms are approximately on a different linkage ofoctahedral and tetrahedral layers closestpacked (Prewitt and Peacor, 1964). The various and thereforeis not a member of this series. This concept structures differ in the periodicity of the tetrahedralchain is discussedmore fully by Koto et al. (1976) and by and in the corresponding arrangement of the octahedrally Thompson (1978). coordinated cations. In this manner they constitute a A number of workers have addressedthe observed structural seriesthat has been classifiedaccording to the temperature,pressure, and compositional stability limits number n of tetrahedrabetweenofsets that intemrpt py- that exist for each pyroxenoid structure (e.g., Akimoto roxene-like chain configurations (Liebau, 1962).Pyrox- andSyono, 19721'Ito, l972;MareschandMottana, l9T6; ene represents one end member of this structuralseries: Ohashi and Finger, 1978;Brown et al., 1980). Akimoto a pyroxenoid with no offsets(n: oo). and Syono (1972) determined that a pyroxenoid of When the entire structural configuration of octahedral MnSiO. composition undergoes successive polymorphic bands and tetrahedral chains is considered,these struc- transformations from rhodonite to pyroxmangite to cli- tures are seen to constitute a polysomatic series, which is nopyroxene and ultimately to garnet-type structures as defined as a group of distinct structures constructed of pressure increases at constant temperature. A similar trend different numbers of slablike portions of end-member is observedwith respectto composition: As mean cation structures-in this case,of wollastonite and clinopyrox- size decreases, structureswith increasingn predominate. ene. All members of the pyroxenoid seriesmay be con- These observations raise an intriguing question: Can 0003-004x/88/0708-0798$02.00 798
Transcript
American Mineralogist, Volume 73, pages 798-808, 1988
Effects of compositional variation on the crystal structures of pyroxmangiteand rhodonite
LrNoa R. hncxxnyCorning Glass Works, Sullivan Park FR-51, Coming, New York 14831, U.S.A.
Cn.q.RLrs W. BunNrr.qlrDepartment of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts 02138, U.S.A.
Ansrnlcr
As adjacent members of the pyroxene-pyroxenoid polysomatic series, constructed offragments of the wollastonite (W) and pyroxene (P) structures, pyroxmangite (WPP) andrhodonite (WP) possess very similar structures with comparable site distortions and cationordering patterns. The two structures respond quite similarly to cation substitutions. Oc-tahedral and tetrahedral distortions generally lessen and the silicate chains straighten aslarger cations substitute for smaller. Both structures exhibit limited stepwise ordering ofcations over the octahedral sites, with large cations preferentially entering the sites on theedges of the octahedral bands. Detailed structural responses to cation substitution generallyparallel those observed in pyroxenes.
The characteristics of the inner octahedral sites strongly influence several structuralparameters, including the sizes and configurations ofthe outer polyhedra. There is, how-ever, no well-defined mean cation size limit that differentiates rhodonite from pyroxman-gite from pyroxene. Structural parameters for both rhodonite and pyroxmangite structureschange smoothly with composition and produce only minor structural adjustments. Theseadjustments, however, produce localized higher-energy structural configurations that clus-ter at the boundaries between the W and P modules of the structures. Such configurationsinclude a strongly kinked tetrahedral chain, short Si-Si distances, and a highly distortedoctahedron. In contrast, the P-P boundary in pyroxmangite is virtually distortion-free.Concentrations of strain energy at W-P boundaries likely play a major role in controllingphase transformations in this system.
INrnooucrroN structed by appropriately stacking these wollastonite (W)Pyroxenes and anhydrous pyroxenoids have the gen- and pyroxene (P) modules, thereby deriving wollastonite
eral chemical formula MSiO3, where M most commonly (W), rhodonite (WP), pyroxmangite (WPP), ferrosilite IIIis Ca, Mg, Fe, and Mn. Their structures consist of single (WPPP), and pyroxene (P). The structures of rhodonitechains of silicate tetrahedra arranged in layers parallel to (n: 5) and pyroxmangite (n : 7) are illustrated in Figure(100) that alternate with layers containing bands of di- l. Bustamite, another three-repeat pyroxenoid, is basedvalent cation octahedra; oxygen atoms are approximately on a different linkage ofoctahedral and tetrahedral layersclosest packed (Prewitt and Peacor, 1964). The various and therefore is not a member of this series. This conceptstructures differ in the periodicity of the tetrahedral chain is discussed more fully by Koto et al. (1976) and byand in the corresponding arrangement of the octahedrally Thompson (1978).coordinated cations. In this manner they constitute a A number of workers have addressed the observedstructural series that has been classified according to the temperature, pressure, and compositional stability limitsnumber n of tetrahedra between ofsets that intemrpt py- that exist for each pyroxenoid structure (e.g., Akimotoroxene-like chain configurations (Liebau, 1962). Pyrox- andSyono, 19721'Ito, l972;MareschandMottana, l9T6;ene represents one end member of this structural series: Ohashi and Finger, 1978; Brown et al., 1980). Akimotoa pyroxenoid with no offsets (n: oo). and Syono (1972) determined that a pyroxenoid of
When the entire structural configuration of octahedral MnSiO. composition undergoes successive polymorphicbands and tetrahedral chains is considered, these struc- transformations from rhodonite to pyroxmangite to cli-tures are seen to constitute a polysomatic series, which is nopyroxene and ultimately to garnet-type structures asdefined as a group of distinct structures constructed of pressure increases at constant temperature. A similar trenddifferent numbers of slablike portions of end-member is observed with respect to composition: As mean cationstructures-in this case, of wollastonite and clinopyrox- size decreases, structures with increasing n predominate.ene. All members of the pyroxenoid series may be con- These observations raise an intriguing question: Can
0003-004x/88/0708-0798$02.00 798
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE
TreLe 1. Crystal-structure refinements of pyroxmangite and rhodonite
Burnham (1971)D. Rothbard (unpub.)Finger and Hazen (1978)Ohashi and Finger (1975)Pinckney and Burnham (1988)Narita et al. (1977)
Finger and Hazen (1978)Murakami and Tak6uchi (1979)Peacor et al. (1978)Ohashi and Finger (1975)Narita et al. (1977)
' Lunar microgabbro." Synthesized by J. lto.t Taguchi mine: regionally metamorphosed manganese ore deposit.
tt Aiiro mine: mineralized manganese ore lens in chert.+ Synthesized by Akimoto and Syono (1972).$ Metamorphosed sedimentary evaporite sequence.
the phase relationships be readily understood in terms ofstructural attributes identifiable in each phase? The sta-bility of micas, for example, is related to the geometricfit of the octahedral and tetrahedral sheets (Hazen andWones, 1972, 1978); if the octahedral layer becomes solarge that normal tetrahedral rotations and polyhedraldistortions can no longer compensate for it, then the sheetsilicate will be rendered less stable relative to alternativestructures. We might anticipate that there could be sim-ilar constraints inherent in the linkages ofthe octahedralbands and tetrahedral chains ofpyroxenoids.
Pntvrous woRK
Although the chemical and structural relations amongthe pyroxenes have been studied intensively (e.g., Cam-eron and Papike, l98l), analogous studies for the pyrox-enoids are relatively scarce. Ohashi and Finger (1975)compared the structures of pyroxmangite and rhodoniteand noted the close topologic and configurational corre-spondence between certain cation polyhedra in the twostructures. Ohashi and Finger (1978) studied the role ofthe octahedral cations in the crystal chemistry of the three-tetrahedral repeat hydrous and anhydrous pyroxenoidsand demonstrated that the different stacking configura-tions oftetrahedral and octahedral layers determine theresulting cation-ordering patterns and limit the extent ofsolid solution.
Very little has been written, however, concerning therelationship between bulk composition and structuralstability of the intermediate pyroxenoids. Liebau (1962)noted that the chain configuration in pyroxenoids is ap-parently a function ofthe average size ofthe octahedrallycoordinated cations. More specifically, based on the ob-servation (e.g., Freed and Peacor, 1967) that pyroxenestability depends on the relative size of the Ml octahe-dron, Tak6uchi (1977) suggested that the sizes of the cat-ions in the inner Ml-like octahedra of pyroxenoids areprimarily responsible for their relative stabilities. Al-
though this correlation was briefly examined by Mura-kami and Tak6uchi (1979), the exact nature ofthe rela-tionship has not been defined. Nor has there been anysystematic study, similar to the plT oxene studies of Ohashiet al. (1975) and Ribbe and Prunier (1971), regarding thespecific efects of compositional variation, temperature,or pressure on these crystal structures.
This paper examines the effects of compositional vari-ation on the crystal structures of pyroxmangite and rho-donite, concentrating primarily on Mn-Mg substitutions.Because there is substantial compositional overlap be-tween the Mn-Mg pyroxmangite and rhodonite stabilityfields along the MgSiOr-MnSiO, join, these compositionsare particularly suitable for examining any developingstructural instabilities. Accordingly, data from previous
crystal-structure refinements of pyroxmangite and rho-
R h o d o n i t e P y r o x m o n g i t e
Fig. l. Projections ofthe rhodonite and pyroxmangite struc-
tures onto (100). Octahedral (M) sites in both structures are iden-
tified by number. After Ohashi and Finger (1975).
P y r o x m o n g i t e
800 PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE
donite (listed in Table l) have been assessed along withthose from a new refinement of a nearly pure MnSiO,pyroxmangite. Data for pyroxferroite (Burnham, l97l)are also included in the analysis.
We employ the CI setting for the pyrxoxenoids; datataken from previous refinements based on the PI settinghave been converted into their CI equivalents. The seriesofAppendix Tables Al through A,5 compiles selected datafor pyroxmangite and rhodonite taken or calculated fromthe crystal-structure refinements listed in Table l.
Srnucrunq.L vARIATToNS
Cell expansion
For the most part, the pyroxmangite and rhodonitestructures respond very similarly to compositional vari-ation. Cell parameters of both structures, for example,increase fairly smoothly with the substitution of Mn forMg, although small amounts of Ca present in natural rho-donites cause marked increases in the a and D cell di-mensrons.
A complete description of lattice strain due to chemicalsubstitution of a larger cation (Ca or Mn) for a smallerone (Mg) requires that a strain ellipsoid be calculated(Ohashi and Burnham, 1973). Using the cell parametersgiven in Appendix Table Al and the computer programsrRArN, written by Y. Ohashi, the magnitudes and ori-entations of strain ellipsoids for several compositionalincrements of pyroxmangite and rhodonite were calcu-lated. Because of the relative complexity and lower sym-metry of these structures, their ellipsoids are somewhatmore difficult to interpret than are those of pyroxenes.Nevertheless, several clear trends emerge. The pyrox-mangites, whose compositions range from (Mno ,rMgo rr)-SiO. to MnSiOr, exhibit two distinct strain-ellipsoid ori-entations depending on Mn content. These are summa-rized in Table 2 as average low- and high-Mn ellipsoids,with the breaking point at about MnroMgro. The orien-
TneLe 2. Principal strain components of expansion due to chem-ical substitution
Fig.2. Cation site occupancy, Mn/(Mn + Mg), as a functionof bulk composition for pyroxmangites along the join MgSiOr-MnSiOr. Points represent ayerage site occupancies over the twoor three sites indicated.
tation of the average strain ellipsoid for Ca-free rhodo-nites, which are all relatively Mn-rich, strongly resemblesthat of the high-Mn pyroxmangites.
The reason for this change in ellipsoid magnitude andorientation becomes clear if we examine cation orderingpatterns. Figure 2 plots M-site occupancies in pyroxman-gite as a function of bulk composition (Mn,Mg)SiOr; threeordering patterns are apparent. Although cation occupan-cies of the Ml, M3, and M4 sites vary linearly with bulkcomposition, the other sites display a limited stepwiseordering pattern, with the M2 and M6 sites showing amarked preference for small cations and the large, seven-coordinated M5 and M7 polyhedra taking up Mn earlyin the series.
Addition of Mn to a low-Mn pyroxmangite produces astrain ellipsoid in which the greatest expansions lie closeto the b-c plane, with the largest expansion about halfwaybetween the b and c axes. This reflects a marked expan-sion of the structure along directions in which the densityof M5-O and M74 bonds is highest. The relatively smallvolume expansion, compared with that of the equiva-lent composition increment (Mno'Mgor)SiOr-(Mnon,"Mg"oo,)SiOr, is due to the ability of the M5 and M7polyhedra, located at the edges ofthe octahedral band, toexpand into the adjacent void space without unduly dis-torting the structure.
As Mn content increases, expansion of the small andinner octahedra begins to contribute most to the overallexpansion. The resulting strain ellipsoid suggests a less
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE 801
distorted expansion or "swelling" ofthe octahedral banddue to the more isotropic expansion of the inner octa-hedra, with the greatest expansions perpendicular to theband and the least expansion parallel to the band becauseof the constraints imposed by the tetrahedral chain. Atthis stage, the direction ofgreatest expansion lies parallelto the "dense zone" ofcations, first described for pyrox-enes by Morimoto et al. (1966) and defined for pyrox-mangites and rhodonites by Aikawa (1979). In pyrox-mangite, the dense zone is parallel to [031].1, which forthe Ajiro sample is oriented 90", 30o, and 79" from the a,b, and c axes, respectively. The direction of largest ex-pansion is thus parallel to the zone having densest con-centration ofM and Si cations.
Silicate chain
Chemical substitutions in the octahedral sites neces-sarily influence the silicate chains as well since they linkthe octahedral bands. The chains respond to their newenvironment with expansion and distortion of individualtetrahedra and by rotation oftetrahedra with respect toeach other. Because the number of crystallographicallydistinct tetrahedra is fairly large, there are many degreesof freedom in the chains; thus substantial structural com-pensations can be achieved with only small adjustmentsin each parameter.
Individual tetrahedra. Mean Si-O bond lengths do notchange appreciably (<0.006 A) witfr the substitution ofMn for Me (App. Table A2). As in almost all single-chainsilicates, the Si-OC Oridging oxygen) bonds are longerthan the nonbridging bonds, and of the latter, the Si-OA(apical oxygen) bonds are longer than the Si-OB bonds.
In both pyroxmangite and rhodonite, tetrahedral dis-tortions decrease as mean octahedral cation size increases(App. Table A4). This effect is seen also in pyroxenes(Cameron and Papike, l98l) and is due to the dispro-portionate lengthening of the OC-OC edge of the tet-rahedron, which opens up the OC-Si-OC angle andthereby lessens the distortion. This is a simple yet effec-tive means of stretching the chain without increasing theaverage Si-O distance. Tetrahedral distortions are rough-ly the same for pyroxmangite and rhodonite of similarcompositions (App. Table A4); they increase along thechain from Sil to Si7, primarily because the Si4 throughSi7 tetrahedra share edges with octahedra. The distor-tions of the non-cdge-sharing tetrahedra are comparableto those in clinopyroxene (e.g., Ohashi, et al., 1975).
Tetrahedral chain angles. As the mean octahedral-cat-ion radius increases, the tetrahedral chain undergoes anoverall gradual straightening by about 2'(see App. TableA5). Not unexpectedly, the "straightest" chain angles (OC-OC-OC closest to 180") are those associated with the Pmodules. The most highly kinked angle is OC5-OC6-OC7in pyroxmangite, and its rhodonite equivalent is OC3-OC4-OC5. This angle is associated with the tetrahedral"triplet" of the W module and is located at the boundarybetween the W and P modules. Although the angle wid-ens by up to 6o as Mn replaces Mg, it remains significantly
more kinked in pyroxmangite than in rhodonite of iden-
tical composition. This suggests that the W-P boundary-and, as will be demonstrated, the W module itself-be-
comes less favorable energetically in pyroxenoids withsmaller mean cation size.
The Si-Si distances within the chains range ftom 2.94to 3.1 1 A. tne longer Si-Si distances are those associatedwith the P modules. The shorter ones are the Si6-Si7distances in pyroxmangite and the equivalent Si4-Si5 dis-tances in rhodonite, as well as the Sil-Si2 distances in
both structures. Both of these distances are located at theW-P module interface. The former is a result of the highlykinked chain angle mentioned above. The latter is theshortest Si-Si distance in each of these pyroxenoids andis a manifestation of the inherent peculiarity of the py-
roxenoid tetrahedral chain as it jogs sideways to follow
the octahedral band, causing two tetrahedra (Sil and Si2)to point in the same direction. Oxygens OBI and OB2thus lie on the same side of the "backbone" of the chainformed by the bridging oxygens. This is the natural con-sequence of connecting a W and a P module.
Octahedral band
The octahedral bands in chain silicates contain two ba-sically different kinds of sites, those on the inside of thebands and those on their edges. The inner octahedra pos-
sess edges defined by the apical oxygens ofthe tetrahedralchain. Except for the Ml octahedra in pyroxenoids, theseoctahedra are situated in the P modules, have environ-ments very similar to that of Ml in pyroxene, and, like
the pyroxene Ml octahedron, are relatively undistorted.The Ml octahedra of pyroxmangite and rhodonite, whichpossess three such "apical edges" because of the lateraljog ofthe tetrahedral chains across them, are the basis ofthe W module and as such have no pyroxene equivalents.The sites on the edges of the bands are distorted poly-
hedra that share edges with tetrahedra and correspond to
the M2 polyhedra in pyroxenes.In light ofthese correspondences, Tak6uchi (l 977) pro-
posed the designation of Mli for the inner octahedra (in-
cluding Ml) and M2i fot the outer polyhedra, where i :
0 , 1,2, ' ' ' andT : 1,2, ' ' ' . The Ml i s i tes in pyroxman-gite, for example, comprise Ml through M3. Althoughwe retain the original nomenclature for each individualsite in this paper, the concept of basically different Mliand M2j polyhedra is useful since both the geometries
and the resulting cation occupancies of the two groups
are quite distinct. The concept of Mllike and M2-likesites also facilitates comparison with pyroxenes.
Inner octahedra. The changes in mean M-O distance(NI5) as a function of site occupancy for the inner oc-tahedra in pyroxmangite and rhodonite are plotted inFigure 3. M4e.-" and M3.n.o are included in the figurealthough they are not true Mli octahedra. Several inter-esting features are apparent in these plots. Whereas theNFO of several of the octahedra, such as M I and M4 ofpyroxmangite, exhibit a linear relationship with cationsubstitution, the curves for M2 and M3 exhibit a slight
802
x A A - l ' \l v l - v 2 1 7
215
2 2 0
M 3 - O z B
2 . 1 4
M 2 - O
) 1 0
M 1 - O2 1 7
2 .15
o 20 40 60 ao 100Mg Occuponcy Mn
Fig. 3. Mean M-O bond distances (NF-O) of the regular oc-tahedra of pyroxmangite and rhodonite as a function of site oc-cupancy Mn/(Mn + Mg).
bend, which suggests that the oxygen framework of thesepolyhedra may be held open by the nature of the structureto a minimum M-O- of about 2.B A. Moreover, for anygiven cation occupancy, the mean bond lengths of therhodonite octahedra are almost always longer than thoseofthe equivalent pyroxmangite octahedra. This effect isparticularly pronounced for M2.
Distortion parameters for the inner octahedra of bothpyroxmangite and rhodonite (App. Table A.4) indicatethat, although distortions generally are slightly greater forMg-rich compositions, the presence of Ca in the outerpolyhedra does not cause significant distortion ofthe in-ner octahedra. Distortion is comparable for topologicallyequivalent octahedra of the two structures except that thedistortion of M2.noo is greater than that of M2o*-n, partic-ularly in angle variance. The M2 octahedra are discussedfurther in the next section.
Finally, it should be noted that the M3 octahedron inpyroxmangite, located at the boundary between adjacentP modules, is consistently less distorted than the other-wise similar M2 octahedron, located at the W-P bound-
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE
( A ) ary. This effect is observed also in the WPPP structure,ferrosilite III (Weber, 1983).
Outer octahedra. A plot of NFO for the outer sitesshows considerable scatter because of the varying coor-dinations of these polyhedra. Whereas M4o"-n and M3,n"oare fairly regular octahedra, the M5o--",.1.d and M7p*.nsites have effective oxygen coordinations described betteras seven than as six, with distortion from ideal increasingin the order M7e--" ( M5,noo < M5o*-n. This distortiondifference is due to a combination of polyhedral anglesand to the slightly different coordination characteristicsof the three polyhedra: M5o*-, and M5.n- are coordinatedby four short and three medium-toJong bonds, whereasM7e--, has four short, two medium, and one very longbond. By virtue of this one very long bond, M7o,-, is theclosest to six-coordinated; in fact, its distortion parame-ters calculated using the six shortest bonds are similar tothose of M2 in clinoferrosilite [(I) : 1.06, osz : 164(Ohashi et al, 1975)1.
Substitution of Mn for Mg does not cause significantdistortion of the large M5 and M7 polyhedra in pyrox-mangite. Addition of Ca to these sites, however, causesa considerable increase in distortion and a closer ap-proach to seven-coordination as short M-O bondslengthen disproportionately relative to long bonds, whichundergo little or no change. The rhodonite M5 polyhe-dron responds to Ca occupancy in the same manner. IJn-like in pyroxmangite, however, the rhodonite M5 poly-hedron responds also to Mg substitution, causing theoxygen configuration to shift to four short, one medium,and two long bonds. This polyhedron, therefore, tendstoward seven-coordination in Mg-rich as well as in Ca-rich rhodonites, although the geometry of the polyhedrais quite different.
The small irregular M6".-" and M4.hod octahedraundergo significant changes in configuration as cation sizeincreases, resulting in a coordination closer to five thanto six in Mn-rich compositions. Five-coordination is par-ticularly pronounced when Ca is present. Such seeminglyanomalous behavior, in which effective coordination de-creases as cation size increases, is readily understood inlight of the fact that, although large cations (except Ca)do enter this site, they are filling neighboring cation po-sitions at a much faster rate (recall Fig. 2). In pyroxman-gite, this causes the tetrahedral chain angle OC5-OC6-OC7 to straighten by up to 6", which in turn pulls theOA7 oxygen atom sharply away from the M6 cation (Fig.4). Charge repulsion also increases the distance betweenthe now-larger M6 cation and the Si7 cation with whichthe M6 cation shares a polyhedral edge. An importantresult of these relative movements is that, as Mn or Cafills the M5 and M7 sites, the OA7 oxygen moves awayfrom the M6 cation faster than the cation itself movesaway from Si7. Eventually the OA7 oxygen is farther fromthe M6 cation than is the Si7 atom; in Ca-rich pyroxfer-roite, these distances are 2.9A2 and 2.873 A, respectively(Burnham, l97l).
The M6 octahedron is a component of the W module
219
2 .19
2 1 7
215
2 1 3
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE 803
M 1 i
Mg sio=,20 oorr"r.uo 80
MnSio,
Fig. 5. Variation of grand mean Mll-O bond length, Mli,as a function of bulk composition for Mg-Mn pyroxmangitesand rhodonites.
seen by correlating their size with other structural param-eters and comparing the correlations with those observedin pyroxenes. It is convenient to define Mfi for pyrox-enoids as the grand mean Mll-O bond length, where Mlioctahedra include Ml througlr M3 in pyroxmangite andM I and M2 in rhodonite (Murakami and Tak6uchi, 1979).A plot of Mfi as a function of bulk composition along ornear the join MgSiO.-MnSiO, for pyroxmangites andrhodonites is given in Figure 5.
Unlike pyroxenes, the cell parameters of natural py-roxmangite and rhodonite follow MTI only approximate-ly, because moderate amounts of Ca in the outer M27sites exert little influence on the Mli ocatahedra but havea marked effect on the cell edges. In most respects, how-ever, the relationship ofMll'- with other structural param-eters parallels those observed in pyroxenes.
More unexpected is the influence of the Mli octahedraon the M2l polyhedra. One significant observation thathas emerged from pyroxene studies is that the sizes ofpolyhedra are frequently constrained by the nature ofthestructure itself and that an important control is the Mloctahedron. Ghose and Wan (1975), for example, ob-served that in MCaSirOu clinopyroxenes (i.e., with con-stant M2 cation occupancy), the average M2-O distanceincreases linearly with increasing mean Ml-O distance,which implies that the size of the outer octahedra maybe ultimately controlled by the size of the inner octahe-dra. Figure 6 indicates that a similar relationship existsfor pyroxmangite and rhodonite as well. Taking the term
@ to represent the grand mean M274 bond length foreach particular structure, two curves of M2j versus Mlihave been plotted for each structure, reflecting two dif-ferent assignments of coordination numbers to the M27polyhedra. Although the plots for pyroxmangite are very
( A )
2 2 2
214
2 1 6
2 1 4
) 1 )
a P x m no Rhod
Fig. 4. Configuration changes ofthe M6 polyhedron in py-roxmangite as Mn/(Mn + Mg) increases. Double arrows indicaterelative movements of atoms. As the tetrahedral chain angleOC5-OC6-OC7 straightens, the OA7 oxygen atom moves awayfrom the M6 cation. The distance between the M6 cation andthe Si7 atom also increases because of increased charge repul-sion, but at a slower rate. Eventually, the M6 cation is closer toSi7 than it is to OA7, resulting in M6 becoming five-coordinat-ed.
and as such is best suited geometrically for larger cations(Ohasi and Finger, 1978). Presence of small cations in thesite causes the extreme kinking of the OC5-OC6-OC7tetrahedral chain angle (discussed above) and contributesto the "unfavorableness" of W modules in structures withsmall mean cation size.
Significance of the Mli octahedra
The pyroxene Ml octahedron does not vary drasticallyamong the different structure tlpes (ortho, clino, and pro-to); it is appropriate to consider it the major buildingblock of the pyroxene structure (Cameron and Papike,1981). Various workers (e.g., Ribbe and Prunier, 1977;Cameron and Papike, 1981) have demonstrated that theMl octahedron determines to a large extent the unit-celldimensions, the relative displacement and kinking of thetetrahedral chain, and the size and configuration of theM2 polyhedron. We therefore might anticipate that thetopologically equivalent Mli octahedra would play anequally central role in the pyroxenoids, and indeed thisis the case.
The significance of the Mli octahedra is most easily
oc5
Pynoxmang j t e
804 PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE
iT:
2.40
230
z z v
M 1 i/ i \ 2 . 1 o
2.OO
1 . 9 0
2 90 3.OO 3r O 3.20 3.30 3 40 3.50
L (A )
Fig. 7. Variation of grand mean bond length MTi with Zparameter (see text) for pyroxmangites and rhodonites.
of the M27 polyhedra are linked to and constrained bythe sizes ofthe inner octahedra.
As this correlation seems to hold true for chain silicatesin general, it may be one way in which the Mli octahedraultimately control the relative stability of the pyroxenoidstructures (including pyroxenes), as suggested by Tak6u-chi(1977). The relationship between the Mll cations andpyroxenoid stability, however, is by no means a simpleone. For example, the large degree of overlap in Mlivalues between pyroxmangites and rhodonites (Fig. 5), asindeed in the bulk composition itself, indicates that al-though a trend does exist, there certainly is no cut-offpoint in Mli size that controls the relative stability ofpyroxmangite and rhodonite.
In an attempt to define the correlation between the Mlications and pyroxenoid stability, Murakami and Tak6u-chi (1979) examined the relationship between MTI andthe average distance between apical oxygens along thetetrahedral chain, which is a qualitative measure of chainextension. They defined a quantity
/ , \L: l> t i + to l /@ + t) ,
\ r l /
where n is the periodicity ofthe tetrahedral chain and Zis essentially the average distance between apical oxygensincluding the apical-apical distance /o across the triplet.They then plotted I versus Mli for a number of pyrox-enoid and pyroxene structures; their plot is reproducedin Figure 7. The set of data points for each structure typedefines an approximately straight line, with very littleoverlap in Z values between structures. Murakami andTak6uchi therefore concluded that there is a critical limitof I for each structure type. They suggested that this limitin Z might be related to /o because of the systematic changein /0 with Mfi. tne lerm lo, however, merely reflects thedistance across the relatively undistorted Ml octahedron,and there is considerable overlap in /o values for pyrox-mangite and rhodonite (see Table 3).
t Bust
Pxmn 4JY Rhooa7
. . X ,j t ' . a
t'! "
' P x
o o
2 1 6 2 1 8
M 1 i
2 2 a 2 . 2 2 2 2 4 ( A )
u 2 )
Fig. 6. Variation of grand mean bond lengths M| with Mll-.(a) Pyroxmangites: the lower curve (filled circles) considers all
M sites as six-coordinated except M6 in CWB-P pyroxferroite,which is taken as five-coordinated. The upper curve (open cir-cles) considers M5 and M7 as seven-coordinated.
(b) Rhodonites: filled triangles take all M sites as six-coordi-nated, whereas open triangles take M5 as seven-coordinated.Significance of the dashed composite curve is explained in thetext.
well defined, the rhodonite data produce breaks in bothcurves. These breaks are a result of the previously dis-cussed change in M5.noo coordination with increasing Mgcontent. Because the sixth and seventh oxygen atoms arethen approximately equidistant from the M5 cation, con-sidering M5.n* as six-coordinated produces artificially lowdata points for the relatively Mg-rich rhodonites MT-Rand FH-R. When M5 is taken as seven-coordinated inthese Mg-rich phases, however, their distances plot onthe lower curve defined by the points of the Mn-rich rho-donites (in which M5 is more nearly six-coordinated).This composite curve is given as the dashed line in Figure6b.
The strong correlation apparent in these curves is evenmore remarkable given that it encompasses not onlyMn + Mg substitution but also Ca-bearing compositionssuch as the Ca-rich pyroxferroite CWB-P. It is clear,therefore, that regardless of bulk composition, the sizes
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE 805
We suggest instead that the lack of overlap in Z values,at least between pyroxmangite and rhodonite, is due pri-marily to the significantly different sizes and distortionsof the M2 octahedron. Although the M2 octahedron inrhodonite is topologically equivalent to that of pyrox-mangite, i.e., located in the P module adjacent to theW-P boundary, it is much larger and more distorted inrhodonite than in pyroxmangite. The difference is inde-pendent of bulk composition and apparently is an inher-ent, structurally controlled phenomenon. The high dis-tortion of this octahedron suggests that the P module maybe less stable in rhodonite than in pyroxmangite of iden-tical composition.
CoNcr,usroNs
Rhodonite (WP) and pyroxmangite (WPP) are adjacentmembers of the pyroxenoid-pyroxene polysomatic series;they possess very similar structures with octahedral andtetrahedral sites that are topologically equivalent and ex-hibit comparable distortions and cation ordering pat-terns. As a result, the two structures respond in much thesame way to compositional variation.
As larger cations replace smaller ones in these struc-tures, octahedral and tetrahedral distortions generallylessen, and the silicate chains straighten. Both structuresexhibit limited stepwise ordering of cations, with largercations preferentially entering the large sites on the edgesof the octahedral bands. Very large cations, such as Ca,appear restricted to these sites and so place limitationson bulk composition. Mean Si-O distances change verylittle. These structural responses to cation substitutionmirror those observed in pyroxenes.
Also as in pyroxenes, the nature of the octahedra onthe inside of the band (Mli octahedra) strongly influencesthe entire structure, including the ultimate size and con-figuration ofthe outer polyhedra. Although there is a gen-eral correlation between cation size and structure type,with the number of P modules per unit cell increasingwith decreasing mean cation size, there is no critical cut-off in mean size that differentiates pyroxmangite fromrhodonite or, for that matter, from pyroxene. Structuralparameters for both pyroxenoid structures changesmoothly with composition and produce only minorstructural adjustments.
These adjustments, however, produce localized higher-energy structural configurations that cluster at the bound-aries between the W and P modules, leaving the P-Pboundaries virtually strain-free. Decreasing mean cationsize in a pyroxenoid, for example, results in smaller Pmodules. The configurations and minimum sizes of Wmodules, however, are constrained by the tetrahedralchains; the chains develop strong kinks at the W-Pboundaries, forcing pairs of Si cations close together toSi-Si distances significantly shorter than the 3.0 A pro-posed by Hill and Gibbs (1979) as the lower limit ofnonbonded Si-Si contacts. Both the W modules and theW-P boundaries therefore become energetically less sta-ble in compositions with smaller mean cation size. It is
TneLe 3. Additional parameters for pyroxmangite and rhodonite
. Mllcomprises M1-M3e, M1-M2'. lvrlr--: mean M-O bond distanceof these octahedra.
-- M2l comprises M4-M7", M3-M5F. All polyhedra are considered to besix-coordinated.
f M2l sites are the same, only M5", M7p, and M5" are considered tobe seven-coordinated.
f lo and L are defined in the text.$ M6 is considered to be five-coordinated.
significant that in pyroxmangite, the P modules remarnrelatively undistorted regardless of composition; nearlyall of the distortion occurs at the W-P boundaries.
Because decreasing cation size in rhodonite leads todistortions that cause higher-energy configurations to de-velop at W-P boundaries, W modules are destabilizedrelative to P modules, and P-P boundary stability isthereby enhanced. This tends to favor nucleation ofad-ditional P modules within the mineral. A mechanism forsuch a nucleation process consistent with this notion hasbeen formulated by Angel et al. (1984) and by Veblen(1985) to explain phase reactions in the pyroxene-p)'rox-enoid system. Conversely, as cation size increases in py-roxmangite, the W module is stabilized and W-P bound-aries become more favorable.
Accumulation of higher-energy strained configurationsat W-P boundaries as cation substitution proceeds sug-gests that a yet-unidentified maximum strain limit maycontribute energetically to phase transformations in thissystem. Indeed, such buildup of distortion-induced strainenergy at W-P interfaces can explain the common obser-vation in transmission electron microscopy (Ried andKorekawa, 1980; Czank and Liebau, 1980; Alario Francoet al., 1980; Czank and Simons, 1983) that excess P mod-ules, and hence excess P-P boundaries, are much morecommon in pyroxenoids than are excess W modules.
AcrNowr,nocMENTS
This research was supported by National Science Foundation Srant EAR-
7920095 to C.W B. We thank D. R. Peacor and R. J. Reeder for helpful
reviews, through which the manuscript was substantially improved.
RrrnnrNcns crrno
Aikawa, N. (1979) Oriented intergrowth ofrhodonite and pyroxmangite
and their transformation mechanism. Mineralogical Joumal, 9, 255-
269.Akimoto, S., and Syono, Y (1972) High pressure transformations in
MnSiOr. American Mineralogist, 57, 7 6-84.
806
Alar io Franco, MA, Jef ferson, DA., Pugh, NJ. , Thomas, J.M., andBishop, A C (1980) Lattice imaging of structural defects in a chainsilicate: The pyroxenoid mineral rhodonite. Materials Research Bulle-t in, 15,73-19
Angel, R J, Price, G.D., and Putnis, A (1984) A mechanism for pyrox-ene-pyroxenoid and pyroxenoid-pyroxenoid transformations Physicsand Chemistry of Minerals, 10,236-243.
Brown, P E, Essene, E J., and Peacor, D R (1980) Phase relations in-ferred from field data for Mn pyroxenes and pyroxenoids. Contribu-tions to Mineralogy and Petrology,74.417425.
Bumham, C W (1971) The crystal structure of pyroxferroite from MareTranquillitatis. Proceedings of the Second Lunar Science Conference,| , 47-57 .
Cameron, M , and Papike, J.J (1981) Structural and chemical variationsin pyroxenes American Mineralogist, 66, l-50.
Czank, M, and Liebau, F. (1980) Periodicity faults in chain silicates: Anew type of planar lattice fault observed with high resolution electronmicroscopy. Physics and Chemistry of Minerals, 6, 85-94
Czank, M, and Simons, B (1983) High resolution electron microscopicstudies on ferrosilite III Physics and Chemislry of Minerals, 9, 229-
Finger, L W, and Hazen, R M (1978) Refined occupancy factors for syn-thetic Mn-Mg pyroxmangite and rhodonite Camegie Instilution ofWashington Year Book 77, 850-853
Freed, R.L., and Peacor, D.R. (1967) Refinement of the crystal structureof johannsenite. American Mineralogist, 52, 7 O9-i 20
Ghose, S., and Wan, C (1975) Crystal structures of CaCoSirOu andCaNiSirOu: Crystal chemical relations in C2lc pyroxenes (abs ). EOS,56. tO76
Hazen, R.M, and Wones, D.R. (1972) The effect of cation substitutionson the physical properties of trioctahedral micas. American Mineral-ogist, 57, lO3-129
-(1978) Predicted and observed compositional limits of trioctahe-dral micas American Mineralogist, 63, 885-892.
Hill, RJ., and Gibbs, GV. (1979) Variation in d(T-O), d(T T) and<T-O-T in silica and silicate minerals, phosphates, and aluminatesActa Crystallographica, B35, 25-30.
Ito, J. (1972) Rhodonite-pyroxmangite periteclic along the join MnSiO,-MgSiO, in air American Mineralogist, 57,865-816
Koto, K , Morimoto, N , and Narita, H. (1976) Crystallographic relation-ships of the pyroxenes and pyroxenoids. Journal of the Japanese As-sociation of Mineralogy, Petrology, and Economic Geology, 71,248-
Liebau, F. (1962) Die Systematik der Silikale, Natumissenschaften, 49,48 l-49 l
Maresch, WV., and Mottana, A (1976) The pyroxmangite-rhodonitetransformation for the MnSiO, composition Contributions to Miner-alogy and Petrology, 55,69-79.
Morimoto, N., Koto, K, and Shinohara, T. (1966) Oriented transfor-mation, johannsenite to bustamite. Mineralogical Journal, 5, 44-64
Murakami, T, and Tak6uchi, Y (1979) Structure ofsynthetic rhodomte,Mno.rrMgo,,rSiOr, and compositional transformations in pyroxenoids.Mineralogical Joumal, 9, 186-304
Narita, H, Koto, K, and Morimoto, N (1977) The crystal structures ofMnSiO, polymorphs (rhodonite- and pyroxmangite-type). Mineralog-ical Journal, 8, 329-342.
Ohashi, Y., and Bumham, CW (1973) Clinopyroxene lattice deforma-tions: The roles of chemical substitution and temperature. AmericanMineralogist, 58, 843-849.
Ohashi, Y., and Finger, LW (1975) Pyroxenoids: A comparison ofre-fined structures ofrhodonite and pyroxmangite Carnegie Institution ofWashington Year Book 74, 564-569.
-(1978) The role of octahedral cations in pyroxenoid crystal chem-istry. I. Bustamite, wollastonite, and the pectolite-schizolite-seranditeseries American Mineralogist, 63, 274-288
Ohashi, Y., Bumham, C W., and Finger, L W (1975) The effect of Ca-Fe substitution on the clinopyroxene crystal structure American Min-eralogist , 60,423434
Peacor, D.R , Essene, E.J., Brown, P.E, and Winter, G A (1978) Thecrystal chemistry and petrogenesis ofa magnesian rhodonite. AmericanMineralogist, 63, | 13'7 -l | 42.
PINCKNEY AND BURNHAM: PYROXMANGITE AND RHODONITE
Pinckney, L.R., and Burnham, CW (1988) High-temperature crystalstructure of pyroxmangite. American Mineralogist, 73, 809-817.
Prewitt, C.T , and Peacor, D.R. (l 964) Crystal chemistry ofthe pyroxenesand pyroxenoids American Mineralogist, 49, 1527-1542.
Ribbe, P H , and Prunier, A R, Jr. (1977) Stereochemical systematics ofordered C2lc silicate pyroxenes. American Mineralogist, 62, 7 lO-720.
Ried, H , and Korekawa, M. (1980) Transmission electron microscopy ofsynthetic and natural fiinferketten and siebenerketten pyroxenoids.Physics and Chemistry of Minerals, 5, 351-365.
Robinson, K , Gibbs, G.V , and Ribbe, P H ( 197 1) Quadratic elongation:A quantitative measure of distortion in coordination polyhedra. Sci-ence ,172 ,567 -570 .
Tak6uchi, Y. (197'7) Designation of cation sites in pyroxenoids. Miner-alogical Journal, 8, 43 1-438,
Thompson, J.B., Jr (1978) Biopynboles and polysomatic series. Ameri-can Mineralogist, 63, 239-249
Veblen, D R (1985) TEM study of a pyroxene-to-pyroxenoid reactionAmerican Mineralogist, 70, 885-901.
Weber, H.-P. (1983) Ferrosilite III, the high-temperature polymorph ofFeSiO,. Acta Crystallographica, C39, l-3
P o l y h e d r a l v o l u a e 6 a n d d l 6 t o r t i o n s w e ! e c a l c u l a t e d u s i n g t h e c o r p u t € rp r o g r a n V O L C A I , w r i t l e n b y L . c . 8 l n g e r . T h e d l a l o r t l o n f r a r a r € l e l E o fn e a n q u a d r a t i c e l o n g a t i o n a n d a n g l e v ! r i r n c e . r e d e f i n e d b y B o b i n 8 o ne t a I ( 1 9 7 1 ) .
1 7 . 5 8 1 7 . 0 0 T a b l e A 5 . l e t r e h e d r a l c h a i n l n g l e s i n p y r o x r a n g i t e a n a l r h o d o n i t e
1 2 . 5 1 1 2 . 1 0 L 3 . 2 3
1 , 0 ? 1 1 . 0 6 5 1 . 1 1 9
1 8 5 , 9 2 L 1 2 . 1 1 2 8 9 . 2 3
5 , 6 8 5 . 4 6 7 . 0 6
1 3 . 9 0 1 3 . 1 S L 4 . 6 2
1 . 0 6 6 1 . 0 6 0 1 . 0 6 6
1 9 6 . 1 2 r ? 9 . 0 3 2 0 7 . 1 6
1 7 . 1 ? L 6 . 3 1 1 8 . 0 1
P y r o x [ a n a i t e 5 :
A n g l e r N a r - P FH-P ROIh-P CWB-P
x ? v ( 6 ) 1 { . 5 3 1 { . 4 0 1 4 . 3 8
\ 1 . 0 5 5 1 0 5 5 1 . 0 6 6
o 2 t l s , B 3 1 8 5 . 5 g r 9 o . 1 g
v ( 7 ) 1 8 . 1 5 1 8 , 0 0 1 ? , 8 1
o c l - 2 - 3o c 2 - l - 4o c 3 - 4 - 5o c 4 - 5 - 6o c 5 - 5 - 7o c 5 - ? - 1
I A n g l e s f o r D e d b y b r l d g i n g o r y g e n . t o i 6 i 9 . 9 . r c a d O C I - 2 - 3 a s t h e a n g l cfo r t r€d by oxygens OC1-OC2-OC3
