SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES • NUMBER 21

Schreibersite Growth and ItsInfluence on the Metallography

of Coarse-StructuredIron Meteorites

Roy S. Clarke, Jr.and Joseph I. Goldstein

SMITHSONIAN INSTITUTION PRESS

City of Washington

1978

Authors Copy

A B S T R A C T

Clarke, Roy S., Jr., and Joseph I. Goldstein. Schreibersite Growth and ItsInfluence on the Metallography of Coarse-Structured Iron Meteorites. Smithson-ian Contributions to the Earth Sciences, number 21, 80 pages, 28 figures, 20 tables,1978. —The role that schreibersite growth played in the structural developmentprocess in coarse-structured iron meteorites has been examined. The availabilityof many large meteorite surfaces and an extensive collection of metallographicsections made it possible to undertake a comprehensive survey of schreibersitepetrography. This study was the basis for the selection of samples for detailedelectron microprobe analysis. Samples containing representative structuresfrom eight chemical Groups I and IIAB meteorites were selected.

Electron microprobe traverses were made across structures representative ofthe observed range of schreibersite associations. Particular emphasis was placedon schreibersite-kamacite interface compositions. An analysis of these data hasled to a comprehensive description of the structural development process.

Massive schreibersite, one of the four major types of schreibersite encoun-tered, may be accounted for by equilibrium considerations. Subsolidus nuclea-tion and growth with slow cooling from temperatures at least as high as 850° C,and probably much higher, explain the phase relationships that one sees inmeteorite specimens. The retention of taenite in the octahedrites establishesthat bulk equilibrium did not extend as low as 550° C. Schreibersite undoubtedlycontinued in equilibrium with its enclosing kamacite to lower temperatures.

A second type of schreibersite to form is homogeneously nucleated rhabdite.It nucleated in kamacite in the 600° C temperature range, either as a conse-quence of low initial P level or after local P supersaturation developed followingmassive schreibersite growth.

A third type of schreibersite is grain boundary and taenite border schreiber-site. It formed at kamacite-taenite interfaces, absorbing residual taenite. Nuclea-tion took place successively along grain boundaries over a range of temperaturesstarting as high as 500° C or perhaps slightly higher. Grain boundary diffusionprobably became an increasingly important factor in the growth of theseschreibersites with decreasing temperature.

The fourth type of schreibersite is microrhabdite. These schreibersitesnucleated homogeneously in supersatuated kamacite at temperatures in the 400°C range or below.

P diffusion controlled the growth rate of schreibersite. The Ni flux to agrowing interface had to produce a growth rate equal to that established by theP flux. This was accomplished by tie line shifts that permitted a broad range ofNi growth rates, and these shifts account for the observed range of Niconcentrations in schreibersite. Equilibrium conditions pertained at growthinterfaces to temperatures far below those available experimentally. Kineticfactors, however, restricted mass transfer to increasingly small volumes ofmaterial with decreasing temperature.

OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded inthe Institution's annual report, Smithsonian Year. SERIES COVER DESIGN: Aerial view of UlawunVolcano, New Britain.

Library of Congress Cataloging in Publication DataClarke, Roy S., Jr.Schreibersite growth and its influence on the metallography of coarse-structured iron meteorites.(Smithsonian contributions to the earth sciences; no. 21)Bibliography: p.1. Schreibersite. 2. Metallography. 3. Meteorites, Iron. I. Goldstein, Joseph I., joint author.

II. Title. III. Series: Smithsonian Institution. Smithsonian contributions to the earth sciences;no.21.

QEl.S227no. 21 [QE395] 550'.8s [552] 77-10840

Authors Copy

Contents

Page

Introduction 1Acknowledgments 2

Historical Background 2Introduction 219th-century Studies of Schreibersite 3Early 20th-century Studies of Iron Meteorites 4Modern Work on Metallic Phases of Iron Meteorites 5Related Materials 8

Experimental 9Results 11

Coahuila 13Bellsbank 16Ballinger 18Santa Luzia 21Lexington County 25Bahjoi 29Goose Lake 30Balfour Downs 31

Discussion 37Equilibrium Considerations of Phase Growth 37

Coahuila 40Ballinger 42Santa Luzia 43Lexington County and Bahjoi 44Goose Lake and Balfour Downs 45Bellsbank 46

Low Temperature Phase Growth and the Equilibrium Diagram 48Diffusion-Controlled Schreibersite Growth 54Cooling Rate Variations 58Interface Data and Schreibersite Distribution 59Interface Data and the a/a + Ph Boundary 66Schreibersite with Cohenite Borders 70

Summary and Conclusions 73Appendix 75Literature Cited 77

Authors Copy

Schreibersite Growth and ItsInfluence on the Metallography

of Coarse-StructuredIron Meteorites

Roy S. Clarke, Jr. and Joseph I. Goldstein

Introduction

Meteorites are currently understood to be theoldest rocks available for scientific study, contain-ing components and structures that span the pe-riod from the final stages of solar nebula conden-sation to the present (Anders, 1962, 1971; Gross-man and Larimer, 1974; Wasson, 1974). They arefragments of parent bodies that accreted frompreexisting aggregations of material during theperiod of planetary system formation some 4.6billion years ago. These parent bodies were subse-quently disrupted into smaller objects that thenhad independent existences in space. Individualfragments eventually intercepted the earth andlanded as recoverable meteorites. Meteorite struc-tures and compositions have undergone varyingdegrees of modification while resident in theirparent bodies, as small bodies in space, on theirpassage through the atmosphere, and on landingon the earth's surface. Further changes resultfrom long residence time on the ground and maycontinue during storage in collections. Bearingevidences of these complex histories, meteoritesare samples not only from a far distant place butalso from a far distant time, having been preservedin a remarkably gentle environment when com-

Roy S. Clarke, Jr., Department of Mineral Sciences, National Museumof Natural History, Smithsonian Institution, Washington, D. C.20560. Joseph I. Goldstein, Department of Metallurgy and MaterialsSciences, Lehigh University, Bethlehem, Pennsylvania 18015.

pared to terrestrial or lunar rocks. As a conse-quence, meteorites have a unique place in thestudy of the development of the planetary system,yielding information that is available from noother source.

Iron meteorites represent only a small fractionof the meteorites that have been observed to fall,about five percent. Fortunately, many specimensfrom ancient falls have been preserved under suchconditions that deterioration has not been severeand are available for study (Buchwald, 1976). Ifthis were not the case, our view of iron meteoritecompositions and structures would be much nar-rower than it is. Only one of the meteorites usedin this study is an observed fall.

The spectacular metallographic structures re-vealed on prepared surfaces of iron meteoriteshave fascinated scientist and dilettante alike formore than 160 years (Perry, 1944). The Widman-statten pattern is the historical distinguishing char-acteristic of the populous octahedrite classes ofiron meteorites (Goldstein and Axon, 1973). At anearly date this structure was understood to resultfrom very slow cooling at relatively low tempera-tures in the meteorite parent body. The tempera-ture range through which this structure developsis now believed to be from 700° to 350° C, temper-atures well below those where silicate systems un-dergo change. Interpretation of metallographicstructures, therefore, gives information on a late,low temperature period in the history of a parent

1

Authors Copy

SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

body, information that is not readily availablefrom other types of meteorite studies.

Modern interpretations of iron meteorite struc-tures have been based on the Fe-Ni equilibriumdiagram (Goldstein and Olgivie, 1965a), as it wasthe only applicable system known with sufficientaccuracy in the low temperature range required.The Widmanstatten pattern, therefore, resultedsimply from taenite transforming to kamacite withdecreasing temperature as a solid state, diffusioncontrolled reaction. Recently the iron-rich cornerof the Fe-Ni-P equilibrium system has been deter-mined experimentally down to 550° C, leading tonew possibilities for investigating meteorite struc-tures (Doan and Goldstein, 1970). In this studythe metallography of a group of schreibersite-con-taining coarse-structured iron meteorites has beenexamined using this low temperature Fe-Ni-P dia-gram. It will be shown that equilibrium growth athigher temperatures combines with diffusion-con-trolled growth at lower temperatures to explainobserved structure more comprehensively thanwas possible using the simpler system. Schreiber-site growth becomes an integral part of the overallstructural development process, and schreibersitemay no longer be considered an "inclusion" thatcan be safely ignored when structural developmentis discussed.

ACKNOWLEDGMENTS

A number of colleagues and associates havecontributed invaluable assistance throughout thecourse of this study. The support and encourage-ment of departmental chairman William G. Mel-son and his predecessor Brian Mason are acknowl-edged with sincere thanks. Erik Randich providedpenetrating discussion of many of the problemsstudied here, and his analysis of ternary diffusionas applied to schreiberate growth and meteoritecooling rates was made available to us prior topublication. A. D. Romig, Jr., of Lehigh Univer-sity made available to us data on the low tempera-ture Fe-Ni-C system prior to publication. EugeneJarosewich and Charles R. Obermeyer III fur-nished unstinting help with the electron micro-probe work. Grover C. Moreland and RichardJohnson prepared many excellent polished sec-tions. Ann Garlington and Kathy P. Porter werescrupulous in their attention to the details ofmanuscript preparation. Dante Piacesi, Jr., of the

Office of Computer Services was responsible forthe preparation of data reduction programs. Agrant from the Secretary's Fluid Research Fundsupported a brief visit to Cambridge University,Cambridge, England, in the fall of 1974, where aserious beginning was made on the interpretationof the data. The support of the GeochemistryProgram, Division of Earth Sciences, National Sci-ence Foundation, through Grant No. DES 74-22518 is also acknowledged. The acronym USNM(for the former United States National Museum)is used for catalog numbers in the National Mu-seum of Natural History, Smithsonian Institution.

This paper was presented to the George Wash-ington University, Washington, D.C., by R. S.Clarke, Jr. in partial fulfillment of the require-ments for the Ph.D. degree.

Historical Background

INTRODUCTION

Meteorites, with their dramatic arrivals on earth,exotic origins, and unusual structures, have at-tracted much more than routine scientific interestever since they became respectable objects of studyat the end of the 18th century. Many of the leadersin the development of 19th-century physical sci-ence made significant contributions to their study,and several of them will be mentioned below. Theolder literature is extensive, but it is mainly obso-lete for other than descriptive and historical pur-poses. A critical review of this material was notattempted, but a survey of selected early develop-ments will be given. Scientific interest in schreiber-site dates from near the beginning of meteoritestudies, and it is interesting to place modern workin this perspective.

The first major development in the metallogra-phy of meteorites was the discovery of the phe-nomenon that has become known as the Widman-statten pattern. This structure was first discoveredby William Thomson, an Englishman living inexile in Naples. He published a French version ofhis findings along with a good illustration in aSwiss journal in 1804. What appears to be anidentical paper, only in Italian, dated 6 February1804, was published in Siena four years later (G.Thomson, 1804, 1808). Modern writers on meteor-itics have been unaware of the earlier paper by

Authors Copy

NUMBER 21

Thomson, and this has led to confusion as towhether Thomson or Von Widmanstatten reallyhad priority. The record seems to be unambigu-ously in favor of Thomson. Marjorie Hooker ofthe U.S. Geological Survey retrieved this 1804paper, and its existence was mentioned in a foot-note in a biographic study of Thomson by C. D.Waterston (1965: 133). The writer ran across Wa-terston's paper among unindexed material in theF. A. Paneth meteorite literature collection thatwas deposited in the Smithsonian Institution in1974.

Alois von Widmanstatten, Director of the Impe-rial "Fabrik-Produkten-Cabinett" (Industrial Prod-ucts Collection), Vienna, discovered the phenom-enon that now bears his name independently in1808. He devoted many years to its study andcirculated his observations privately, drawing thephenomenon to the attention of students of mete-orites. Von Widmanstatten's observations were fi-nally published in 1820 by his co-worker, Carl vonSchreibers (1820). The history of the discovery ofthe Widmanstatten pattern has been reviewed byPaneth (1960) and Smith (1960, 1962). Smith'sreview contains excellent reproductions of earlyprints and a lengthy translation of Schreibers'description of Widmanstatten's work. These earlyreproductions of iron meteorite structures by Wid-manstatten not only represent an advance in thescience of meteoritics but also one in the art ofprinting.

During the 19th century descriptive studies ofmeteorites advanced with the broadening base ofunderstanding in mineral chemistry and relatedphysical science. Many of the iron meteoritesknown today were already represented in collec-tions, and a number of them had been carefullydescribed in the literature. The major phases ofiron meteorites had been characterized and givenmineral names: kamacite (low-nickel Ni-Fe, fer-rite, a-iron, b.c.c), taenite (high-nickel Ni-Fe,austenite, y-iron, f.c.c), plessite (fine-grained in-tergrowth of kamacite and taenite), troilite (FeS),cohenite (Fe3C), and schreibersite ((Fe,Ni)3P) wereall known and their bulk compositions understood.Accessory minerals such as daubreelite (FeCr2S4),chromite (FeCr2O4), and graphite (C) were alsoknown. The Widmanstatten pattern was under-stood to have formed by slow cooling under crys-tallographicly controlled conditions, kamacite

plates forming parallel to the faces of an octahe-dron, giving the name octahedrite to those meteor-ites that displayed this pattern on polished andetched surfaces.

19TH-CENTURY STUDIES OF SCHREIBERSITE

Schreibersite was first characterized in 1832 bythe great Swedish chemist J. J. Berzelius (1832a,b,1833, 1834). He isolated brittle, silver-white grainsof a magnetic material from an acid insolubleresidue of the B©humilitz, Bohemia, coarse octa-hedrite. Berzelius' elaborate analytical procedureyielded a surprisingly good analysis for the time,demonstrating that his material was an iron-nickel-phosphorus compound containing about 14 per-cent phosphorus and considerably more nickelthan the bulk of the meteorite. In 1834 he reportedsimilar analyses for schreibersite from the Elbo-gen, Bohemia, medium octahedrite and the Kras-nojarsk, Siberia, pallasite (the historic Pallas Iron).It is also interesting to note that Berzelius' conceptof the Widmanstatten pattern included the ele-ment phosphorus, as is indicated by the followingquotation:

Es mochte wahrscheinlich seyn, dass die sogenannten Widman-stddtschen Figuren einer, der hier analysirten Schuppen analo-gen, Verbindung zwischen Eisen, Nickel und Phosphor, zuzu-schreiben seyen, welches zu erforschen ich aus Mengel anMaterial verhindert bin.

Berzelius (1832b:297)

Early descriptions of schreibersite occurrenceswere confounded by inadequate separation tech-niques, poor analytical methods, and complica-tions arising from both the varied morphologyand the wide range in composition that is typicalof this mineral. C. U. Shepard (1846, 1853) ana-lysed schreibersite and attempted to assign a min-eral name to it. His suggestion, however, wassupported by inadequate data. Credit for namingschreibersite belongs to A. Patera, whose recom-mendation was reported by Haidinger (1847). Thename honors Professor Karl von Schreibers (1775-1852), director of the Imperial Cabinet, Vienna, apioneer worker on meteorites and colleague ofVon Widmanstatten.

The writings of J. Lawrence Smith, a distin-guished chemist of the middle part of the lastcentury (Silliman, 1886; Phillips, 1965), includenumerous examples of early work on meteorites.

Authors Copy

SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

His description of the Tazewell, Tennessee, finestoctahedrite contains a lengthy discussion of hisobservations (Smith, 1855). At this early date herecognized the name schreibersite and applied itto the correct mineral. He reports having identi-fied schreibersite by visual inspection in a numberof iron meteorites from "the Yale College Cabi-net," where it had previously been thought to bepyrite. Smith's low nickel value for Tazewellschreibersite is suspect and his attempt at a for-mula was unsuccessful, but he clearly recognizedthe species and understood which elements wereits major constituents. This work was undoubtedlydone, at least in part, in the chemical laboratoryof the Smithsonian Institution. This laboratorywas organized by Professor Smith in 1854, andSecretary Joseph Henry reported that during theyear "he also made a series of analyses of meteor-ites, among which were fourteen specimens fromthe cabinet of James Smithson, the founder of theInstitution" (Goode, 1897:614).

Confusion in nomenclature, however, persistedin the literature for many years. Von Reichenbach(1861) discussed schreibersite under the descriptiveterm "Ganzeisen" and also used the name lamper-ite. Both of these names are now part of thesynonymy of schreibersite. Rose (1865) introducedthe synonym rhabdite into the literature, a namederived from the Greek word for "rod." Thisterm is still in common use by meteoriticists, butonly when morphological distinctions are of inter-est. Rhabdite is the term used for the small schrei-bersite crystals occurring in the kamacite of ironmeteorites. In polished sections these small schrei-bersite crystals have rhombohedral cross sectionsthat in some cases are elongated, suggesting rodsor needles.

The early literature on meteoritic phosphides,schreibersite and rhabdite, was reviewed in detailby Cohen (1894). Included were a number ofanalyses of schreibersite and rhabdite isolatedfrom various meteorites by Cohen and his co-workers, as well as selected analyses from theliterature. This work established that schreibersiteand rhabdite are morphological variations of thesame mineral species, the rhabdite form having asomewhat higher nickel concentration than thatobserved in large schreibersite inclusions. A simi-lar review was given a few years later by Farrington(1915), using much of the same data Cohen usedbut with several later analyses.

The 19th-century work on meteoritic phos-phides that culminated with Cohen's and Farring-ton's reviews was limited by the capabilities ofclassical descriptive mineralogy and classical ana-lytical chemistry. A plateau had been reached indescriptive iron meteorite studies in general thatbecame an accepted norm that was exceeded byfew workers during the first half of the 20thcentury. Most iron meteorite descriptions pub-lished during this period employed neither con-cepts nor techniques beyond those available toCohen. The descriptive literature grew in volumeand in quality, but new interpretations lagged.Powerful petrographic techniques were being de-veloped and used by metallurgists and ore micro-scopists, but they were not much used by studentsof iron meteorites. The availability of good trans-mitted light microscopes in the hands of petrogra-phers who were accustomed to working with sili-cate minerals resulted in stony meteorites beingpreferred subjects of investigation.

EARLY 20TH-CENTURY STUDIES OF

IRON METEORITES

The emergence of modern physical chemistryand physical metallurgy during the decades adja-cent to the turn of the century had a markedeffect on the future course of iron meteoriteinvestigations. Descriptive studies following the19th-century pattern continued to comprise thebulk of the published literature and, indeed, con-tinue to be published today as essential documen-tation. More modern experimental approaches,however, became significant, resulting in a bodyof knowledge more readily interpretable in termsof origins and developmental histories of meteor-ites than was previously possible.

Actually, the roots of this work go back to amuch earlier period. Pioneering synthetic experi-ments were made by Michael Faraday as early as1820. He prepared metals of meteoritic composi-tions in the course of an investigation to improvesteel (Stodart and Faraday, 1820). At a later pe-riod, Daubree (1868) described synthetic experi-ments related to meteorites. Sorby (1877, 1887),the father of metallography, made astute observa-tions, some of which were based on artificial alloysof meteoritic compositions. Unfortunately, Sorby'spioneering efforts in the metallography of mete-orites were not pursued. The following quotation

Authors Copy

NUMBER 21

from his 1877 paper reflects his perceptive under-standing of the structures he observed:

These facts clearly indicate that the Widmanstatt's Figuring isthe result of such a complete separation of the constituentsand perfect crystallisation as can occur only when the processtakes place slowly and gradually. . . . Difference in the rate ofcooling would serve well to explain the difference in thestructure of some meteoritic iron which do not differ inchemical composition; . . . we are quite at liberty to concludethat they may have been melted . . . .

During the same period the well-known Frenchmineralogist, Meunier (1880), reported the resultsof his synthetic work on meteoritic systems. Apaper by Brezina (1906) is an excellent summariz-ing statement of this early period. It contains anumber of good photomicrographs of iron mete-orites, an unusual feature for papers of the time,and it discusses the development of these struc-tures in terms of exsolution phenomena (solidstate transformation). He also gave a successionfor the formation of the constituents of iron mete-orites, starting with olivine in pallasites: olivine,daubreelite, troilite, graphite, schreibersite, co-henite, chromite, swathing kamacite, kamacitebands, taenite, and plessite. This is a remarkablyinformative listing for the time, demonstratingcareful petrographic observation and keen insight.

One of the perennial problems of meteoriteresearch has been that of drawing the attention ofthose with the knowledge and facilities to makesignificant contributions to the meteorite speci-mens and the important and interesting scientificproblems they represent. Despite Sorby's enthusi-asm for meteorite studies, metallurgists showedlittle interest and had to be attracted into thefield. An example of this process is an invitedpaper read before the Vienna meeting of the Ironand Steel Institute in 1907. Professor FrederickBerwerth's (1907) paper, "Steel and MeteoriticIron," was based on his knowledge of the meteor-ite collection of the Imperial Natural History Mu-seum, Vienna. It was an open invitation to metal-lurgists to become involved in this field of study.

During the early decades of this century, syn-thetic studies were combined with analyses basedon equilibrium phase diagrams. The work of Os-mond and Cartaud (1904), Benedicks (1910), andBelaiew (1924) are examples. Unfortunately, thetools were not available at the time to determinesufficiently accurate equilibrium diagrams. As aresult, many of the conclusions drawn were seri-

ously flawed. These studies, however, pointed theway for future workers.

The introduction of X-ray diffraction analysisinto metallurgical and mineralogical research inthe 1920s had an important influence on meteoritestudies. Young (1926) published an early studyestablishing that kamacite precipitates from tae-nite, not the converse as many had previouslythought. X-ray examination of the Widmanstattenpattern continued for over a decade with deepen-ing understanding and some controversy (Mehland Derge, 1937; Derge and Komnel, 1937; Owen,1938; Smith and Young, 1938, 1939).

Rudolf Vogel's contributions to metallurgicalstudies in meteorites deserves special mention, hispublished papers having covered the period 1925to 1967. They treated a broad range of topicsrelated to meteoritics, eight of them being ofparticular interest in this context (Vogel, 1927,1928, 1932, 1951, 1952, 1957, 1964; Vogel andBaur, 1931). These papers deal with the role ofphosphorus in the development of meteoriticstructures. His 1928 paper discusses schreibersiteand rhabdite precipitation in terms of a hypothet-ical ternary Fe-Ni-P system. His 1932 and 1952papers discuss the same material in terms of exper-imentally derived ternary diagrams.

A watershed publication that represents the cul-mination of this earlier period of iron meteoriteresearch is Perry's (1944) monograph on the met-allography of meteoritic iron. Although manyphotographs of iron meteorites had been pub-lished in the older literature, most of these wereexternal views of complete specimens or macro-photographs of meteorite slices prepared for ex-hibit purposes. Perry's publication was the firstcomprehensive and systematic collection of ironmeteorite photomicrographs. The variety of me-tallic structures in iron meteorites was presentedto the scientific public in a way that has hadsignificant impact up to the present. Sorby hadpointed out the power of the metallographic ap-proach in 1877, but it was not seriously pursueduntil E. P. Henderson drew the publisher andmeteorite collector S. H. Perry into this field inthe 1930s (Henderson and Perry, 1958: ii).

MODERN WORK ON METALLIC PHASES OF

IRON METEORITES

Modern work on the metallic phases of ironmeteorites includes studies that may be grouped

Authors Copy

SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

into the following categories for convenience: (1)interpretations of the Widmanstatten pattern interms of cooling histories of meteoritic parentbodies, (2) compositional studies utilizing traceelements and structure as a means of identifyingparent bodies, (3) schreibersite growth as a key tostructure development and cooling history, (4)descriptive studies and literature reviews. Thesestudies are interrelated and provide essential back-ground for the work to be discussed below. Briefmention of major contributions in these areas willbe given here.

Progress in developing accurate low tempera-ture phase diagrams has played an essential rolein modern iron meteorite studies. The Fe-Ni dia-gram of Goldstein and Ogilvie (1965a) has beenused by meteoriticists for over 10 years (Figure 1).Electron microprobe techniques were employedby these workers to improve upon earlier equilib-rium diagrams (Owen and Sully, 1939; Owen andLui, 1949) and to reconcile the Fe-Ni diagramwith observations on meteorites (Agrell et al.,1963). Buchwald (1966) studied the Fe-Ni-P systemat low temperature using X-ray techniques andproposed a diagram that he used to discuss mete-orite structure development. More extensive workon the Fe-Ni-P system based on longer coolingperiods and electron microprobe measurementswas reported by Doan and Goldstein (1970), andtheir diagram is basic to the interpretation of theexperimental observations discussed below.

The modern literature on the growth of theWidmanstatten pattern is extensive. Massalski andPark (1962) used the Fe-Ni equilibrium diagram tocalculate initial temperatures of kamacite precipi-tation and final equilibrium temperatures, usingareal distribution of kamacite and taenite and alever rule calculation. Wood (1964) and Goldsteinand Ogilvie (1965b) considered the nonequilibriumnature of iron meteorite structures, and combinedphase diagram and kinetic considerations to derivecooling rates and to estimate parent body sizes.Both papers recognize Widmanstatten patterngrowth as diffusion controlled and not subject tobulk equilibrium consideration. Wood (1964) usesthe relationship between Ni buildup at the centersof taenite bands and kamacite band widths tocalculate cooling rates. The Goldstein and Ogilvie(1965b) approach used a detailed analysis of diffu-sion gradients within taenite at taenite-kamacite

interfaces. Both approaches gave comparable cool-ing rates within reasonable limits of error. Anumber of other papers were published duringthis period that contributed significantly to ourknowledge of the structural detail of meteoriticiron (Short and Anderson, 1965; Goldstein, 1965;Reed, 1965b; Axon and Boustead, 1967). Coolingrate and thermal history studies were pursued byGoldstein and co-workers (Short and Goldstein,1967; Goldstein and Short, 1967a, b; Fricker et al.,1970). Goldstein and Doan (1972) discussed theeffect of phosphorus on the formation of theWidmanstatten pattern and reported the first lab-oratory production of an artificial Widmanstattenpattern. They conclude that cooling rate calcula-tions are not seriously affected by phosphorus. Acomprehensive review of Widmanstatten patternstudies has been published by Goldstein and Axon(1973).

Compositional studies of iron meteorites havetraditionally gone hand-in-hand with classification,bulk Ni values and metallographic structure beingthe primary considerations in assigning classifica-tion categories. Modern versions of this type ofclassification have been given by Buchwald andMunck (1965) and Goldstein (1969). The groupingof iron meteorites on the basis of trace elementcontent was introduced by Goldberg et al. (1951)and Lovering et al. (1957). Wasson and co-workersin a series of papers (Wasson, 1974; Scott andWasson, 1975, 1976) have extended this approachto define genetic groups in a large number of ironmeteorites, based on Ni, Ga, Ge, and Ir contentsand structural considerations. Sixteen iron mete-orite groups have been characterized, represent-ing 12 genetically related meteorite groups and anumber of individual meteorite parent bodies.

iRecent work on schreibersite growth in ironmeteorites may be dated from the observations ofHenderson and Perry (1958). Using conventionalchemical analysis, they observed unusually low Niconcentrations in swathing kamacite bordering thelarge low-Ni (12%) schreibersite in the TombigbeeRiver meteorite, a meteorite they also showed tocontain low-Ni (20%) rhabdite. They explainedtheir observations in part on the migration of Feand Ni atoms in the solid state and suggested thatmuch of the phosphide may have separated fromthe liquid. Somewhat later, the first electron mi-croprobe measurements of schreibersite composi-

Authors Copy

NUMBER 21

900 1

800

700

600

500

400

300

-

0

-

-

i i i i

\

\

\

\

11

111w

11

1 1 1 1 1 1

a+y

1

r

i 110 20 30 40

ATOMIC % NICKEL

FIGURE 1. —LOW temperature Fe-Ni phase diagram of Goldstein and Ogilvie (1965).

50

tions were reported by Adler and Dwornik (1961)on schreibersite and individual rhabdite crystalsfrom the Canyon Diablo meteorite.

Goldstein and Ogilvie (1963) reported electronmicroprobe analyses of the composition of various-sized schreibersites in the Canyon Diablo, Breeceand Grant meteorites. They observed a size-com-position correlation, the smaller the schreibersitethe higher the Ni content, and showed the pres-ence of a zone of Ni depletion (swathing zone)even around very small schreibersites (rhabdites).They suggested that most schreibersite grew bysolid state precipitation; although massive schrei-bersites showing no relation to the Widmanstattenpattern probably formed directly from the liquidstate. Size and composition of the schreibersiteswere explained as dependent upon nucleationtemperature and time available for growth of theprecipitate. A growth analysis based on diffusion

kinetics demonstrated that larger schreibersitesnucleated between 700° and 500° C, and that smallschreibersites (rhabdites) formed between 500° and400° C. They also demonstrated that P solubilitywas greater in kamacite than in taenke.

Reed (1965a) emphasized the important role Pplays in the formation of iron meteorite structures.He determined the range of Ni values in schreiber-sites and rhabdites in a large number of meteor-ites, finding Ni values as low as 14% in largeschreibersites and as high as 50% in small taenite-border schreibersites. Ni depletion at kamacite-schreibersite interfaces was also demonstrated.The precipitate size and Ni content relationshipswere confirmed, and Reed pointed out that size-for-size rhabdite grains in the Canyon Diablo me-teorite are 7% higher in Ni than those in theCoahuila meteorite. The author recognized theapproximate nature of chemical analysis values

Authors Copy

8 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

for total P, suggesting that in most cases the actualvalues are higher than the measured values. Hediscussed the sequence of phase formation in me-teorites based on Vogel and Baur's (1931) phasediagram and stressed the need for more preciseinformation on phase equilibria.

In a subsequent paper, Reed (1967) measured Pdistribution in the Mount Edith meteorite, dem-onstrating the practicality of measuring Ni and Pdistribution profiles in kamacite, taenite, andschreibersite. Later Reed (1969) discussed the roleof P in relation to Widmanstatten pattern forma-tion and cooling rate calculations, and reportedmeasurements of P content in the kamacite of 61meteorites.

Doan and Goldstein (1969) discussed the forma-tion of phosphides in the various meteorite classi-fication groups on the basis of their newly deter-mined Fe-Ni-P phase diagram. They estimatedtotal P values from examination of the largestavailable polished sections of a number of meteor-ites, and used them as the basis of a discussion ofschreibersite precipitation reactions in various me-teorite composition ranges. They inferred thermalequilibrium in iron meteorites down to 650° C,with schreibersites forming at higher temperatureswhile rhabdites form from kamacite at low temper-atures when Ni and P diffusion are severely lim-ited.

Comerford (1969) discussed schreibersite andcohenite occurrences in iron meteorites andpointed out that rhabdites, schreibersites that nu-cleated within kamacite, may be more reliable fordiffusion analysis than grain boundary schreiber-sites. Axon and co-workers have discussed schrei-bersite occurrences in a broad range of meteorites(Axon and Faulkner, 1970; Axon and Waine, 1971,1972; Axon and Smith, 1972). The papers of Axonand Waine (1971, 1972) discuss schreibersite mor-phology, mode of occurrence, and relationship toother minerals in great detail. Hornbogen andKreye (1970) have reported detailed metallo-graphic studies on the Coahuila and Gibeon mete-orites and have discussed Ni and P solid statereactions in some detail. Reed (1972) has reportedanalyses of schreibersite in the Oktibbeha Countymeteorite, the highest Ni schreibersite known inmeteorites (65.1%), and De Laeter et :al. (1973)have reported on schreibersite in the Redfieldsmeteorites, the lowest Ni schreibersite on record(7.0%).

Modern reviews that are particularly importantto the investigations at hand have been mentionedpreviously. The paper by Goldstein and Axon(1973) is a comprehensive review of the develop-ment of the Widmanstatten pattern. The recentbook by Wasson (1974) is a broad treatment ofmeteorites and includes an excellent summary ofthe work on chemical groupings of iron meteoriteswith extensive listings of individual iron meteoriteanalyses and classification. The treatise on ironmeteorites by Buchwald (1976) has been of greathelp. Much of the descriptive material on individ-ual meteorites had been available to us in manu-script, and Dr. Buchwald spent many hours teach-ing the senior author the rudiments of iron mete-orite metallography during his two-year stay(1968-1970) at the Division of Meteorites, Smith-sonian Institution.

RELATED MATERIALS

Schreibersite occurrences are not restrictedsolely to iron meteorites. It is an accessory mineralin pallasites, mesosiderites, and enstatite chon-drites, and has been reported in several ordinarychondrites, and a few carbonaceous chondritesand achondrites (Ramdohr, 1973; Powell, 1971;Malissa, 1974). Schreibersite is also known as aproduct of combustion in the coal mines of Com-mentry and Cranzac, France (Palache et al., 1944:125), as a cavity mineral in iron slags (Spencer,1916), and associated with metallic iron in theDisko Island, Greenland, basalts (Pauly, 1969).

Schreibersite has been observed by a number ofworkers in metal particles in lunar soil and rocks.Individual particles have been described in greatdetail, and Ni/Co ratios have been used to identifymetal of meteoritic origin (Goldstein et al., 1970;Goldstein and Yakowitz, 1971; Axon and Gold-stein, 1972; Goldstein et al., 1972). Nickel concen-tration in metal and schreibersite in contact witheach other have also been used to suggest finalequilibrium temperatures and to imply coolinghistories. As lunar studies have continued, it hasbecome apparent that a greater proportion ofschreibersite-containing lunar metal than had pre-viously been thought has been derived from proc-esses other than simple disruption of impactingmeteorites (Axon and Goldstein, 1973; El Goresyet al., 1973; Brown et al., 1973; Carter and Pado-vani, 1973; McKay et al., 1973; Goldstein andAxon, 1973; Gooley et al., 1973).

Authors Copy

NUMBER 21 9

A second phosphide mineral, barringerite(Fe,Ni,Co)2P, has been reported from the Ollaguepallasite (Buseck, 1969). This material is probablya secondary product of some type and not an en-dogenous meteoritic mineral (Buchwald, 1976:105).

The phosphide mineral, schreibersite, is theonly phosphorus mineral that is observed as aminor phase in most iron meteorites, and it is theonly phosphorus-containing mineral that will beconsidered in this paper. There are, however,nine phosphate minerals known in meteorites,and seven of these have been identified as traceconstituents in iron meteorites (Fuchs, 1969; Bild,1974). They are chlorapatite, Ca5(PO4)3Cl; grafton-ite, (Fe,Mn)3(PO4)2; sarcopside, (Fe,Mn)3(PO4)2;whitlockite, Ca3(PO4)2; brianite, Na2MgCa(PO4)2;panethite, Na2Mg2(PO4)2; and farringtonite,Mg3(PO4)2. These minerals are observed in a vari-ety of associations: as inclusions? in graphite-troi-lite-silicate nodules, in simple troilite nodules orin troilite-chromite nodules, as inclusions in or incontact with schreibersite, and associated withother minor minerals or isolated in metal. Thephosphate-phosphide association has been used tocalculate equilibrium oxygen fugacity in iron me-teorites and pallasites (Olsen and Fredriksson,1966; Olsen and Fuchs, 1967). In the iron meteor-ites considered in this paper, phosphate mineralsare either absent or isolated within inclusions thatrepresent a higher temperature stage in the devel-opment of the meteorite structure. No evidencehas been developed in this work to indicate thatthe observed phosphide-metal equilibria have beenaffected by the presence of phosphates.

Experimental

Two types of data are needed in order to inter-pret iron meteorite structural development as aconsequence of cooling through the subsolidusFe-Ni-P system. Accurate bulk compositions arethe first requirement, and good bulk Ni valuesare available from the literature. The problem ofobtaining good bulk P values has not been solvedsatisfactorily, and estimated values were found tobe useful. This aspect of the problem will bediscussed in detail below. The second type of datarequired are P and Ni gradients within kamaciteand schreibersite, and particularly P and Ni valuesat kamacite-schreibersite interfaces. Experimental

procedures for obtaining these values were devel-oped in the electron microprobe laboratory of theDepartment of Mineral Sciences, National Mu-seum of Natural History, Smithsonian Institution.Reed (1967, 1969) had reported traverse data onthe Mount Edith meteorite and had demonstratedthe practicality of measuring the low P levels inkamacite with the microprobe. Doan and Gold-stein's (1970) experience in their phase diagramstudies was also encouraging. Ideally, one re-quired step-by-step traverses over many meteoritestructures, each traverse consisting of a large num-ber of individual analyses for P and Ni. Ni concen-trations were in a range that presented no unusualanalytical problems, but the expected low levels ofP required special care.

An A.R.L. EMX electron microprobe (AppliedResearch Laboratory, Sunland, California) wasused in this work. Simultaneous measurementswere made for Ni Ka (LiF crystal) and P Ka (ADPcrystal) radiation at an X-ray takeoff angle of52.5°. The instrument was operated at an acceler-ating voltage of 20 KV, with an approximately 0.1fia sample current on 100% Fe and a beam size ofapproximately 1 /xm. Average count rates mea-sured on 50% Ni and calculated to 100% Ni were30,000 cts/sec, with a peak to background ratio of230. Count rates for 100% P measured on schrei-bersite were 25,000 cts/sec with a peak to back-ground ratio of 915. Standard statistical assump-tions suggest that for the 20 second countingperiods employed, a detectability limit for P of150 ppm for a single determination would beachieved, within a satisfactory range for the prob-lem at hand (Ziebold, 1967). Neighboring P deter-minations within a given traverse generally agreedwithin ±50 ppm.

The standards used were 100% Fe, 100% Ni, aseries of Fe-Ni alloys of known composition, and afragment of a large schreibersite from the CanyonDiablo meteorite. The 100% Fe was used for deter-mining both Ni and P background counts. TheFe-Ni standards were prepared by R. E. Ogilvieand obtained from J. I. Goldstein. They wereused to derive an empirical curve that was incor-porated into a data reduction program to correctNi determinations. The schreibersite standardcontained 12.9% Ni, 0.26% Co, and.was assumedto contain 15.5% P. A linear relationship betweenP counts in schreibersite and concentration in anunknown was assumed. This was thought to cause

Authors Copy

10 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

no serious error in low level P measurements, butit does introduce errors in P determination inschreibersite. In this study, P values for schreiber-site were only used to identify the mineral, andfor this purpose corrections seemed unnecessary.

Sections were prepared for study by standardmetallographic procedures. Pieces of meteoritewere mounted in one-inch diameter Bakelite,ground on a graded series of sandpapers, andpolished with 3 /xm diamond paste and finallyLinde B alumina compound. The sections werethen lightly etched with 0.5% nital (0.5 ml nitricacid in 100 ml of 95% ethanol). Light etching wasessential to reveal structures that were examinedand to permit their being relocated and analyzedin the microprobe. It would have been impossibleto obtain the mass of data required using unetchedsections. While deep etching is known to causeerrors in microprobe measurements at phase inter-faces, the light etching used here did not seem tocause a problem. A schreibersite-kamacite-taenitestructure in the Carleton meteorite was selected totest this point. As nearly identical traverses aspossible, considering that the section had to beremoved from the microprobe, were made beforeand after etching. The results were identical withinreasonable experimental error. Areas selected formicroprobe analysis were photographed in detail,permitting the recording of the exact location ofmicroprobe traverses.

The electron microprobe data was accumulatedusing a step-scan procedure. Ni and P countswere measured simultaneously and recorded onpunch cards along with step length. The datawere computer processed, resulting in a secondset of cards containing calculated P values, empir-ically corrected Ni values, and step length. Thesecards in turn were used to obtain computer gen-erated plots. Photographs of the data plots asreceived from the computer are given in Figures 2and 3. The only additions to these diagrams arethe vertical lines that have been added to helpidentify the structural elements. Figure 2 is atraverse across a structure in the Ballinger, Texas,meteorite, catalog number USNM 824. The label-ing indicates that this is the fifth traverse madeover structures in that meteorite, the data havingbeen taken on 8 September 1972. Two symbolsare used to indicate P concentration, the ordinatescale being zero to 0.7% P for low values and zeroto 70% P for high values. The traverse started in

kamacite and crossed a very thin, elongated phos-phide (4 /mm wide) at the edge of a taenite lamella.It passed into kamacite and after traveling 275 /Amentered a rhombohedral phosphide (rhabdite) 30/u,m across and then passed into kamacite again.The two unusually high P values in the kamaciteare undoubtedly due to unseen phosphides thatwere included in the excitation volume of themicroprobe beam. Figure 3 is a similar plot forthe Lexington County Meteorite. In this case onlya portion of the complete traverse is reproduced(700 out of 1600 fim). The traverse started inkamacite, entered a phosphide at a taenite border,passed into kamacite, a second phosphide, and asecond taenite lamella. Interface concentrationvalues and concentration gradient shapes andlengths were read from these plots.

Data obtained in this way are subject to severalsources of error beyond those that apply to anindividual point analysis in a homogeneous mate-rial. Stability of the microprobe over the longperiods required to obtain lengthy traverses was aproblem. Due to the survey nature of this investi-gation, compromises were made in favor of largenumbers of traverses and for lengthy traverses.The microprobe beam current was monitored andkept within narrow limits, but this does not neces-sarily prevent all instrumental drift. Data wererejected when drift became an obvious problem,but normally measurements on standards or re-peated measurements on the same structures gaveidentical results within reasonable limits of error(± 0.005% P, ±0.1% Ni).

The schreibersite-kamacite interface measure-ments are subject to errors due to both finitebeam size and to interface geometry. The finitebeam produces measured interface profiles thatare less sharp than the actual profile. It is unlikelythat this is a serious error in determining theinterface value of P in kamacite, as this is onlyread to the nearest 0.01%. The interface value ofNi in kamacite is probably increased slightly bythe proximity of high Ni in the adjoining schrei-bersite. More serious errors may result from inter-face geometry, as there is no reason to assumethat the interfaces measured were perpendicularto the surface of the section. The small size ofmost of the structures measured, combined withtheir large number, made it impractical to attemptmeasuring slopes and applying corrections. Repro-ducibility of measurements on the same structure

Authors Copy

NUMBER 21 11

BRLLINGER 824 05 8 SEPT 72

ELEMENTrn = P ( 0 - . 7 )© = P ( 0 - 7 0 )* = NI ( 0 - 7 0 )

B 5 a

me a

50 100 150 200 250 300 350 400 450 500 550

DISTflNCE (MICRONS)

FIGURE 2. —Computer plotted Ni-P traverse across structure in Ballinger meteorite.

and on similar structures within the same meteor-ite indicated that this is not too serious a problemfor the study at hand.

Results

Eight coarse-structured iron meteorites coveringa range of bulk Ni and bulk P contents wereselected for Ni and P measurements at kamacite-schreibersite interfaces (Table 1). Four structuraland five chemical classification categories are rep-resented in the group. The Bellsbank meteoritecontains very low total Ni, while Balfour Downs isone of the highest Ni members of Group IA.Coahuila contains only about 0.3 wt. %P, whileBellsbank contains exceptionally high P. The othermeteorites in the group range between these limits

in Ni and P contents. The data in Table 1 weretaken from Wasson (1974) and Buchwald (1976).Buchwald's values combine, where appropriate,chemical analytical values with the results of plan-ometric analyses of large surfaces. Contributionsof Ni and P due to large schreibersite inclusionsare included in this approach, resulting in esti-mates more representative of the meteorite as awhole than uncorrected chemical values. Wasson'sNi values are chemical values determined on smallsamples taken for trace element analysis. A com-plete review of the literature of each of thesemeteorites and a discussion of their detailed met-allography has been given by Buchwald (1976).

Microscopic examination of metallographic sec-tions of these eight meteorites led to the selectionof representative schreibersite-containing struc-

Authors Copy

12 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

LEXINGTON COUNTY 3334 03 30 SEPT 72

ELEMENTm = P ( 0 - . 7 )O = P (0-70)A = NI t 0 -70 )

HO O a

100 ISO 200 250 300 360 400 450 600 550

DISTRNCE (MICRONS)

FIGURE 3. —Computer plotted Ni-P traverse across structure in Lexington County meteorite.

tures for detailed study by the electron microprobestep-scan procedure described above. Several mi-croprobe traverses were made for each meteorite,and the data obtained on Ni and P values atkamacite-schreibersite interfaces and on Ni and Pgradients in kamacite are summarized below. Aseparate table accompanies the description of trav-erses for a given meteorite, the format used beingthe same in each case (Tables 2-9). The first twocolumns give the specimen number, which in-cludes the USNM catalog number, and thetraverse number. Column three contains two typesof information. The first entry for each traverse isenclosed in square brackets and gives the sequenceof phases traversed and the length of the traverse.The abbreviations used here are Ph for the various

morphologies of schreibersite, a for kamacite andy for taenite. The following entries for a giventraverse identify the specific schreibersites mea-sured and give the length of schreibersite tra-versed. The next column, headed %NiSch, givesthe weight percent Ni in schreibersite either at akamacite-schreibersite interface or within a largeschreibersite. Where an interface measurementhas been made, the schreibersite Ni value will befollowed in the next two columns by the weightpercent Ni in kamacite at the interface (%Niot) andthe weight percent P at the interface (%Pa). Thenext two sets of two columns each indicate thelength of observed Ni and P gradients away fromthe interface and the Ni and P values at which agradient was no longer distinguishable. Factors

Authors Copy

NUMBER 21 13

TABLE 1. — Composition and classification of selected meteorites

Meteorite

Coahuila

Bellsbank

BaTMnger

Santa Luzia

Lexington County . . . .

Bahjol

Goose Lake

Balfour Downs

•Buchwald (1976)

**Wasson (1974)

Weight Percent%P*

ChemicalClassification**

StructuralClassification*

0.3

2

0.4

0.9

0.3

0.4

0.3

5.6

5.3

6.5

6.6

-7.0

7.7

8.3

8.4

5.49

4.13

6.19

6.3

6.69

7.95

8.00

8.39

IIA

IRANOM

IA-AN

I IB

IA

IA-AN

IA-AN

IA

H

H

Og

Ogg

0g

0g

0m

Og

such as impingement and termination of a partic-ular traverse influence these final observed Niand P values. In the final three columns the Niand P values from columns three through six havebeen converted to atomic percent for use later.

COAHUILA

The Coahuila, Mexico, meteorite is an exampleof an ordinary hexahedrite (IIA), similar in com-position and structure to a number of other hexa-hedrites. It consists of single crystal kamacite con-taining profuse small schreibersites generally ofrhabdite morphology, and occasional largerschreibersites. The larger schreibersites sometimesborder troilite or troilite-daubreelite inclusions,and in rare instances are associated with cohenite.

Section 3298 is dominated by six troilite-dau-breelite inclusions, ranging in size from 5 to 0.3mm in diameter. Half to three-quarters of thelength of the borders of these sulfides are rimmedwith schreibersite. One small sulfide is borderedfor most of its circumference with schreibersite,the remaining border being rimmed with cohenite.Kamacite in the vicinity of these inclusions containsmany subgrain boundaries and is free of schreiber-site. Microrhabdites measuring less than a fewmicrons in maximum length and generally consid-erably smaller are present in profusion away fromthese inclusions. Somewhat larger rhabdites arepresent in several areas aligned along Neumannbands. Several lamellar schreibersites, 5 to 10 /amwide and extending to lengths that are a majorpart of a millimeter or more, are also present. Asummary of the traverses of this section is given

below and specific measurements are listed inTable 2.

3298, traverse 1: crossed a 150 /xm wide schreibersite borderinga 2 x 2 mm troilite-daubreelite inclusion and then extendedinto the surrounding kamacite for nearly 1 mm.

3298, traverse 2: crossed a 40 yum wide schreibersite bordering a0.5 x 0.5 mm troilite-daubreelite inclusion and then ex-tended 350 (im into the surrounding kamacite.

3298, traverse 3: crossed a 100 /u.m wide schreibersite borderingthe 5 x 5 mm troilite-daubreelite inclusion and then ex-tended for 350 /xm into surrounding kamacite. This traverseis reproduced at the upper left in Figure 4.

Section 1641(1) primarily contains regions ofkamacite with evenly distributed rhabdite, the in-dividual rhabdites averaging 200 fim2 in cross-sectional area. Also present in areas of clear kam-acite are several large schreibersite inclusions andschreibersite-cohenite inclusions. The largestschreibersite inclusion has an area of 0.25 mm2. A0.1 x 0.6 mm schreibersite is bordered by cohen-ite, and an adjacent, slightly smaller schreibersiteis partially surrounded by cohenite. One cohenitearea, 0.2 x 0.3 mm, contains six small schreiber-sites. A 0.4 x 1.4 mm cohenite area contains anumber of small schreibersites and surrounds acentral schreibersite measuring 0.04 x 0.6 mm.An 80 X 150 fim daubreelite is surrounded byschreibersite. Two lamellar schreibersites, eachapproximately 0.02 X 1.0 mm, are present, andone of these is partially bordered with cohenite.

1641(1), traverse 4: crossed a large skeletal schreibersite enclos-ing a kamacite area. The traverse first crossed 450 fjun ofkamacite, then 120 /urn of schreibersite, 70 jxm of includedkamacite, a second 70 (im schreibersite crossing, and termi-nated after passing through 360 /tm of kamacite. The schrei-bersite was of uniform composition, and the measurements

Authors Copy

14 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 2. —Coahuila schreibersite-kamacite interface measurements

SpecimenNo.

3298

3298

3298

1641(1)

1641(1)

1641

1641

1641

1641

TraverseNo.

1

2

3

4

5

6

7

8

9

Structure traversed & schreibersites measured

[Ph border of troilite-daubreelite inciusion-a,1.1 mm]

Schreibersite, 150 pm wide

[Ph border of troilite-daubreelite inclusion-a,400 ym]

Schreibersite, 40 ym wide

[Ph border of troilite-daubreelite inclusion-a,450 ym]

Schreibersite, 100 um wide

[a-Ph-a-Ph-a, 1 mm]

Schreibersite 260x500 ym with 60x170 umincluded kamacite

120 ym traverse

70 ym traverse

[a-Ph-a, 350 um]

Rhabdite, 25 ym traverse

Enter

Exit

[a-Ph-a, 750 um]

Rhabdite (45x75 um), 45 um traverse

[a-Ph-a, 800 um]

Rhabdite (50x70 um), 70 ym traverse

Exit

[o-Ph-a-Ph-a, 850 ym]

Rhabdite (50x70 ym), 50 um traverse

Rhabdite (25x30 ym), 25 ym traverse

[a-Ph-a, 200 ym]

Rhabdite (20x40 um), 25 ym traverse

Weight

*NiSch

23.5

25.0

25.0

23.5

23.5

34.5

34.5

27.5

27.0

27.0

29.0

33.0

Percent3SNi %P

a a

3.4

3.6

3.7

3.53.5

4.94.9

3.7

4.0

4.13.7

4.5

0.07

0.07

0.08

0.08

0.08

0.06

0.06

0.07

0.07

0.07

0.06

0.05

Ni GradientLength(ym) Wt.%Ni

600

250

250

200250

5050

175

100

150200

50

5.5

5.3

5.3

5.35.3

5.35.3

5.3

5.3

5.35.3

5.3

P GradientLength(um) Wt.%P

800

350

300

50200

Flat

100

100

Flat

Flat100

Flat

0.16

0.15

0.15

0.10

0.10

0.06

0.07

0.08

0.07

0.07

0.08

0.05

Atomic

*NiSch

20.1

21.4

21.4

20.1

20.1

29.7

29.7

23.6

23.1

23.1

24.9

28.4

Percent

%N1 %Pa a

3.2

3.4

3.5

3.33.3

4.74.7

3.5

3.8

3.93.5

4.3

0.13

0.13

0.14

0.14

0.14

0.11

0.11

0.13

0.13

0.13

0.11

0.09

in Table 2 are for the initial and final schreibersite-kamaciteinterfaces.

1641(1), traverse 5: crossed 100 fim of kamacite, a 25 fim widerhabdite and terminated after another 200 fim of kamacite.This traverse is reproduced at the lower left in Figure 4.

Section 1641 is a neighboring specimen to1641(1) and is similar in metallography. Evenlydistributed rhabdites averaging about 200 jam2 inarea are the primary feature. Random largerschreibersites, schreibersite-daubreelite, andschreibersite-cohenite inclusions similar to thosedescribed above are present, although they areneither as numerous nor as large as the largestmentioned above. Two unusually large rhabditesand two smaller ones were selected for measure-ment.

1641, traverse 6: crossed 400 fim of kamacite, a 45 fim widerhabdite, and 300 fim of kamacite. Values reported in Table2 are an average of the two similar interface measurements.This traverse is reproduced upper right in Figure 4.

1641, traverse 7: crossed 230 fim of kamacite, entered therhabdite from an area of disturbed kamacite, traversed 70

fim of rhabdite and then 0.5 mm of kamacite. Only themeasurements on leaving the rhabdite were usable.

1641, traverse 8: crossed 370 fim of kamacite, 50 fim of rhabdite,25 fim of kamacite, 25 fim of rhabdite, and 350 fim ofkamacite. The large rhabdite traversed here is the same onetraversed in number 7 but from a near perpendicular direc-tion. Values reported in Table 2 are averages of two similarinterface measurements.

1641, traverse 9: crossed a smaller rhabdite approximately 1mm from traverses 6 through 8. 90 fim of kamacite werecrossed, followed by 25 fim of rhabdite and 80 fim ofkamacite. Average interface values are listed in Table 2.This traverse is reproduced at the lower right in Figure 4.

Ni and P concentration-distance profiles typicalof the Coahuila data outlined above are given inFigure 4. These diagrams were prepared fromtracings of the computer plotted step-scan data.Ni concentrations are indicated by the solid lines,and P concentrations multiplied by 100 by dashedlines. The Ni profiles are normally readily repro-duced from the data charts without ambiguity.The low level P data contains more scatter; draw-ing the appropriate line, and particularly selecting

Authors Copy

-NUMBER 21 15

5O

45-

40-

omUJ0 .

£ 3OOUJS

25

20-

15-

, 0

25

100 (Jm

0.08

3.7

P x 100

Ni

3296

28

0.07

3.7

1641

400MICRONS

N4.9

300MICRONS

0.05,,4.5

FIGURE 4. —Ni and P profiles at kamacite-schreibersite interfaces in Coahuila: upper left, traverse3; upper right, 6; lower left, 5; lower right, 9.

Authors Copy

16 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

the kamacite interface P level, is more subjective.A smoothing procedure relying on eyeball exami-nation was used. Ni values for schreibersite andNi and P concentrations in kamacite at the kama-cite-schreibersite interface are indicated at an ap-propriate place on the diagrams. The bulk Nicontent of the meteorite is also given and indicatedby a short dashed line.

The four profiles (Figure 4) illustrate the ob-served range of Ni in Coahuila schreibersite andthe shape and level of the Ni and P gradients inthe surrounding kamacite. They are given fromupper left to lower right in order of decreasing Pconcentration at the kamacite-schreibersite inter-face . Three of the profiles are of large rhabdites(traverses 6, 5, 9, Table 2), while the one at theupper left is of a 100 /xm wide schreibersite border-ing a large troilite-daubreelite inclusion (traverse3, Table 2). The high P interface value and thehigh level in the surrounding kamacite are ofparticular interest. The three rhabdites appear toincrease in Ni, both with decreasing cross-sectionalarea and with increasing interface Ni in kamacitevalue. The interface P values do not seem tocorrelate with either Ni in schreibersite or withinterface Ni in kamacite values. Ni and P gradientsextend for several hundred microns around thelarger schreibersites but appear to be much morerestricted in extent around the smaller ones. The

33% Ni schreibersite has an essentially flat P con-centration at 0.05%. For these four cases, thelength and steepness of the P gradient decreaseswith decreasing P at the interface.

BELLSBANK

The Bellsbank, South Africa, meteorite is aphorphorus-rich hexahedrite. Massive and largeskeletal schreibersite crystals, with individual facesover a centimeter in length and areas larger than2 cm2, dominate polished surfaces and appear tobe distributed throughout this meteorite (photo-graph in Henderson, 1965). The kamacite matrixcontains a network of lamellar schreibersite. Theserarely intersecting planar crystals may exceed acentimeter in length, but are normally only 5 to 10/u,m thick. Occasional large rhabdites and subgrainboundary schreibersites are present. Microrhab-dites are present in profusion in kamacite areasaway from the other forms of schreibersite. TheBellsbank meteorite has experienced severe terres-trial weathering. Many of the lamellar schreiber-sites have been replaced with corrosion products,and corrosion along kamacite-schreibersite inter-faces is common.

Section 2162 contains parts of two massiveschreibersites surrounded by schreibersite-freekamacite areas. Two lengthy subgrain boundaries

TABLE 3. —Bellsbank schreibersite-kamacite interface measurements

SpecimenNo.

2162

2162

2162

2162

2162

2162

TraverseNo.

1

2

3

4

5

6

Structure traversed & schreibersites measured

[a-Ph, 3.6 mm]

Massive schreibersite (7x2 mm), 1 mm traverse

[a-Ph-a, 900 um]

Large rhabdite along sub-grain boundary,40x50 um, diagonally across body.

[a-Ph-a, 200 um]

Same rhabdite as traverse 2, parallel to 50 pmdirection

Enter

Exit

[a-Ph-a, 300 um]

Rhabdite along sub-grain boundary, 15x120 um

[a-Ph-a, 300 um]

Rhabdite along sub-grain boundary, 15x120 um,repeat of traverse 2162-4

[a-Ph-a, 130 ym]

Large rhabdite Isolated in micro-rhabditearea of a, 30x60 pm

Weight

*NiSch

12.4

22.0

22.0

22.0

22.6

22.8

18.6

Percent

SNia

1.6

3.2

3.3

2.7

2.6

2.6

2.8

XPa

0.09

0.09

0.09

0.05

0.05

0.05

0.09

Ni GradientLength

(um)

900

20

40

40

80

90

40

Wt.SSNI

3.9

4.5

4.2

4.2

4.3

4.3

4.2

P GradientLength

(um)

1000

50

Flat60

100

100

30

Wt.JSP

0.20

0.18

0.09

0.12

0.20

0.18

0.20

Atomic

*N1Sch

10.6

18.8

18.8

18.8

19.3

19.5

15.9

Percent

XN1a

1.5

3.1

3.1

2.5

2.5

2.5

2.7

%Pa

0.16

0.16

0.16

0.09

0.09

0.09

0.16

Authors Copy

NUMBER 21 17

MICRO-RHABOITES

400MICRONS

Px 100\

400MICRONS

FIGURE 5. —Ni and P profiles at kamacite-schreibersite interfaces in Bellsbank: top, traverse 1;lower left, 6; lower middle, 3; lower right, 4 and 5. (Dashed line indicates bulk Ni value.)

Authors Copy

18 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

cross the kamacite and are sites of small lamellar,irregular-shaped, and rhabdite-shaped schreiber-sites. The kamacite matrix contains profuse micro-rhabdites. Oxidation products are present at sitespreviously occupied by lamellar schreibersite, andalong some Neumann bands and possibly cleavageplanes.

2162, traverse 1: crossed 2.6 mm of kamacite and entered alarge schreibersite and continued for 1 mm. The traversewas perpendicular to a straight kamacite-schreibersite inter-face, and the Ni content of the schreibersite was uniform forthe length of the measurement (reproduced at top of Figure5). Numerical data for this and succeeding Bellsbank trav-erses are given in Table 3.

2762, traverse 2: crossed 650 fim of kamacite, a rhabdite at asubgrain boundary and continued into surrounding kamacitefor 200 /urn. Average values for the two interface values aregiven in Table 3.

2162, traverse 3: crossed 70 fim of kamacite, the same rhabditemeasured above, and passed into 70 fim of kamacite.Traverse 2 crossed diagonally, while this one passed throughthe middle parallel to the 50 im edge. Two differentinterface values for Ni and P in kamacite were obtained,perhaps influenced by the presence of the subgrain boundary(Table 3 and lower middle of Figure 5).

2162, traverse 4: crossed 170 fim of kamacite, the narrowdirection of an elongated rhabdite situated along a subgrainboundary, and 100 fim of kamacite (lower right in Figure 5).

2162, traverse 5: a remeasurement of traverse 4.2762, traverse 6: crossed a large rhabdite surrounded by a

region of clear kamacite within a microrhabdite area (lowerleft in Figure 5).

An attempt was made to determine Ni in othersmall schreibersites in Bellsbank. A number ofmeasurements were made on schreibersites rang-ing down to a minimum of a few microns inwidth. Twelve measurements were obtained wherethe P value indicated that the microprobe beamhad been completely within a schreibersite. TheNi values ranged from 22% to a maximum of32%. Interface measurements were not attemptedon these small bodies.

The four profiles in Figure 5 represent thetraverses described above and illustrate the shapesand levels of the Ni and P gradients in the sur-rounding kamacite. The profile at the top is theinterface portion of traverse 1. The low value forNi in the schreibersite combines with an unusuallylow interface value for Ni and a high interfacevalue for P in kamacite. The rhabdite profile atthe lower left has the same P interface value, buthigher Ni in both kamacite and schreibersite. TheP gradient in this case is particularly severe. Therhabdite at the lower right has similar Ni values to

the one on the left, but the P interface value isconsiderably lower. The middle profile has uni-form Ni in the schreibersite but has two differentsets of P and Ni kamacite interface values, theonly such observation made in this study.

BALLINGER

The Ballinger, Texas, meteorite is a low-Nicoarse octahedrite. Figure 6 is a photograph ofthe slice studied, USNM 824. The area marked PSshows where material was taken for section prepa-ration. Section 824 is from the back of this surfaceand contains a large schreibersite similar in sizeand shape to the one at the lower right in Figure6. Section 824(1) is the outlined surface after repol-ishing. The overall structure consists of largekamacite grains and three areas of heiroglyphicschreibersite. A Widmanstatten pattern is sug-gested in some areas of the structure, but it iscertainly not well developed. The large schreiber-site on the left is completely surrounded by rimsof partially decomposed cohenite. The centergroup of schreibersite crystals is partly borderedby partially decomposed cohenite, while the schrei-bersite on the right is completely free of cohenite.Areas of this surface containing the best Widman-statten pattern development are well away fromthe large schreibersite inclusions. It is interestingto note that the lower left-hand edge of this slicecontains a rim of kamacite heat altered to a2, aremnant from ablation heating. This structure hasnot been previously observed in the Ballingermeteorite.

Section 824 contains a large schreibersite 0.8mm long and averaging 1 to 2 mm in width,surrounded by a narrow area of schreibersite-freekamacite. The kamacite matrix contains a profu-sion of both Neumann bands and rhabdites. Neu-mann band bending, particularly at the interfaceswith the large schreibersite, suggests mild mechan-ical distortion in Ballinger. The kamacite containsnumerous grain boundaries that are the sites ofgrain boundary schreibersites. Terrestrial oxida-tion has penetrated areas of this section.

824, traverse 1: crossed three areas of the large schreibersite.500 fim of kamacite were crossed, followed by 350 fim ofschreibersite, 800 fim of kamacite, 400 fim of schreibersite,450 fim of kamacite, 670 fim of schreibersite, and finally 500fim of kamacite. The three schreibersite areas traversed areessentially free of Ni gradients and they all have approxi-

Authors Copy

NUMBER 21 19

PS

1 cmI 1

FIGURE 6. —Macrostructure of a slice of Ballinger meteorite (large dark inclusions are schreibersite,the one on the left enclosed in partially decomposed cohenite; area marked PS was used formicrostructure examination [824(1)]; and edge marked a2 contains a rim of ablation recrystallizedkamacite).

mately the same Ni value. Numerical data for this andsucceeding Ballinger traverses are given in Table 4.

824, traverse 2: crossed two rhabdites along a grain boundary 2mm away from the massive schreibersite. 400 fim of kamacitewere crossed, followed by two 40 fim rhabdites separated by25 fim of kamacite, and finally 300 fim of kamacite. Becauseof impingement effects, the interface values in Table 4 arean average of the two similar external measurements.

824, traverse 3: crossed two areas of a 1 x 0.4 mm skeletalschreibersite 3.5 mm from the massive schreibersite. Thevalues given in Table 4 are averages of the two externalinterface values.

Section 824(1) is from a typical Widmanstattenarea of the Ballinger meteorite, the area outlined

in Figure 6. The kamacite matrix contains Neu-mann bands and rhabdites in a wide range ofsizes. Kamacite lamellae are separated by grainboundaries that are alternately sites of grainboundary schreibersites and taenite-plessite bands.

824(1), traverse 4: crossed 230 fim of kamacite and entered asmall schreibersite embedded in a taenite border, passedthrough a narrow band of taenite, and then 220 fim ofkamacite.

824(1), traverse 5: crossed 150 fim of kamacite, a schreibersiteembedded in taenite, 20 fim of taenite, 270 fim of kamacite,a large rhabdite, and finally 120 /AITI of kamacite (see Figure2 in experimental section).

Authors Copy

20 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 4. —Ballinger schreibersite-kamacite interface measurements

SpecimenNo.

824

824

824

824(1)

824(1)

824(1)

824(1)

824(1)

TraverseNo.

1

2

3

4

5

6

7

8

Structure traversed & schreibersites measured

[a-Ph-a-Ph-a-Ph-a, 3.7 mm]

Massive schreibersite, 350 ym

Enter

Exit

Massive schreibersite, 400 ym

Enter

Exit

Massive schreibersite, 670 ym

Enter

Exit

[a-Ph-o-Ph-a, 850 ym]

Two large rhabdites, 40x40 um each,separated by 25 um. Values recordedare averages of similar values.

[a-Ph-a-Ph-o, 700 um]

Skeletal schreibersite 1x0.4 mm. Valuesrecorded are average of similar valuesfrom the two exterior Interfaces.

[a-Ph-Y-o» 480 um]

Rhabdite embedded in taenite border,15x20 um

[a-Ph-Y-a-Ph-a, 650 ym]

Rhabdite embedded In taenite border, 5 umwide

Large rhabdite (35x25 um), 250 ym into afrom above

[ct-Ph-Y-a, 300 um]

Rhabdite embedded in taenite border,10x15 um

[a-Ph-a-Ph-a-Ph-a-Ph-a, 1.4 mm]

Grain boundary schreibersite in sequencewith taenite, 20x15 ym

Rhabdite 50 ym from above, 40x50 ym

Rhabdite 400 ym from above, 40x40 ym

Rhabdite 600 ym from above, 30x50 ym

[a-Ph-a, 1.0 mm]

Grain boundary schreibersite in sequencewith taenite, 15x140 ym

Weight

*N1 S c h

19.5

19.5

19.5

19.3

19.7

19.5

34.5

35.8

49.8

49.5

41.5

51.0

47.0

44.5

40.0

40.0

48.8

Percent

a

2.93.0

2.62.6

...

2.6

4.2

4.8

6.0

6.0

5.2

6.3

5.75.75.05.0

6.0

%Pa

0.06

0.06

0.06

0.06

0.06

0.06

0.04

0.05

0.03

0.03

0.05

0.03

0.03

0.04

0.05

0.05

0.03

N1 GradientLength(ym)

500300

200150

150500

200

150

50

50

50

50

50100100100

100

Wt.%N1

5.24.3

4.34.0

4.05.0

6.0

6.8

7.2

7.2

7.2

7.2

7.27.27.27.2

7.0

P GradientLength(um)

150200

200150

100150

100

200

50

50

50

50

50100200200

300

Wt.SP

0.10

0.09

0.08

0.08

0.08

0.08

0.09

0.08

0.05

0.05

0.07

0.06

0.06

0.07

0.09

0.09

0.08

Atomic

*NiSch

16.6

16.6

16.6

16.5

16.8

16.6

29.7

30.8

43.1

42.9

35.8

44.2

40.7

38.4

34.5

34.5

42.3

Percent

XN1a

2.82.9

2.52.5

2.52.5

4.0

4.6

5.7

5.7

5.0

6.0

5.45.44.84.8

5.7

XPa

0.11

o.n

o.no.n

o.n0.11

0.07

0.09

0.05

0.05

0.09

0.05

0.05

0.07

0.09

0.09

0.05

824(1), traverse 6: crossed 80 fj.m of kamacite, a taenite borderschreibersite, 15 fim of taenite, and 150 fim of kamacite.

824(1), traverse 7: crossed 140 /u,m of kamacite, a grain boundaryschreibersite in sequence with taenite, 50 fim of kamacite, alarge rhabdite, 300 (im of kamacite, a large rhabdite, 620jum of kamacite, a large rhabdite, and finally 80 /um ofkamacite.

824(1), traverse 8: crossed 90 jxm of kamacite, a grain boundaryschreibersite in sequence with taenite, and 1 mm of kamacite.

The massive schreibersite in Ballinger containshigher Ni than similar sized schreibersite in Bells-bank, and the Ni in kamacite interface values arehigher while the P interface values are lower.

Large rhabdites in Ballinger contain more Ni thansimilar ones in Coahuila or Bellsbank, and theirinterface values are generally higher in Ni andlower in P. Grain boundary schreibersite tends tocontain more Ni than kamacite matrix rnabdites,and taenite border schreibersite contains the high-est Ni values observed. As the Ni values in theschreibersite go up, interface values in kamacitetend to go up for Ni and down for P. Ni and Pgradients extend over relatively large distancesaround the massive schreibersite and over muchshorter distances in the higher Ni forms of schrei-bersite .

Authors Copy

NUMBER 21 21

SANTA LUZIA

The Santa Luzia, Brazil, meteorite is a P-richcoarse octahedrite, a chemical Group IIB meteor-ite. The structure of a typical section is illustratedin Figure 7. Large, hieroglyphic schreibersites withbroad areas of swathing kamacite dominate thestructure. Two such schreibersite areas are shownin the figure, the central one bordering a largetroilite. The second schreibersite area could alsobe associated with a lens-shaped troilite nucleusthat lies outside the plane of this section. Theareas of structure between the troilite-schreiber-site-swathing kamacite areas contain a coarse Wid-manstatten pattern, particularly well developed inthe top of Figure 7. These Widmanstatten areasare reminiscent of the schreibersite-free areas ofthe Ballinger meteroite. Weathering has pene-trated deeply into Santa Luzia, particularly alongthe major grain boundaries separating the swath-ing kamacite areas from the areas of Widmanstat-ten pattern.

Figure 8 is a sketch of the section of the photo-graph (Figure 7), indicating how this slice wassampled for detailed metallographic and electronmicroprobe analysis. Four metallographic sectionswere prepared from the areas indicated by dashedlines in the sketch. Section 1618 is from an area ofWidmanstatten pattern, 1618P from the schreiber-site-swathing kamacite area, and 1618PS and1618PS(1) from the sulfide-schreibersite-swathingkamacite area. The numbered arrows in the sketchindicate the location of the traverses describedbelow. Lengthy traverses were made in each ofthe two hieroglyphic phosphide areas, with shortones being made in the Widmanstatten patternarea.

Section 1618P contains massive schreibersite andswathing kamacite (Figure 8). Clear kamacite sur-rounds the large schreibersite, grading into kama-cite containing a profusion of microrhabdites.Neumann bands are most obvious in the transitionareas between clear kamacite and microrhabditeareas. Occasional subgrain boundaries are presentwithin the kamacite, and small schreibersites ofvarious morphologies are associated with them.

1618P, traverse 1: crossed 1.3 mm of massive schreibersite, 2.0mm of kamacite, 0.7 mm of massive schreibersite, 0.5 mm ofkamacite, 0.2 mm of massive schreibersite, and 2.6 mm ofkamacite. Numerical data for this and succeeding Santa

Luzia traverses are given in Table 5.1618P, traverse 2: crossed 1.0 mm of kamacite, 0.4 mm of

massive schreibersite, 1.9 mm of kamacite, 0.8 mm of massiveschreibersite, 0.9 mm of kamacite, and 0.2 mm of massiveschreibersite.

1618P, traverse 3: crossed 0.2 mm of massive schreibersite, 1.0mm of kamacite, and 0.2 mm of massive schreibersite.

Section 1618PS contains troilite, massive schrei-bersite and areas of swathing kamacite (Figure 8).Its metallographic characteristics are similar tothose of section 1618P.

1618PS, traverse 4: crossed 0.2 mm of troilite, 2.9 mm ofkamacite, 140 /tm of massive schreibersite, 0.6 mm of kama-cite, 2.9 mm of massive schreibersite, and 0.5 mm of kama-cite.

1618PS, traverse 5: crossed 0.4 mm of troilite, 0.7 mm ofmassive schreibersite, 4.6 mm of kamacite, 2.5 mm of massiveschreibersite, and 1.8 mm of kamacite. A representativeinterface profile for this traverse is reproduced in Figure 9.

1618PS, traverse 6: crossed 0.6 mm of massive schreibersite and1.9 mm of kamacite.

Section 1618PS(1) contains an outer edge of themassive schreibersite and spans the swathing kam-acite zone, containing a segment of its exteriorboundary. A detailed electron microprobe traversewas not made on this sample, but sufficient mea-surements were taken to establish the Ni and Pconcentration patterns for the swathing zone. Inthe direction away from the schreibersite andtoward the exterior grain boundary, the Ni inkamacite increased from slightly more than 2% atthe schreibersite interface to approximately 5%within the first 0.5 mm. Over the next 7 mm theNi concentration increased smoothly to approxi-mately 7%. It is in the first 0.2 mm of kamacitesurrounding the schreibersite that the Ni concen-tration increases most rapidly, and it is this zonethat is free of microrhabdites. The P concentrationincreases to approximately 0.1% within this sameregion of approximately 0.2 mm bordering theinterface. Upon entering the zone of microrhab-dite precipitation, P concentration falls off in kam-acite to a uniform level of approximately 0.05%for the remainder of the swathing zone. Late-stage microrhabdite precipitation seems to haveeffectively removed the P gradient that must havebeen produced within the swathing zone duringthe massive schreibersite growth, while apparentlyonly slightly modifying the Ni gradient.

The clear kamacite surrounding the centralschreibersite within the swathing zone is 100 to

Authors Copy

22 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

FIGURE 7. —Etched surface of Santa Luzia meteorite, USNM 1618: center, troilite surrounded byschreibersite and swathing kamacite; left, schreibersite area in swathing kamacite.

FIGURE 8. —Sketch of Santa Luzia meteorite slice in Figure 7, (coarse pattern, troilite; fine pattern,schreibersite; lines enclose areas of swathing kamacite; broken lines, locations of metallographicsections; numbered arrows, positions of electron microprobe traverses; stars, positions of residualtaenite areas).

200 /xm wide. This clear kamacite grades intokamacite containing profusion of microrhabdites,

and it is in this transitional area that Neumannbands are frequently best developed. Microrhab-

Authors Copy

NUMBER 21 23

TABLE 5. — Santa Luzia schreibersite-kamacite interface measurements

SpecimenNo.

1618P

1618P

1618P

1618PS

1618PS

1618PS

1618

1618

1618

TraverseNo.

1

2

3

4

5

6

7

8

9

Structure traversed & schreibersites measured

[Ph-a-Ph-a-Ph-a, 7.3 mm]

Massive schreibersite,

Start

Exit

Massive schreibersite,

Enter

Exit

Massive schreibersite,

Enter

Exit

[a-Ph-a-Ph-a-Ph, 5.3 mm]

Massive schreibersite,

Enter

Exit

Massive schreibersite,

Enter

Exit

Massive schreibersite,

Enter

Stop

[Ph-a-Ph, 1.4 mm]

Massive schreibersite,

Start

Exit

Massive schreibersite,

Enter

Stop

[Troilite-a-Ph-a-Ph-a, 7.4

Massive schreibersite,

Enter

Exit

Massive schreibersite,

Enter

Exit

[Troll1te-Ph-a-Ph-a, 1 cm]

Massive schreibersite,

Start

Exit

Massive schreibersite,

Enter

Exit

[Ph-a, 2.5 nin]

Massive schreibersite,

Start

Exit

[a-Ph-a, 350 ym]

Traverses across grainsite, 12x170 ym

TipCenter

Tip

[a-Y-Ph-a, 400 ym]

1.3 mm traverse

0.7 mm traverse

250 ym traverse

430 ym traverse

800 ym traverse

200 ym traverse

220 ym traverse

200 ym traverse

mm]140 ym traverse

2.9 mm traverse

0.7 mm traverse

2.5 mm traverse

0.6 mm traverse

boundary schreiber-

Taenite border schreibersite, 10x20 ym

[a-Y-Ph-a, 350 ym]

Taenite border schreibersite, 15x40 ym

Weight

*N1Sch

16.0

16.5

16.5

16.5

17.9

17.9

16.0

16.0

16.2

16.2

16.1

16.1

16.0

16.0

16.2

16.2

17.3

17.3

16.6

17.1

16.6

16.6

16.4

16.6

16.6

16.6

45.5

44.7

45.7

48.0

48.5

Percent

a

2.4

2.3

2.4

2.4

2.4

2.3

2.3

2.2

2.4

2.3

2.3

2.3

2.6

2.3

2.4

2.5

2.5

2.5

2.5

2.3

5.8

6.06.2

6.4

6.5

%Pa

0.07

0.06

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.08

0.06

0.07

0.08

0.07

0.07

0.07

0.07

0.03

0.04

0.04

0.03

0.03

Ni GradientLength(ym)

500

600

200

200

1000

900

1000

600

400

400

200

300

600

300

300

500

900

500

600

600

50

100

50

100

70

Wt.«N1

4.9

4.9

3.8

3.8

5.3

4.0

3.7

3.8

4.2

4.2

4.0

4.0

4.8

3.5

3.5

5.0

4.6

4.6

5.2

4.8

7.3

7.4

7.4

7.4

7.5

P GradientLength(ym)

500

400

100

Flat

800

500

300

400

300

200

300

300

800

300

200

250

700

800

800

500

100

100

100

150

150

Wt.*P

0.15

0.15

0.08

0.07

0.16

0.09

0.09

0.09

0.09

0.09

0.11

0.11

0.16

0.08

0.08

0.14

0.14

0.14

0.18

0.13

0.10

0.10

0.10

0.10

0.10

Atomic

%N1Sch

13.6

14.1

14.1

14.1

15.3

15.3

13.6

13.6

13.8

13.8

13.7

13.7

13.6

13.6

13.8

13.8

14.8

14.8

14.4

14.6

14.2

14.2

14.0

14.2

14.2

14.2

39.4

38.6

39.6

41.5

42.0

Percent

a

2.3

2.2

2.3

2.3

2.3

2.22.2

2.1

2.3

2.2

2.2

2.2

2.5

2.2

2.3

2.4

2.4

2.4

2.4

2.2

5.5

5.75.9

6.1

6.2

vn#ra

0.13

0.11

0.13

0.13

0.13

0.13

0.13

0.13

0.13

0.13

0.13

0.13

0.14

0.11

0.13

0.14

0.13

0.13

0.13

0.13

0.05

0.07

0.07

0.05

0.05

Authors Copy

24 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 5—continued

SpecimenNo.

772

772

772

772

TraverseNo.

10

11

12

13

Structure traversed & schreibersites measured

[a-Ph-a-Ph-a, 3.6 Itm]

Massive schreibersite, 650 um traverse

Enter

Exit

Massive schreibersite, 600 um traverse

Enter

Exit

[a-Ph-a, 2.3 mm]

Grain boundary schreibersite, 400x450 um

Enter

Exit

[a-Ph, 750 um]

Grain boundary schreibersite, 70 um traverse

Enter

Stop

[a-Ph-Y-a, 1.4 mm]

Taenite border schreibersite, 15x20 um

a-border

Y-border

Weight

*N1Sch

19.3

19.3

19.7

19.7

28.0

28.0

36.2

36.2

47.0

47.5

Percent

%N1a

2.8

2.8

2.8

2.8

3.7

3.9

4.6

6.0

%Pa

0.07

0.07

0.07

0.07

0.06

0.06

0.05

0.03

N1 GradientLength(um)

800

250

250

80

1400

500

500

200

Wt.%N1

5.8

4.7

4.7

5.8

6.9

6.3

7.2

7.4

P GradientLength(um)

500

250

250

600

1000

300

500

200

Wt.%P

0.14

0.10

0.10

0.15

0.11

0.12

0.09

0.07

Atomic

*N1Sch

16.5

16.5

16.8

16.8

24.0

24.0

31.2

31.2

40.7

41.1

Percent

SN1a

2.7

2.7

2.7

2.7

3.5

3.7

4.4

5.7

0.13

0.13

0.13

0.13

0.11

0.11

0.09

0.05

dites appear to increase in number per unit areafor the first few hundred microns away from theclear kamacite. As distance increases further, thereappears to be a slight increase in both concentra-tion and coarseness of microrhabdites. Coarsermicrorhabdites are occasionally associated withsubgrain boundaries within the kamacite. Thereare also occasional aligned microrhabdites, as ifthey had precipitated along Neumann bands thatare no longer present. Rare lamellar schreibersites2 fxm or less in width and as long as 200 ju,m arepresent. Near the outer edge of the swathing zoneoccasional segments of grain boundary type schrei-bersite protrude into the swathing kamacite or areobserved enclosed within it. Four areas that con-tain residual taenite as well as associated schreiber-site were observed within the two swathing zones(Figure 8), and close to their exterior borders.

The swathing zone boundary has suffered se-vere terrestrial weathering and now contains sec-ondary iron oxides that have penetrated into theadjoining kamacite. Remnant grain boundaryschreibersite is present, but it is encased in oxidesprecluding interface measurements. The amountof remnant schreibersite suggests that originallyas much as 50% of the length of the grain bound-ary may have been occupied by schreibersite. ANi determination on one of these schreibersitesgave a value of 29%.

Section 1618 is from an area of the large sectioncontaining coarse-structured Widmanstatten pat-tern (Figure 8). Parts of four kamacite grains 4 to5 mm wide are present. The general metallogra-phy of the section is similar to that described forBallinger, with the exception that only microrhab-dites are present within kamacite areas. Measure-ments described below were made along a 1.5 mmgrain boundary containing taenite and taenite-plessite areas in sequence with grain boundaryschreibersite.1618, traverse 7: combines data from three short (350 fim)

traverses perpendicular to the long direction of a singlegrain boundary schreibersite. It measured 170 fim in lengthand was separated from taenite by 10 to 20 fim at either end.Measurements were made near the two ends and in themiddle of the schreibersite. The interface profile for thecenter traverse is given in Figure 9.

1618, traverse 8: crossed 200 fim of kamacite, 10 fim of schrei-bersite, 15 fim of taenite, and 170 fim of kamacite. Thistaenite border schreibersite was embedded in the taenitelamella at one end of the schreibersite in traverse 7, approx-imately 200 fim away. The schreibersite interface profile isgiven in Figure 9.

1618, traverse 9: crossed 200 fim of kamacite, 15 fim of schrei-bersite, 25 fim of taenite, and 120 fim of kamacite. Thelocation was 200 fim farther along the same taenite measuredin traverse 8.

Section 772 is from a separate small specimen ofthe Santa Luzia meteorite. It combines both theswathing kamacite and Widmanstatten pattern

Authors Copy

NUMBER 21 25

400MICRONS

FIGURE 9. —Ni and P profiles at kamacite-schreibersite interfaces in Santa Luzia meteorite, USNM1618: traverses 5, 7, and 8 as indicated.

areas measuYed above on separate sections in onemicroprobe section of 2 cm2. Figure 10 is a scaledrawing indicating the areas of interest. A massiveschreibersite approximately 7 mm in length issurrounded by clear kamacite that in turn is en-closed in microrhabdite-containing kamacite. Amajor grain boundary containing grain boundaryschreibersite is within 3 mm of one end of themassive schreibersite. Leading off of that grainboundary is another that contains taenite as wellas grain boundary schreibersite, within 6 mm ofthe massive schreibersite. Measurements havebeen made in these areas as indicated in Figure10.

772, traverse 10: crossed 0.9 mm of kamacite, 0.6 mm of massiveschreibersite, 0.6 mm of kamacite, 0.6 mm of massive schrei-bersite, and 0.9 mm of kamacite. Part of this traverseincluding the first kamacite-schreibersite interface is given inFigure 11.

772, traverse 11: crossed 1.4 mm of kamacite, 0.4 mm of grainboundary schreibersite, and 0.5 mm of kamacite. The kama-cite-schreibersite interface on leaving the schreibersite isgiven in Figure 11.

772, traverse 12: crossed 0.7 mm of kamacite and 50 fim ofschreibersite. The Ni and P profiles are reproduced inFigure 11.

772, traverse 13: crossed a taenite boundary schreibersite (seeinset in Figure 10). The major features of this traverse aregiven in Figure 11.

The data on the Santa Luzia meteorite may beof particular significance from the standpoint ofclassification considerations. The schreibersite re-lationships observed for the large swathing zonesand their included schreibersite are in many wayssimilar to the Bellsbank meteorite. The schreiber-site relationships in the Widmanstatten areas ofSanta Luzia are similar to those observed in Ballin-ger. Could the differences in these two structures(Figures 6, 7) be due primarily to differences ininitial P concentration?

LEXINGTON COUNTY

The Lexington County, South Carolina, mete-orite is a coarse octahedrite of intermediate Ni

Authors Copy

26 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

r

1-2

mm

772FIGURE 10. —Scale drawing of schreibersites in an area of Santa Luzia meteorite section, USNM 772(location of electron microprobe traverses indicated; traverse 13 is of a taenite border schreibersite,indicated in more detail in the inset).

Authors Copy

NUMBER 21 27

0.06

3.8

400MICRONS

50-

4 5 -

4 0 -

ZiJ 35-

E

L

r 30-s>uS

25-

20-

15-

10-

5 '

PxlO

772 *I2

3 6

0.054.6 N

772 * 13

4 7

1

—

6.0

0.0

-

--

-

FIGURE 11. — Ni and P profiles at kamacite-schreibersite interfaces in Santa Luzia meteorite, USNM772: traverses 10 through 13 as indicated. (Dashed line indicates bulk Ni value.)

Authors Copy

28 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCEES1,

3 3 3 4

FIGURE 12. —Sketch of Lexington County section, USNM 3334, prepared from photomosaic of areastudied (arrows, electron microprobe traverses; large structure at left, massive schreibersitesurrounded by cohenite; subgrain boundaries, grain boundary schreibersites, taenite, and taeniteborder schreibersite are indicated).

content. Polished and etched surfaces reveal largeareas of well-developed Widmanstatten pattern.The kamacite of these areas contains rhabdites ofvarious sizes, grain boundary schreibersites, grainboundary and residual taenite and plessite, andNeumann bands. Oxidation has severely pene-trated the exterior surface of the specimen andinvaded the interior of the meteorite along majorgrain boundaries. Occasional skeletal schreiber-sites in the millimeter size range are present,normally completely surrounded by cohenite.These large schreibersites tend to be surroundedby areas of swathing kamacite that interrupt theregularity of the Widmanstatten pattern. Isolatedpatches of cohenite are also present.

Section 3334 contains a typical Lexington CountyWidmanstatten pattern area and a large schreiber-site surrounded by cohenite (6 x 0.3 mm). Edgesof the section and major grain boundaries havebeen invaded by oxidation, but the areas selectedfor microprobe analysis appear fresh. The pathsof the traverses listed below are indicated in Figure12 and numerical data are given in Table 6.

3334, traverse 1: crossed 1.4 mm of kamacite, a 30 fim schreiber-site partially embedded in cohenite at a cohenite-kamaciteinterface, 60 fim of cohenite, 90 fim of schreibersite embed-ded in cohenite, 0.3 mm of cohenite, 0.2 mm of massiveschreibersite, 0.1 mm of cohenite, 0.2 mm of kamacite, 50fim of cohenite, 0.2 mm of massive schreibersite, 80 fim ofcohenite, and 0.8 mm of kamacite.

3334, traverse 2: crossed 1.0 mm of kamacite, 50 fim of a largegrain boundary schreibersite, and 1.5 mm of kamacite. •

3334, traverse 3: crossed 0.8 mm of kamacite, a 15 fim widetaenite border schreibersite, 25 im of taenite, 0.3 mm ofkamacite, a 15 fim wide schreibersite in sequence withtaenite, 10 fim of kamacite, a 40 fim taenite, and 0.4 mm ofkamacite. The data for an abbreviated section of this traverseare given in Figure 3.

3334, traverse 4: crossed 0.1 mm of kamacite, a 10 fim widetaenite border schreibersite, 25 fim of taenite, and 50 fim ofkamacite.

3334, traverse 5: crossed 50 fim of kamacite, a 5 fim widetaenite-border schreibersite, 20 /xm of taenite, and 60 fim ofkamacite.

Figure 12 is a tracing of a photomosaic of thestudied area in Lexington County, section 3334.The path of the traverses, the phases crossed, andthe nature of the structure are indicated diagram-matically. This section illustrates a general prob-lem encountered in obtaining kamacite-schreiber-site interface data in the more carbon-rich coarse-structured octahedrites. The large, low-Ni schrei-bersites in these meteorites are frequently com-pletely surrounded by cohenite. Their Ni valuesmay be determined, but kamacite interfaces arenot present. Traverse 1, however, illustrates aninteresting schreibersite-cohenite relationship thatwill be observed later in other meteorites. Coheniteborders around large schreibersites occasionallyinclude small schreibersites, and more frequentlyhave small schreibersites at cohenite-kamacite in-

Authors Copy

NUMBER 21 29

TABLE 6. —Lexington County schreibersite-kamacite interface measurements

SpecimenNo.

3334

3334

3334

3334

3334

TraverseNo.

1

2

3

4

5

Structure traversed & schreibersites measured

[a-Ph-cohen1te-Ph-coheni te-Ph-coheni te-a-cohen1te-Ph-cohen1te-a, 3.5 mm]

Schreibersite embedded in cohenite at a-Interface, 25x120 ym

Schreibersite embedded in cohenite, 140x280ym

Enter

Exit

Massive schreibersite surrounded by cohen-ite (6x0.3 mm), 200 vm traverse

Enter

Exit

Massive schreibersite surrounded by cohen-ite (6x0.3 mm), 170 ym traverse

Enter

Exit

[a-Ph-a, 2.5 mm]

Grain boundary schreibersite, 500x40 ym

[a-Ph-Y-a-Ph-a-Y-a, 1.6 mm]

Taenite border schreibersite, 15x40 ym

Schreibersite near taenite, 10x20 ym

[a-Ph-Y-a, 200 ym]

Taenite border schreibersite, 5x10 ym

[a-Ph-Y-a, 140 ym]

Taenite border schreibersite, 5x10 ym

Weight

*NiSch

34.8

33.0

32.5

25.0

24.6

24.1

24.1

39.3

50.0

52.5

50.5

51.0

PercentXN1a

4.4

5.0

6.06.2

6.2

6.3

%

0

0

00

0

0

.04

.04

.03

.03

.03

.03

Ni GradientLength(ym)

150

300 .

5050

50

30

Wt.XNi

6.0

7.0

7.27.4

7.3

7.4

P GradientLength(ym)

200

200

150100

50

30

Wt.%P

0.07

0.06

0.06

0.05

0.05

0.05

Atomic

*NiSch

29.9

28.3

27.9

21.4

21.1

20.6

20.6

33.9

43.3

45.6

43.6

44.2

Percent

%N1a

4.2

4.8

5.75.9

5.9

6.0

XPa

0.07

0.07

0.05

0.05

0.05

0.05

terfaces. The Ni values in these schreibersitesincrease away from the large schreibersite. In thiscase, the large schreibersite contains approxi-mately 25% Ni, the small one completely embed-ded in cohenite has an apparent Ni gradient from32.5% to 33.0% increasing in the direction towardkamacite, and the kamacite interface schreibersitecontains 35'% Ni. The measurements in Table 6are otherwise comparable to those reported abovefor similar structures.

BAHJOI

The Bahjoi, India, meteorite is an observed fallof 1934, the only one included in this study. It is atypical Group I meteorite of above average Nicontent. Bahjoi contains complex silicate-troilite-graphite-chromite nodules surrounded by rims ofschreibersite, and, in turn, cohenite. Grain bound-ary schreibersites, rhabdites of various sizes, andtaenite border schreibersites are present. Taeniteand plessite are present in greater abundance thanin the previously examined meteorites, and pearl-itic and martensitic forms are common. An area

of the carbide haxonite was observed within apearlitic plessite area.

Section 1807(Sil) contains part of a complexsilicate-containing inclusion surrounded by schrei-bersite and cohenite, and small areas of kamacite.

1807(Sil), traverse 1: started within the complex inclusion, cross-ing first an island of kamacite surrounded by cohenite. Thephases in sequence were: 0.1 mm of cohenite, 0.2 mm ofkamacite, 0.1 mm of cohenite, 1.5 mm of schreibersite, 0.5mm of cohenite, 0.1 mm of schreibersite containing a slightNi gradient, and finally 1.0 mm of kamacite. The centralpart of this traverse is illustrated in Figure 13 and the data isgiven in Table 7.

Section 1807 is of a typical Widmanstatten areaof Bahjoi, free of large inclusions. Rhabdites,grain boundary schreibersites, and taenite borderschreibersites are present in abundance. Manytaenite-plessite areas are present in a variety ofmorphologies. One of these martensitic areas con-tains a 0.1 x 0.1 mm area of haxonite.

1807, traverse 2: crossed 30 /xm of grain boundary schreibersite,0.7 mm of kamacite, a 15 fim wide taenite border schreiber-site, 20 /xm of taenite, and 10 fim of kamacite.

The observations made on the Bahjoi meteorite

Authors Copy

30 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 7. —Bahjoi schreibersite-kamacite interface measurements

SpecimenNo.

1807(Sil)

1807

TraverseNo.

1

2

Structure traversed & schreibersites measured

[Cohenite-a-cohenite-Ph-cohenite-Ph-a, 3.5 mm]

Massive schreibersite, 10x4 mm, surroundedby cohenite and containing -30% silicate

Enter

Exit

Schreibersite 100 ym wide embedded incohenite at a-border

Enter

Exit

[Ph-a-Ph-a, 750 um]

Grain-boundary schreibersite, 35x250 um

Taenite-border schreibersite, 20x60 wm

Weight

*N1Sch

18.8

18.9

32.3

33.5

38.8

49.0

Percent

%M

4.2

5.2

6.2

%Pa

0.04

0.04

0.03

Ni GradientLength( m)

600

150

30

Wt.XNI

6.5

7.2

6.2

P GradientLength( m)

200

150

100

Wt.SP

0.06

0.06

0.06

Atomic Percent

%Ni,. . XN1 %PSch a a

16.0

16.1

27.7

28.8

33.4 5.0 0.07

42.4 5.9 0.05

are mainly similar in character to examples previ-ously described. Figure 13 is a composition profilebased on data from a part of traverse 1. Part ofthe massive schreibersite and its cohenite borderare included. Ni values are low in cohenite andincrease slightly upon moving away from the largeschreibersite. The small exterior schreibersite isenclosed on three sides by cohenite and contains ameasurable Ni gradient. This type of associationis common in high C Group I meteorites.

GOOSE LAKE

The Goose Lake, California, meteorite is a highNi member of Group I with a medium octahedritestructure. Polished and etched surfaces revealmany large cohenite bordered, skeletal schreiber-site inclusions, and occasional complex silicate -troilite-graphite-schreibersite-cohenite associa-tions. The cohenite has undergone partial decom-position to kamacite and graphite. Grain boundaryschreibersites, rhabdites in a range of sizes, andtaenite border schreibersites are present. Taenite-plessite areas are more abundant than in themeteorites examined previously and are presentin a variety of forms. Weathering has invaded thestructure along grain boundaries and is present incracks within the large schreibersites.

Section 1332 is dominated by two areas of cohen-ite bordered schreibersite of approximately 5 X 10mm each (photograph in Doan and Goldstein,1969:765). One smaller cohenite-schreibersite in-clusion is also present. Kamacite areas aroundthese inclusions are comparatively free of other

structures for distances up to 2 mm. Beyond this,normal Widmanstatten areas are present.

1332, traverse 1: began at the border of a taenite area, crossed2.4 mm of kamacite, and entered a 30 /xm wide grainboundary schreibersite near its contact with a taenite borderof a plessite area. Numerical data for this and succeedingGoose Lake traverses are given in Table 8.

1332, traverse 2: crossed 430 /xm of kamacite, 50 fim of schrei-bersite embedded in a cohenite rim of a large schreibersite,90 fim of cohenite, 0.2 mm of schreibersite, 50 pm ofcohenite, and 0.9 mm of kamacite.

1332, traverse 3: began by crossing a 40 fim wide schreibersiteassociated with a taenite-plessite area, crossed 2.5 mm ofkamacite, entered a 40 /xm wide schreibersite embayed incohenite, 150 fim of cohenite, and 0.3 mm of massiveschreibersite.

1332, traverse 4: started near the initial point of traverse 3within the taenite border, crossed 0.2 mm of kamacite, a 50fim wide grain boundary schreibersite, and 0.7 mm ofkamacite.

Section 1332(2) is similar to number 1332 in thatit also contains a 5 X 10 mm skeletal schreibersite-cohenite area. This section contains an unusuallylarge scrfreibersite that is only partially surroundedby cohenite, thus permitting kamacite-schreiber-site interface measurements. Its longest dimensionis 1.4 mm and has a maximum thickness of 0.6mm. Morphology suggests that it is an unusuallylarge grain boundary schreibersite rather than amassive schreibersite of the skeletal variety. Theareas away from these inclusions contained normalWidmanstatten pattern.

1332(2), traverse 5: crossed 0.1 mm of kamacite, 0.1 mm cohen-ite, 1.8 mm of massive schreibersite, 170 fxm cohenite, 2.1mm of kamacite, 20 /tin of cohenite, 0.1 mm of large grainboundary schreibersite, and 0.8 mm kamacite.

Authors Copy

NUMBER 21 31

4.20.04

400MICRONS

FIGURE 13. —Ni and P profile for Bahjoi meteorite, USNM 1807: left, massive schreibersitebordered by cohenite; right, small schreibersite partially enclosed in cohenite, but also with an edgein contact with kamacite.

1332(2), traverse 6: crossed 0.4 mm of kamacite, 25 fim ofcohenite, 1.2 mm of schreibersite containing a small Nigradient over part of its distance, and 0.5 mm of kamacite.This traverse crosses the length of the large schreibersite innumber 5.

1332(2), traverse 7: crossed 40 /*m tip of large schreibersite, 1.2mm of kamacite, 0.2 mm of second tip of same schreibersite,and 0.4 mm of kamacite. This is the same large schreibersitetraversed in numbers 5 and 6.

1332(2), traverse 8: crossed 0.5 mm of kamacite, 150 (im of thethird tip of the large schreibersite measured in traverses 5through 7, and 0.5 mm of kamacite.

1332(2), traverse 9: crossed 0.4 mm of kamacite, 150 /xm of asmall schreibersite near the large schreibersite in traverses 5through 8, and 0.3 mm of schreibersite.

1332(2), traverse 10: crossed 0.2 mm of kamacite, 30 im of agrain boundary schreibersite, and 0.2 mm of kamacite.

1332(2), traverse 11: crossed 0.3 mm of massive schreibersiteclose to traverse number 5, 0.1 mm of cohenite, a 20 imwide schreibersite embedded in cohenite, 0.1 mm of kama-cite, and 50 fim of cohenite.

The associations measured in Goose Lake aresimilar in character to those described above for

other meteorites. The schreibersite precipitates inthis case, however, exist in an environment that ismore Ni-rich.

BALFOUR DOWNS

The Balfour Downs, Western Australia, mete-orite is one of the highest Ni members of GroupI. The first prepared surface of this specimen wasperpendicular to the one shown in Figure 14,along its bottom edge. This area of approximately12 cm2 was free of large inclusions, having theregular structure of the upper right-hand area inFigure 14. On the basis of this casual observation,Balfour Downs was assumed to be a low P mediumoctahedrite. Actually, it contains localized massiveschreibersite enclosing significant quantities of sil-icates. These schreibersites are normally enclosedin cohenite and surrounded by swathing kamaciteareas that interrupt the regular Widmanstatten

Authors Copy

32 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 8. —Goose Lake schreibersite-kamacite interface measurements

SpecimenNo.

1332

1332

1332

1332

1332(2)

1332(2)

1332(2)

1332(2)

1332(2)

1332(2)

1332(2)

TraverseNo.

1

2

3

4

5

6

7

8

9

10

11

Structure traversed & schreibersftes measured

[y-a-Ph-y, 2.4 mm]

Schreibersite at y-border, 25 ym wide

[a-Ph-cohenite-Ph-cohenite-a, 1.6 mm]

Schreibersite embaying cohenite, 55 ym tra-verse

Enter

Exit

Massive schreibersite enclosed in cohenite(3.4x0.2 mm), 260 ym traverse

[Ph-a-Ph-cohenite-Ph, 3.1 mm]

Schreibersite near taenite, 40x70 ym

Schreibersite embaying cohenite, 30x50 ym

Massive schreibersite (6x2 mm), 300 ymtraverse

[y-a-Ph-a, 950 um]

Grain boundary schreibersite (60x250 ym),60 ym traverse

[a-cohen1te-Ph-cohen i te-a-cohen1te-Ph-a,5.3 mm]

Cohenite bordered massive schreibersite(2x6 mm), 1.8 mm traverse

Enter

Middle

Exit

Large schreibersite with partial coheniteborder (1.4x0.6 mm), 100 um traverse

[a-cohenite-Ph-a, 2.2 mm]

Large schreibersite with partial coheniteborder (1.4x0.6 mm), 1.2 mm traverse

Enter

Middle

Exit

[Ph-a-Ph-a, 1.8 mm]

Large schreibersite of traverses 5 and 6

40 um traverse

200 ym traverse

Enter

Exit

[a-Ph-a, 1.2 mm]

Large schreibersite of traverses 5-7,150 ym traverse

Enter

Exit

[a-Ph-a, 0.9 mm]

Small schreibersite near traverses 5-8(250x400 ym), 150 ym

Enter

Exit

[a-Ph-a, 0.5 mm]

Grain boundary schreibersite (0.03x1.6 mm)associated with plessite, 30 ym traverse

[Ph-cohenite-Ph-a-cohenite, 0.7 mm]

Massive schreibersite (same as traverse 5),0.4 mm traverse

Start

Exit

Schreibersite (20x60 ym) embedded incohenite-a border

Weight

*N1Sch

52.0

35.0

34.0

25.0

46.5

36.5

18.5

41.0

18.7

18.2

18.5

31.3

29.3

27.3

27.3

31.0

27.0

27.0

29.1

29.1

27.2

27.2

44.0

18.3

18.3

37.3

Percent

a

6.0

4.1

6.0

4.3

5.1

3.8

3.6

3.9

3.3

3.5

3.7

3.8

3.1

3.6

5.8

4.5

%Pa

0.03

0.05

0.04

0.05

0.05

0.06

0.07

0.06

0.06

0.06

0.06

0.06

0.06

0.06

0.05

0.06

N1 GradientLength(ym)

25

100

50

150

100

200

500

400

600

300

500

400

300

250

100

150

Wt.*N1

7.2

6.4

7.05.7

7.2

5.7

5.8

6.0

6.2

6.8

6.5

6.5

6.0

5.5

7.4

6.0

P GradientLength(ym)

100

50

100

200

Flat

100

250

200

150150

150

150

100

100

200

100

Wt.iSP

0.07

0.06

0.07

0.08

0.05

0.08

0.09

0.08

0.08

0.09

0.09

0.08

0.07

0.07

0.06

0.09

Atomic

*N1Sch

45.1

30.1

29.2

21.4

40.2

31.4

15.8

35.4

15.9

15.5

15.8

26.9

25.1

23.4

23.4

26.6

23.1

23.1

25.0

25.0

23.3

23.3

38.0

15.6

15.6

32.1

Percent

%Ma

5.7

3.9

5.7

4.1

4.9

3.6

3.4

3.7

3.13.3

3.5

3.6

3.0

3.4

5.5

4.3

%Pa

0.05

0.09

0.07

0.09

0.09

o.n

0.13

0.11

o.n0.11

o.no.n

0.13

0.13

0.09

0.11

Authors Copy

NUMBER 21 33

FIGURE 14. —Etched surface of Balfour Downs meteorite, USNM 3202: left-hand side of the slice,large schreibersites, several containing silicates; (outlined area indicates polished section that wasprepared for detailed study).

Authors Copy

34 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 9. —Balfour Downs schreibersite-kamacite interface measurements

SpecimenNo.

3202

3202

3202

3202

TraverseNo.

1

2

3

4

Structure traversed & schreibersites measured

[Ph-a-Ph-a-Ph-Y, 3.3mm]

Massive schreibersite,

Start

Exit

1550 ym traverse

Residual schreibersite 30x800 ym, asso-ciated with taenite, crossed at twolocations

[Ph-a-Ph-a, 3.1 mm]

Massive schreibersite,

Start

Exit

Massive schreibersite,

Enter

Exit

[o-Ph-o-cohen1te-Ph, 2.3 n

Massive schreibersite,

Enter

Exit

Massive schreibersite,

Enter

End

[Ph-cohenite-a-cohenite-PI

Massive schreibersite.

Start

Exit

Massive schreibersite,

Enter

Stop

600 ym traverse

500 ym traverse

400 ym traverse

1350 ym traverse

,2.1 mm]

800 ym traverse

250 ym traverse

Weight

*N1Sch

20.5

21.0

41.041.0

21.3

21.5

21.5

21.5

21.0

21.0

20.0

20.0

19.5

19.5

20.5

21.0

Percent

*N1a

2.9

5.35.5

2.6

2.83.0

2.92.8

%Pa

0.05

0.040.03

0.05

0.05

0.05

0.05

0.05

Ni GradientLength(ym)

1150

10050

300

2501500

250150

Wt.%N1

6.7

00 00

4.8

4.87.0

5.34.8

P GradientLength(ym)

200

100

300

200300

200150

Wt.fcP

0.09

0.09

0.09

0.09

0.11

0.10

0.08

Atomic

*N1Sch

17.5

17.9

35.435.4

18.2

18.4

18.4

18.4

17.9

17.9

17.1

17.1

16.7

16.7

17.5

17.9

Percent

a

2.8

5.15.2

2.5

2.72.9

2.72.7

%Pa

0.09

0.070.05

0.09

0.09

0.09

0.09

0.09

pattern (center left in Figure 14). Grain boundaryschreibersite, schreibersite associated with taenite-plessite areas, and rhabdites in a variety of sizesare present. Taenite-plessite areas are present in avariety of forms and Neumann bands are com-mon.

Section 3202 was prepared from the outlinedarea in Figure 14. It was selected particularly forthe presence of massive schreibersite that waspartially free of cohenite borders. Figure 15 is aphotograph of a photomosaic of a portion of thissection. The paths of the four traverses describedbelow are indicated in Figure 16, a sketch basedon Figure 15.

3202, traverse 1: crossed 1.5 mm of massive schreibersite, 4.5mm of kamacite, a 40 fim wide schreibersite surrounding ataenite area, 0.1 mm of kamacite, recrossed a 30 fim widthof the same schreibersite previously crossed, 50 /urn ofkamacite, and 50 fim of taenite. Numerical data for this andsucceeding Balfour Downs traverses are given in Table 9.

3202, traverse 2: crossed 0.6 mm of massive schreibersite, 0.6

mm of kamacite, 0.5 mm of massive schreibersite, and 1.5mm of kamacite.

3202, traverse 3: crossed 0.3 mm of kamacite, 0.4 mm of massiveschreibersite, 0.2 mm of kamacite, 80 jxm of cohenite, and1.3 mm of massive schreibersite. The path of this traverse isindicated in Figures 15 and 16 and the electron microprobeprofile is shown in Figure 17.

3202, traverse 4: crossed 0.8 mm of massive schreibersite, 0.1mm of cohenite, 0.8 mm of kamacite, 0.1 of cohenite, and0.2 mm of massive schreibersite.

Also indicated in Figure 16 are Ni concentrationsin various areas of the two massive schreibersites.One is tempted to conclude that a slight Ni gra-dient exists, with the lowest Ni values being adja-cent to cohenite. The differences, however, aresmall, and this should undoubtedly be looked atmore systematically in the future. The Ni and Pprofile reproduced in Figure 17 illustrates severalinteresting features. The Ni concentration in kam-acite is low close to these large schreibersites, asare the Ni interface values. There also appears tobe a Ni gradient in cohenite, with Ni increasingaway from the massive schreibersite.

Authors Copy

NUMBER 21 35

FIGURE 15. —Photograph of photomosaic of area of Balfour Downs meteorite examined, USNM3202: large schreibersite inclusions in kamacite, partially bordered by cohenite. (Line indicatesposition of traverse 3; see Figure 17.)

Authors Copy

36 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

I mm

3202

FIGURE 16. —Sketch based on Balfour Downs section illustrated in Figure 15 (arrrows indicatelocation of traverses; numbers give schreibersite Ni concentrations in weight percent)

Authors Copy

NUMBER 21 37

0.05

2.9

600MICRONS

FIGURE 17. —Ni and P profiles in Balfour Downs meteorite, USNM 3202, traverse 3 in Figure 16(schreibersite inclusion at the right enclosed in cohenite).

Discussion

EQUILIBRIUM CONSIDERATIONS OF PHASE GROWTH

The total amount of P present in an iron mete-orite has a marked effect on the structural devel-opment process, and consequently on the finalstructure observed in prepared sections. Studiesof Widmanstatten pattern growth (Wood, 1964;Goldstein and Ogilvie, 1965b; Goldstein and Axon,1973) have demonstrated that octahedrite meteor-ites cooled from high temperatures and were inequilibrium down to 700° C. Schreibersite growthstudies (Goldstein and Ogilvie, 1963; Doan andGoldstein, 1969) led to similar conclusions concern-ing equilibrium temperatures for iron meteoritesin general. In a discussion of the effects of P onthe development of the Widmanstatten pattern,Goldstein and Doan (1972) estimate that equil-brium in iron meteorites was maintained down to650° C, and probably to 600° C. Recent work byRandich (1975) suggests the schreibersite-kamacite

equilibrium is maintained down to 500° C fordiffusion distances as great as 1000 fim. Previouswork, therefore, suggests that many of the schrei-bersite precipitates observed in iron meteoritesnucleated and grew under equilibrium conditionsduring a significant part of their cooling history.It is the purpose of this section to examine schrei-bersite growth and related structural developmentin the selected group of coarse-structured ironmeteorites in terms of the equilibrium Fe-Ni-Pdiagram of Doan and Goldstein (1970). How farcan equilibrium considerations take us in explain-ing observed structures?

Before proceeding, however, brief mentionmust be made of the effects of S, C, and Co onthe Fe-Ni-P system. These elements are present iniron meteorites in quantities that may be compa-rable to or in excess of P. The role of S has beendiscussed recently by Goldstein and Axon (1973)and Wai (1974). Goldstein and Axon (1973) pointout that S may be important in lowering the

Authors Copy

38 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

temperature of metal melts, thereby mobilizingmetal without melting silicates. This effect, how-ever, would only extend down to temperatures inthe range of 1000° C. At lower temperatures,material of Fe-FeS eutectic composition becomesenclosed in solidifying metal, and the solubility ofS in the metallic phases becomes negligibly small.Sulfur is removed from the metal by this process,becoming isolated in complex sulfide nodules (ElGoresy, 1965).

The situation with respect to carbon is muchmore complex. Graphite or graphite-troilite nod-ules present in many meteorites represent hightemperature segregations of C as well as S. Thepetrography of these complex associations hasbeen described by El Goresy (1965). Comparativelylarge amounts of C, however, remain soluble inthe metal to lower temperatures, temperatureswithin the schreibersite growth range. Brett (1967)has discussed the occurrence of cohenite ((Fe,Ni)3C)in meteorites and described its formation in termsof phase equilibria using an extrapolated Fe-Ni-Cdiagram. He proposed that cohenite formedwithin a narrow range of bulk Ni values in thetemperature range of 650° to 610° C. Scott (1971a)has discussed carbides in iron meteorites in greatdetail from a petrographic point of view. He hasalso reported two new carbides, haxonite((Fe,Ni,Co)23C6) (Scott, 1971b), and a phase desig-nated W-carbide ((Fe,Ni,Co)2.5C), both of whichcontain considerably more Ni and Co in theirstructure than is observed in cohenite. Scott(1971a) suggests that cohenite may have formed atsomewhat lower temperatures than those sug-gested by Brett (1967). Both Scott (1971a) andBuchwald (1976) point out that cohenite is muchmore widely distributed in meteorites than previ-ously realized. Cohenite and schreibersite are fre-quently observed in intimate association (Drake,1970; Clarke, 1969; Buchwald, 1976), with cohenitenormally encapsulating large schreibersite inclu-sions in the high C Group I meteorites. Theseschreibersite-bordering cohenites normally encloseoccasional small later-formed schreibersites as hasbeen illustrated above (Figure 13). This morphol-ogy suggests the possibility of concurrent schrei-bersite-cohenite growth after major amounts ofschreibersite have formed. Carbon, therefore, isseen to be a component of the system of directconcern here. Experimental data on carbon's ef-fect on the Fe-Ni-P system at low temperatures is

not available, and it has been assumed that it canbe ignored here except in cases where its physicalpresence in the form of cohenite or other carbidesis important. This assumption may well requirereevaluation,when appropriate experimental databecomes available.

Another simplifying assumption made in thisstudy is to ignore the presence of Co, an elementpresent in iron meteorites in the range of 0.3% to0.7% (Moore et al., 1969). Co distributes itselfbetween the three phases of interest, following Feand seemingly having little effect on the equilibriainvolved.

In the following sections, individual meteoriteswill be discussed in terms of structural develop-ment on equilibrium cooling through the low tem-perature portion (995° to 550° C) of the Fe-Ni-Psystem. The Doan and Goldstein (1970) isothermalsections were recalculated on an atom percentbasis, and four of these sections plotted on atriangular coordinate system are given in Figure18.

Equilibrium in the system at hand may be de-fined as (1) the phases present are homogeneous,that is to say, free of measurable concentrationgradients, and (2) the compositions of individualphases are related to the bulk composition by thephase diagram. In the case at hand, this meansthat (a) the bulk composition is the composition ofthe single phase present when it lies within thekamacite (a) or taenite (y) field of the diagram;(b) the bulk composition is on a tie line joining thecompositions of the two phases present in thekamacite + schreibersite (a + Ph), the kamacite +taenite (a + y), and the taenite + schreibersite (y+ Ph) fields of the diagram; and (c) the bulkcomposition is related to the three phases presentby the compositions at the corners of the triangularkamacite + taenite + schreibersite (a + y + Ph)field of the diagram. The proportions of thephases present in the two- and three-phase fieldsmay be calculated using lever rule principles. Cal-culations of this type were carried out for each ofthe meteorite compositions considered and arepresented in a series of tables to follow (Tables10-15). Temperatures, the proportion of phasespresent, and their Ni contents are given. Thesedata will be discussed for each meteorite in termsof observed structures. Selected compositions rep-resenting these meteorites are plotted on a seriesof isothermal sections of the iron-rich corner of

Authors Copy

NUMBER 21 39

OS?

Authors Copy

40 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 10. —Equilibrium cooling of composi-tions appropriate to the Coahuila meteorite(percentage of phases present at indicatedtemperatures with their Ni contents)

°c

995975925875850750700650600550

°C

995975925875850750700650600550

0.5

%a

6075939999.7

1.1

%a

125082949898

Atomic

*Y

100100100100100402571

Atomic 5

%y

1001001001008850185

i P, 5

SPh

0.3

5 P, 5

SPh

122

.3 Atomic

a

3.34.04.45.25.2

.3 Atomic 3

a

3.53.84.55.05.25.2

t N1

%N

5555569

1214

. N1

%H

5555569

11

Y

.3

.3

.3

.3

.3

.5

.2

Y

.3

.3

.3

.3

.6

.9

.1

2 N iPh

14

*NiPh

9.01313

the Fe-Ni-P system from 850° to 550° C in Figure19.

COAHUILA. —The Coahuila hexahedrite is a me-teorite of both low Ni (5.6 weight %, 5.3 atomic%) and low P (0.3 weight %, 0.5 atomic %). Thesebulk composition values are well established in theliterature and are generally consistent with themajor features of the observed structure. Thepresence of occasional large schreibersites as de-scribed above suggests, however, that the P valuemay be*somewhat low. For this reason the calcula-tions given in Table 10 were made using two Pvalues, the literature value (0.5 atomic %) and oneapproximately twice as high (1.1 atomic %). Theequilibrium phases present, their amounts in molepercent, and their Ni contents in atomic percentare listed in Table 10 for 10 temperatures between995° to 550° C.

Due to Caohuila's low P (0.5 atomic %), thepredicted structural development sequence usingthe Fe-Ni-P diagram is essentially the same as thatderived using the Fe-Ni diagram. At 995° C (Table10), taenite of the bulk meteorite composition isthe only phase present, and this situation contin-ues until the meteorite cools to below 850° C. Asthe temperature drops an additional 200° C, tae-

nite transforms almost completely to massive singlecrystal kamacite. The last of the taenite transformsto kamacite below 600° C, in the temperaturerange where schreibersite first nucleates. This se-quence accounts for kamacite containing homoge-neously nucleated rhabdite, the most prominentfeature of Caohuila. This low temperature ofinitial schreibersite nucleation and the smallamount formed, however, make it difficult tounderstand the presence of the occasional largerschreibersites mentioned above.

If a somewhat higher P content is assumed,significant differences are revealed. With an as-sumed content of 1.1 atomic % P (Table 10),transformation of taenite to kamacite starts morethan 50° C higher, extends over a slightly largertemperature range, and is complete at a highertemperature than in the previous case. Schreiber-site precipitates, while more than 5% taenite re-mains in the structure, and a total of 2% schreiber-site is present at 600° C. In this case, one wouldexpect schreibersite to nucleate at taenite-kamaciteboundaries, but no direct evidence for this hasbeen developed. The actual equilibrium P valuemay well lie between the two selected values,perhaps 0.8 atomic %. This amount of P wouldproduce upon cooling a somewhat smaller quantityof schreibersite, initially nucleating at a tempera-ture below 650° C, and consuming any remainingresidual taenite. Both sufficient P and adequatetime at higher temperatures would be available topermit growth of the isolated larger schreibersitesobserved, and the bulk of the meteorite wouldappear to have formed under conditions of lowertotal P.

The 5.3 atomic % Ni and 0.8 atomic % P com-positions are plotted on a series of isothermalsections in Figure 19. At 850° C, Coahuila's com-position is just inside the kamacite-taenite field ofthe diagram, yielding a structure containing asmall amount of kamacite within a taenite matrix.The field moves in relation to this compositionwith decreasing temperature, until at 650° C onlya small amount of taenite remains in equilibriumwith a kamacite matrix. At 600° C, the compositionis within the kamacite-schreibersite field, perhapshaving passed through the corner of the kamacite-taenite-schreibersite field. By 550° C, the composi-tion lies well within the kamacite-schreibersite fieldof the diagram, with approximately 98% kamacite

Authors Copy

NUMBER 21 41

850°C

0

0100% Fe

A B E L L S B A N K

• COAHUILA

• SANTA LUZIAS-BALLINGER

• LEXINGTON COUNTY- BAHJOI

A G O O S E LAKE-BALFOUR DOWNS

5.0

5.3

6.2

6.2

7.1

7.8

%P3.6

0.8

2.7

1.4

1.8

2.7

10 15ATOMIC %Ni

20

FIGURE 19. — Meteorite bulk compositions plotted on series of isothermal sections of Fe-Ni-Psystem.

Authors Copy

42 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

containing 5.1 atomic % Ni and 0.5 atomic % P,being in equilibrium with 2% schreibersite contain-ing 13 atomic % Ni and 25 atomic % P.

The development sequence described above ac-counts for the major features of Coahuila struc-ture. Equilibrium cooling of this bulk compositionproduced large single crystals of kamacite contain-ing minor inclusions. The earliest-formed schrei-bersite nucleated and grew on preexisting sulfideinclusions where available, or perhaps at kamacite-taenite interfaces. The great majority of the indi-vidual schreibersite crystals, however, may be as-sumed to have nucleated homogeneously withinkamacite. If this process had stopped at 550° C,the meteorite would contain kamacite of 5.1 atomic% Ni and 0.5 atomic % P, and schreibersite of 13atomic % Ni. Phase composition measurementsfrom the literature and those reported above indi-cate that diffusion controlled growth continued tolower temperatures, modifying phase composi-tions significantly, while producing only subtlechanges in the gross structural features. Thesechanges in phase compositions and subtle differ-ences in structure will be discussed in some detaillater in this paper.

BALLINGER. —The Ballinger meteorite is a lowNi octahedrite probably containing more P thanthe analytical value given in Table 1 (0.4 weight %P, 0.7 atomic % P). The Ballinger photograph(Figure 6) shows three areas containing largeschreibersite inclusions, perhaps suggesting thatthe amount of P should be doubled. The calcula-tions given in Table 11, therefore, were madeusing 6.2 atomic % Ni (6.5 weight %) and both 0.7atomic % and 1.4 atomic % P.

During equilibrium cooling in the low P case(Table 11), taenite is the only phase present untilkamacite appears around 800° C. At 600° C, theequilibrium structure contains 89% kamacite, 10%taenite, and only 1% schreibersite. Upon coolingto 500° C, all of the taenite would be expected totransform, producing an equilibrium structure of99% kamacite and 1% schreibersite. Both the pres-ence of the massive schreibersites and residualtaenite in the actual structure argue against thisinterpretation. The growth of such large schrei-bersites requires more P, and the presence oftaenite demonstrates that equilibrium cooling wasnot maintained down to 550° C.

The 1.4 atomic % P calculation for Ballinger(Table 11 ) leads to a more reasonable interpreta-

TABLE 11. —Equilibrium cooling of composi-tions appropriate to the Ballinger meteorite(percentage of phases present at indicatedtemperatures with their Ni contents)

"C

995975925875850750700650600550

°C

995975925875850750700650600550

0

%a

1560788999

1

%a

543062778896

.7 Atomic

*Y

10010010010010085402210

.4 Atomic

%y

10010010095956836208

% P, 6

%Ph

11

% P, 6

%Ph

122344

.2 Atomic

*N1a

3.74.04.95.26.0

.2 Atomic

XN1a

3.94.04.04.55.05.25.7

* N1

XN1y

6.26.26.26.26.26.69.11114

* N1

%N1y

6.26.26.26.36.37.29.11114

*N1

1315

XN1

45791315

Ph

Ph

.4

.3

.1

.0

tion of the structure. Small amounts of kamaciteform in taenite around 900° C, and by 850° C thetaenite matrix contains 4% kamacite and 1%schreibersite, a situation that might be expected tolead to the growth of several rather large schrei-bersites on subsequent cooling rather than a num-ber of small ones. By 750° C, one-third of themeteorite is kamacite, undoubtedly the swathingkamacite surrounding the large schreibersites thatare present. As the temperature falls to 650° C,the meteorite transforms to 77% kamacite, andthe schreibersite doubles its Ni content to 9.0atomic %, while increasing its amount by one-half.As the temperature drops to 600° C, kamaciteincreases in volume, schreibersite increases in bothamount and in Ni content, and taenite shrinks to8%. The higher P content seems to be moreconsistent with the observed structure and couldactually be a little low. The presence of taeniteand high Ni schreibersite in the actual structure(see above) demonstrates that equilibrium coolingdid not persist down to 550° C, and that nonequi-librium phase growth continued to lower temper-atures. As in the case with Coahuila, however, thegross features of the structure are accounted forby equilibrium cooling of this particular composi-tion (Figure 19).

Authors Copy

NUMBER 21 43

SANTA LUZIA. —The Santa Luzia meteoritestructure is dominated by schreibersite and schrei-bersite-troilite inclusions surrounded by broadareas of swathing kamacite (Figure 7). The litera-ture values for its bulk composition, 6.3 atomic %Ni and 1.6 atomic % P (6.6 weight % Ni and 0.9weight % P, Table 1), are very close to thoseassumed for the high P Ballinger composition(Table 11), only a 0.1% increase in Ni and 0.2% inP. This apparently minor change, however, hasimportant ramifications for the structural devel-opment process in Santa Luzia. In Ballinger theexpected sequence of phases with falling tempera-ture,

y —> a + y —» a + y + Ph,

is followed. In Santa Luzia this sequence is inter-rupted around 850° C, where over a range oftemperature taenite is in equilibrium with schrei-bersite, the sequence of phases being

TABLE 12. —Equilibrium cooling of composi-tions appropriate to the Santa Luzia meteorite(percentage of phases present at indicatedtemperatures with their Ni contents)

Ph Ph.

The growth of significant amounts of schreibersitewithin taenite subsequent to the disappearance ofpreexisting kamacite produced conditions that re-sulted in the development of large volumes ofswathing kamacite and the unusual schreibersitemorphology observed in the meteorite.

The actual bulk P value for Santa Luzia may besomewhat larger than the 1.6 atomic % value givenin Table 1. The calculations summarized in Table12 indicate, however, that the same sequence ofphases would be expected whether 1.6 atomic %or 2.7 atomic % P is assumed. In either case,kamacite and taenite would have been in equilib-rium at 925° C. It is reasonable to assume that thiskamacite nucleated within taenite at the preexist-ing troilite-taenite grain boundaries, with kamacitelamellae radiating away from the troilite. Uponcooling to 875° C, the proportion of kamacite wasmarkedly reduced at both P levels. This smallamount of kamacite presumably remained local-ized near the troilite nodules. The taenite inter-faces with troilite and the kamacite-taenite inter-faces acted as nucleation sites for schreibersiteupon further cooling, producing a skeletal struc-ture that sets the pattern for subsequent schreiber-site growth. Nearly half in one case, and morethan half in the other, of the schreibersite thatformed upon cooling the meteorite to 500° C

°c

995975925875850750700650600550

°C

995975925875850750700650600550

1

%a

103

2660768795

2

%a

4350354

2760768891

.6 Atomic

%y

1001009096987137218

.7 Atomic 3

*Y

57506590936632163

t p, 6

%Ph

1233355

i P, 6

SPh

6778899

.3 Atomic

%Nia

4.54.0

4.04.55.05.25.8

.2 Atomic 3

*N1a

5.35.44.94.0

4.04.55.05.25.4

i Ni

SN1Y

6.36.36.46.46.37.29.1

1114

i Ni

XN1Y

6.87.17.06.46.37.29.11114

%NiPh

4.44.55.37.19.01315

*NiPh

4.44.55.37.19.01314

formed at the relatively high temperature of 850°C. Upon further cooling Ni and P continued todiffuse into the central schreibersite areas, P con-tributing to additional growth and Ni contributingto both growth and Ni enrichment. As a conse-quence of this localized growth of schreibersite, alarge surrounding volume of low Ni metal wasproduced, the swathing kamacite zone of the finalstructure. This proposed growth sequence ex-plains the morphology of these large sc(hreibersitestructures without the necessity of invoking precip-itation from the liquid, a process that would re-quire a bulk composition of more than 4.9 atomic% P at high temperatures.

The broad swathing zone developed as the me-teorite cooled from 850° C down into the 700° to650° C range (Figure 7). This transformation takesplace easily at these temperatures as only a rela-tively small amount of Ni movement is requiredfor both schreibersite growth and enrichment,and for production of the taenite border at theouter edge of the zone. This expanding taeniteborder acts as a barrier to isolate the large centralschreibersites from the bulk material beyond theswathing zone. As the temperature drops anddiffusion becomes increasingly restricted, condi-tions are met for the nucleation of schreibersite at

Authors Copy

44 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

this taenite border. By 550° C the taenite borderhas been transformed and replaced with a contin-uous grain boundary, approximately half of whichis occupied by grain boundary schreibersite. Thisis amply supported by examination of Santa Luziaspecimens. The swathing zone boundary containsa large amount of schreibersite and is, with a fewminor exceptions, free of residual taenite. Theswathing zone itself is also free of large schreiber-sites and residual taenite, with the exception ofseveral small areas very close to the swathing zoneborder. These appear to have been small areas ofresidual taenite that became trapped within theswathing zone at a very late stage and may havebeen part of the taenite border system at highertemperatures. Competition for P from neighbor-ing schreibersites appears to be the reason thatthese areas were not converted completely toschreibersite.

Schreibersite equilibration with taenite in the850° C temperature range results in a redistribu-tion of P in the meteorite leading to the develop-ment of areas of coarse Widmanstatten pattern(Figure 7). At these high temperatures P diffuseswith ease, lowering its level even in the distantbulk metal to approximately 1.0 atomic %. Subse-quent to this, the swathing zone developmentdescribed above took place, resulting in the isola-tion of that part of the meteorite from the areaswhere the coarse Widmanstatten pattern formed.The areas beyond the swathing zones developedan effective bulk composition similar to that of thelow P Ballinger case described above. Widmanstat-ten pattern nucleated in these areas at tempera-tures below 750° C, and grain boundary schreiber-site probably began nucleating at kamacite-taeniteinterfaces around 600° C. The observed metallog-raphy of these areas is consistent with this expla-nation.

The bulk composition of Santa Luzia is plottedon the diagrams in Figure 19. As in the previouscases, it is obvious from the observations reportedin the results section that bulk equilibrium was notmaintained down to 550° C, and that phase modi-fication persisted to considerably lower tempera-tures.

LEXINGTON COUNTY AND BAHJOI. —The bulkcompositions of the Lexington County and Bahjoimeteorites appear to be poorly established (Table1). They are structurally similar, carbon- and sul-fur-rich members of Group I with intermediate

TABLE 13. —Equilibrium cooling of composi-tions appropriate to the Lexington Countyand Bahjoi meteorites (percentages of phasespresent at indicated temperatures with theirNi contents)

°c

995975925875850750700650600550

°C

995975925875850750700650600550

0.

%a

40707999

1.

%a

510

Tr42607894

5 Atomic

%y

100100100100100100603020

8 Atomic 3

%y

1009590989796543516

t P, 7.

%Ph

0.31

i p, 7.

0.32344556

1 Atomic

%M a

4.14.95.26.9

1 Atomic

%N1a

5.55.3

(4.2)4.55.05.26.5

S N1

%My

7.17.17.17.17.17.19.21214

i Ni

%N1y

7.17.27.37.17.27.29.11114

SUM

1316

%M

5.5.5.5.7.9.1315

Ph

Ph

203310

bulk Ni. The P values of Lexington County (0.5atomic % P, 0.3 weight % P) are too low to bereconciled with the amount of schreibersite ob-served in prepared sections. The LexingtonCounty bulk Ni value is also probably somewhathigher than 7.0 weight % Ni when the contributionof large schreibersites is taken into consideration.Much of the kamacite in this meteorite contains7.0 weight % Ni, and some of it contains slightlymore Ni. Bahjoi also contains isolated large schrei-bersites, requiring its effective bulk P to be greaterthan 0.3 weight % P. For these reasons, the equilib-rium phase calculations in Table 13 were made atthe intermediate Ni value of 7.1 atomic % (7.5weight % Ni), and at 0.5 atomic % P and 1.8atomic % P (1.0 weight % P). These calculationssuggest that the actual P value probably lies be-tween the two, probably greater than 1 atomic % P.

Structure development in the cooling meteoritedoes not start until below 750° C in the low P case(Table 13), taenite being the only phase present athigher temperatures. Schreibersite precipitationdoes not begin until below 650° C, a temperatureat which more than 70% of the structure hastransformed to kamacite. At 600° C, only 0.3%

Authors Copy

NUMBER 21 45

schreibersite is present. Under these conditionsschreibersite would have precipitated at kamacite-taenite interfaces throughout a well-establishedWidmanstatten pattern. P could not move throughthe Widmanstatten pattern to precipitate as schrei-bersite borders at isolated sulfide interfaces.Schreibersite in these meteorites must have precip-itated initially prior to extensive Widmanstattenpattern development, requiring a greater bulk Pcontent.

The 1.8 atomic % P value used in Table 13 issomewhat higher than necessary to account forthe observed structure. With this amount of P inthe system, kamacite would precipitate by 975° Cand schreibersite would be present by 925° C. Ifthe P content were reduced to 1.4 atomic %,neither kamacite nor schreibersite would be pres-ent in the system at 925° C or above. At 875° C,however, schreibersite would be present growingwithin taenite. Kamacite would not nucleate untilbelow 750° C. The important consideration here isthat within a reasonable range of P contents,schreibersite nucleates within taenite, undoubtedlyat preexisting taenite interfaces with troilite ortroilite-silicate-graphite inclusions if they are avail-able. Major quantities of schreibersite grow underthese conditions prior to kamacite nucleation andthe onset of Widmanstatten pattern development.The 1.8 atomic % P value is used in Figure 19,and it can be seen there that reasonable reductionin the amount of P changes only the amounts ofthe phases present. Nucleation temperatures forboth schreibersite and kamacite would be loweredsomewhat, and smaller amounts of P would pro-duce smaller quantities of schreibersite. Neitherof these meteorites would appear to contain asmuch as 6% schreibersite, but certainly they bothcontain more than 1 %.

The structural development sequence in Lexing-ton County and Bahjoi is similar to that of SantaLuzia. With falling temperature a few massiveschreibersites grow within taenite, followed bynucleation and growth of a swathing kamacitezone. The Ni values in these massive schreibersites(16 atomic % Ni for Bahjoi, 21 atomic % Ni forLexington County) require that Ni enrichmentcontinued to below 550° C. At some temperaturebelow 650° C, however, massive schreibersitegrowth is stopped by precipitation of cohenite.Cohenite contains no detectable P and, therefore,isolates the massive schreibersite from a source of

P for continued growth. Ni gradients within thecohenite borders (Figure 13) suggest that Ni maycontinue to be supplied to the schreibersitethrough the cohenite. Under these circumstances,a distinct swathing zone morphology is not devel-oped, but the surrounding kamacite does retain alow Ni level and is comparatively free of later-stage schreibersite precipitation.

Within the bulk metal surrounding these even-tual cohenite-schreibersite inclusions, initial Wid-manstatten pattern development started above700° C, following its normal sequence. Bulk Ni islow enough that a coarse-structured octahedritepattern formed. Both of these meteorites haveschreibersite morphologies and distributions simi-lar to a meteorite like Ballinger, but they containmore grain boundary taenite and residual taenite-plessite areas, particularly Bahjoi.

GOOSE LAKE AND BALFOUR DOWNS. —The GooseLake and Balfour Downs meteorites are two ofthe highest Ni members of Group I (Table 1), andthey are meteorites that contain significantamounts of C and S. The analytical P values fromthe literature again appear to be too low to explainthe observed metallography. A planometric esti-mate of 0.9 weight % P (1.6 atomic % P) for GooseLake was given by Doan and Goldstein (1969), andthis figure appears to be consistent with the ob-served structure. Phase calculations are given inTable 14 using a pair of compositions to representthe two meteorites, 7.9 atomic % Ni (8.3 weight %Ni), and 1.4 and 2.7 atomic % P (0.75 and 1.5weight % P), bracketing what was undoubtedlytheir effective bulk compositions.

When 1.4 atomic % P is assumed (Table 14),schreibersite nucleates within taenite at a temper-ature above 850° C. If 2.7 atomic % P is assumed,schreibersite nucleates above 975° C in the pres-ence of 20% kamacite. No structural evidence forinitial precipitation at kamacite-taenite interfaceshas been observed, supporting the suggestion thatthe actual P level was in the neighborhood of 1.6atomic % P. This means that schreibersite nu-cleated around 900° C, undoubtedly at taenite-troilite boundaries, were they available. The resultwas growth of a large amount of schreibersite intaenite, with nucleation of a kamacite swathingzone following at a temperature somewhat above700° C. As the temperature fell, a large supply ofNi was required to keep these massive schreiber-sites in equilibrium, a process that seems to have

Authors Copy

46 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

TABLE 14. —Equilibrium cooling of composi-tions appropriate to the Goose Lake and Bal-four Downs meteorites (percentage of phasespresent at indicated temperatures with theirNi contents)

°c

995975925875850750700650600550

°C

995975925875850750700650600550

1

%a

25477092

2

%a

2023

25477091

.4 Atomic

%y

10010010010099987250264

.7 Atomic

*Y

807596959392674421

% P, 7

XPh

123344

% P, 7

%Ph

245788999

.9 Atomic

XN1a

4.55.05.27.1

.8 Atomic

%N1a

6.26.3

4.55.05.27.0

% N1

XN1Y

7.97.97.97.97.98.09.1111418

% N1

%N1Y

8.28.37.87.97.97.99.11114

*NiPh

6.67.07.19.01316

%N1Ph

6.05.76.06.26.67.19.01316

continued down to below 550° C (17 to 24 atomic% Ni in cohenite-enclosed massive schreibersite).The size of the massive schreibersites requires thatthe cohenite borders precipitated at temperaturesbelow 750° C. Within this same 700° to 550° Ctemperature range, the bulk of the metal awayfrom the large schreibersites transformed to nor-mal Widmanstatten pattern. Grain boundaryschreibersite, rhabdites, and taenite border schrei-bersites were formed during this part of the se-quence.

In Figure 19 the 2.7 atomic % P value is used. Itis obvious here that the same sequence of phasesis obtained if the P value is somewhat reduced.The higher Ni is the significant structural devel-opment factor in this case.

BELLSBANK. —The Bellsbank meteorite is anom-alous, unusually low in Ni and high in P. Thegross segregation of schreibersite in its structuremakes it unusually difficult to arrive at a satisfac-tory bulk P content. For this reason the literaturevalues of 5.3 weight % Ni and 2 weight % P werebracketed in Table 15 by assumed compositions of5.3 weight % Ni and both 1 and 3 weight % P (5.0atomic % Ni and 1.8 and 3.6 atomic % P, and 4.9atomic % Ni and 5.3 atomic % P). The intermedi-

ate P level proved to be the most reasonable interms of explaining the observed structure andwill be considered first.

For the composition 5.0 atomic % Ni and 3.6atomic % P (Table 15), kamacite is the only equilib-rium phase present from 1100° C to below 975° C.At a temperature somewhat below 975° C taenitenucleates, and a small amount of it is in equilib-rium with kamacite at 925° C. By 875° C, taenitehas grown to be nearly as abundant as kamaciteand the structure contains 6% schreibersite, un-doubtedly having nucleated at preexisting kama-cite-taenite grain boundaries as well as aroundsulfide inclusions. At 700° C most of the taenitehas disappeared, and at 650° C, 90% kamacite isin equilibrium with 10% schreibersite. The schrei-bersite continues to increase in Ni as it cools,reaching a value of 11 atomic % Ni at 600° C andbelow, with a Ni content in the kamacite of 4.2atomic percent. Both of these Ni values are ingeneral agreement with observed values, perhapsa little on the high side.

If the 1.8 atomic % P value is assumed, kamaciteand taenite are the phases present in the equilib-rium structure from 1100° C down to below 850°C. Schreibersite first appears somewhat above 750°C, and is present in the amount of 2% when thattemperature is reached. Nucleation under thesecircumstances would be expected to distributeschreibersite along kamacite-taenite borders, thesephases being present in approximately equalamounts. Taenite disappears from the structurebelow 700° C, and by 600° C, 3% schreibersite is inequilibrium with kamacite. The final Ni contentsof both kamacite and schreibersite are somewhathigher than in the 3.6 atomic % P case, the higherP situation agreeing better with actual measure-ments on the meteorite.

When 5.3 atomic % P is assumed, the bulkcomposition lies within the kamacite + liquid fieldof the phase diagram from 1100° C to somewhatabove 1010° C. At 995° C, 5% schreibersite is inequilibrium with kamacite, and this schreibersiteremains constant in amount down to 925° C. Dur-ing the next 50° drop in temperature one-third ofthe structure transforms to taenite and the amountof schreibersite more than doubles to 13%. Thetaenite decreases in volume with further coolingand disappears below 750° C. By 700° C the equilib-rium phases are again kamacite and schreibersite,

Authors Copy

NUMBER 21 47

TABLE 15. —Equilibrium cooling of composi-tions appropriate to the Bellsbank meteorite(percentage of phases present at indicatedtemperatures with their Ni contents)

°c

995975925875850750700650600555

°C

995975925875850750700650600550

°C

995975925875850750700650600550

1

%a

10303035556888979695

3

%a

1001008553506385908888

5

U

95959552496283838180

.8 Atomic

%y

9070706545309

.6 Atomic

%Y

154142285

.3 Atomic

%y

353622

X P, 5

%Ph

23345

% P, 5

%Ph

68910101212

X P, 4

XPh

55513151617171920

.0 Atomic

%N1a

4.04.23.83.54.04.04.54.94.74.6

.0 Atomic

SN1a

5.05.04.74.04.04.04.54.64.24.2

.9 Atomic

%Nia

4.84.84.84.04.04.04.44.43.83.8

% Ni

%N1Y

5.15.35.55.86.37.29.1

% N1

*N1Y

6.66.46.37.29.1

% Ni

SN1Y

6.46.37.2

«H1 p h

5.37.18.91213

*NiPh

4.44.45.37.18.01111

%Ni p h

6.46.05.24.44.45.37.07.51010

schreibersite being present in the amount of 17%.With further cooling, schreibersite increases inquantity and in Ni content, with a rather markeddecrease in kamacite Ni content. The final Nicontent of the schreibersite is possibly slightlylower than the measured value, the quantity ofschreibersite seems too high, and the kamacite Nivalues too low. The 3.6 atomic % P figure, theone plotted in Figure 19, seems to be the bestchoice of the three. A somewhat higher P value,perhaps 4 atomic %, would probably be evenbetter, however. This would mean that schreiber-site would initially nucleate in the presence of lesstaenite, and that final massive schreibersite andkamacite would have slightly smaller Ni values.The observed low temperature schreibersite mor-phology suggests that an extensive Widmanstattenpattern was not present when schreibersite nu-cleated .

The phase growth calculations that have beendiscussed above account for the major structuralfeatures of the selected group of meteorites. Ap-propriate compositions cooled under equilibriumconditions to approximately 600° C would producethe observed proportion of phases and accountfor the general character of the structures. Coa-huila is single crystal kamacite containing a fewisolated moderate-size schreibersite inclusions.Ballinger has a coarse Widmanstatten pattern in-terrupted by large schreibersites that may be sur-rounded by cohenite. Santa Luzia contains massiveschreibersite areas surrounded by giant swathingkamacite, separating areas of coarse Widmanstat-ten pattern. The higher Ni members of the grouphave a finer Widmanstatten pattern with massiveschreibersite morphology explainable in terms ofthe amounts of P and C present. Bellsbank con-tains massive schreibersite and low Ni kamaciteand has similarities to the giant swathing kamaciteareas in Santa Luzia.

It is interesting at this point to summarize equi-librium data given above and compare it withwhat would be expected in terms of Widmanstat-ten pattern growth from the binary Fe-Ni system.Here Widmanstatten pattern growth is construedin its conventional usage of simply taenite trans-forming to kamacite, ignoring the presence of Peither in solution or in schreibersite. The compar-ison is made in Table 16, where the first twocolumns give the compositions of the meteoritesas plotted in Figure 19, the compositions that bestfit the structural arguments given above. Thesecolumns are followed by sets of data for theternary system and for the binary system (Figure1). In each set the first column gives equilibriumtemperatures at which Widmanstatten pattern de-velopment starts (y —>a + y) if undercoolingeffects are ignored, probably a reasonable assump-tion when P is present. The second column givesthe expected temperatures at which taenite wouldbe completely transformed to kamacite (a + y —»a) if equilibrium were maintained down to theselow temperatures. The third column gives theexpected temperature range for Widmanstattenpattern growth (ATWPG). The last two columnsshow the effect of P on increasing the tempera-tures of kamacite nucleation (TaNuc) and taenitedisappearance (TyDis) over what would be ex-pected from the binary system.

Authors Copy

48 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES*

TABLE 16. —Temperatures (° C) for Widmanstatten pattern development in the ternarysystem as compared to the binary system

Meteorite

Bellsbank

Lexington County& Bahjoi

Goose Lake& Balfour Downs • •

Atomic

Ni

5.0

5.3

6.2

6.2

7.1

7.8

%

P

3.6

0.8

1.4

2.7

1.8

2.7

Fe-NI-P System

(-930°)"

860°

880°

840°

750°

730°

c+Y-a

* 690°

610°

580°

580°

565°

555°

A T W G P

240°

250°

300°

260°

185°

175°

Fe-N1

Y-a+Y <»

775°

765°

755°

755°

740°

725°

System

+Y+*

610°

580°

570°

570°

480°

ATWGP

165°

185°

185°

185°

260°

AT due

Sue(155°)**

95°

125°

85°

10°

5°

to P

T Y D 1 S

80°

30°

10°

10°

85°

* For this composition WPG starts with the reaction** Nucleation of Y within a.

The presence of P in the coarsest structuredmeteorites, Bellsbank through Santa Luzia, signifi-cantly increases the temperature at which Wid-manstatten pattern growth starts, increases thetemperature at which transformation of taenitewould be expected to be complete, and also signifi-cantly increases the temperature range over whichgrowth takes place. These three factors all workin the direction of producing larger kamacite crys-tals than would be expected from the binary dia-gram for the same cooling rate. In the coarse-structured meteorites, Lexington County throughBelfour Downs, those with a well-developed Wid-manstatten pattern, the amount of P has a lessdramatic effect. The important point here is thatkamacite nucleation takes place at significantlylower temperatures than is the case for meteoriteswith less Ni. The nucleation temperature is alsovery close to that which would be expected fromthe binary diagram. This leads to much morerestricted conditions for kamacite growth and theretention of larger amounts of taenite and plessitein the structure, thereby producing the finer andmore regular Widmanstatten pattern that is ob-served in these meteorites.

Details of these various structures as given inthe results section, however, are not explainedsatisfactorily by this discussion. Taenite and ples-site are retained in significant amounts in most ofthese meteorites, demonstrating that equilibriumcooling did not extend to 550° C. Diffusion con-trolled growth, modifying the equilibrium struc-ture in significant ways continued to lower temper-atures. It is this aspect of meteorite structure thatwill be discussed below.

Low TEMPERATURE PHASE GROWTH AND THEEQUILIBRIUM DIAGRAM

The preceding section has shown that the majorstructural features of the selected meteorites maybe explained on the basis of nucleation and growthof phases in a system responding to changingequilibrium conditions with decreasing tempera-ture. Bulk equilibrium may have pertained downto 550° C or below for Bellsbank and Coahuila,but the other six meteorites were not in completeequilibrium at that temperature (Figure 19). Hadthey been, their structures would consist only ofkamacite and schreibersite, and the taenite thatthey contain would have transformed. All eight ofthe meteorites also contain Ni and P concentrationgradients in kamacite adjacent to schreibersite (Ta-bles 2-9), although the recent work of Randich(1975) suggests that these gradients formed atmuch lower temperatures. Some of the massiveand all of the smaller schreibersites have Ni con-centrations that are greatly in excess of thosepermitted for schreibersite in equilibrium withkamacite at 550° C (Figures 18, 25). These obser-vations are explained by the fact that as tempera-ture drops the atomic mobility necessary to main-tain bulk equilibrium is lost. Elemental diffusionbecomes slower and compositional changes takeplace over much shorter distances, resulting incomposition gradients. Kinetic factors must nowbe considered as well as equilibrium states. Uponcooling under these restrictive.conditions, schrei-bersite continues to grow in volume, may increasein Ni content, and late-formed schreibersites nu-cleate and grow. Taenite shrinks in volume andincreases in Ni, while kamacite decreases in Ni.

Authors Copy

NUMBER 21 49

Phase interfaces are still in equilibrium in thesense that interface compositions represent tielines within the Fe-Ni-P system (see below). Schrei-bersite-kamacite interface measurements on mete-orites, therefore, give equilibrium information rel-evant to the Fe-Ni-P systems at temperatures wellbelow those that are available experimentally.Combining this low temperature equilibrium datawith both an analysis of the kinetic record presentin the form of diffusion gradients and a detailedknowledge of specimen petrography may be ex-pected to provide new possibilities for the elucida-

t ion of the structures at hand.With these ideas in mind, it will be helpful first

to examine the Fe-Ni-P diagram asking whatchanges would be expected with decreasing tem-perature. Meteorite data (Reed 1965a, 1967, 1969,and Tables 2-9 above) indicate major changes inthe diagram on cooling to low temperatures. Thetop part of Figure 20 may be used as a startingpoint for a discussion of this transition. It is aschematic isothermal section of the ternary systemat a temperature slightly below 550° C. The scaleon the Fe-P axis has been interrupted so that thelower part of the diagram could be enlarged forclarity. The low P concentrations that are observedin kamacite and taenite make it impossible to showthe three fields along the Fe-Ni axis and the threeschreibersite-containing fields using the samescale.

Compositional data on schreibersite-kamaciteand schreibersite-taenite associations suggest thefollowing trends in the ternary diagram with de-creasing temperature. The constant P content ofschreibersite requires that the schreibersite com-position corner of the three-phase field (point Din Figure 20) lie on the 25 atomic % P line regard-less of temperature. Ni values for schreibersiteembedded in taenite borders, an association thatapproximates the schreibersite corner of the three-phase field, range from 41 to 45 atomic %. PointD, therefore, must move at least that far to theright with decreasing temperature. Higher Ni con-tents in schreibersite have been measured (Reed,1972), but these represent compositions in thetaenite-schreibersite field, while lower Ni contentsrepresent compositions in the kamacite-schreiber-site field. Because of the steepness of the Niconcentration gradients, no serious attempt hasbeen made to obtain taenite-schreibersite interface

compositions, but incidental measurements (Fig-ures 9, 11) may be used to bracket the position ofpoint C. The Ni value in taenite associated withschreibersite is less than the schreibersite Ni value,perhaps by about 5 atomic %. This suggestion issupported also by Reed's (1972) measurements onthe Octibbeha County meteorite and by similarmeasurements by the authors on this meteorite(60 atomic % Ni in schreibersite in contact with 55atomic % Ni in taenite). Point C moves to theright, but trails behind point D. The P values intaenite at these interfaces are also not accuratelyknown, but sufficient data is available to say thatthey are a fraction of the P value in kamaciteassociate with the same schreibersite (see discus-sion below and Figure 22). The Ni and P contentsfor point B may be estimated from the kamacite-schreibersite interface values of these taenite-embedded schreibersites. They range from 5.7 to6.2 atomic % Ni, with constant P values of 0.05atomic %. In summary the diagram changes asfollows with decreasing temperature: point Dmoves to the right to approximately 45 atomic %Ni with point C moving to approximately 40 atomic% Ni and to a P value that is much smaller than0.05 atomic %. As a consequence the a + Ph anda + y + Ph fields greatly expand in area mainlyat the expense of the y + Ph field. The threesmall fields along the Ni axis also lose a largeproportion of their areas, but they still retainconsiderable importance for the interpretation ofmeteorite structures.

The preceding illustrates the direction and ex-tent of change in the ternary system with decreas-ing temperature, but it says nothing about whatthat temperature range might be. Do meteoritestructures continue to develop to very low temper-atures or is growth effectively concluded at tem-peratures not far below 550° C? The bottom sectionof Figure 20 suggests an approach that may beuseful here. We know that D' lies on the 25 atomic% P line and can assume that it moves continuouslyto the right with decreasing temperature. Thepoints on the Fe-Ni axis can be obtained from theestimated portion of the Goldstein and Ogilvie(1965) binary Fe-Ni diagram down to about 350°C. If A', B', and C could be estimated as afunction of temperature, we would have much ofthe information needed for extrapolated isother-mal sections.

Authors Copy

50 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

0100 %Fe

20 30ATOMIC %Ni

40

C'

20ATOMIC % Ni

30 40

FIGURE 20. —Fe-Ni-P system below 550°C: top, schematic isothermal section; bottom, parameters ofuse in extrapolating to lower temperatures.

Authors Copy

NUMBER 21 51

TABLE 17. —Experimental values of P concentration inkamacite and taenite in equilibrium with schreibersite

TABLE 18. —Temperature versus P saturation concentra-tion in kamacite and taenite taken from Figure 22

Fe-P System* Fe-Ni-P System**

T°C

1010995975925875850750700650600550

1283126812481198114811231023973923873823

1/T ("KxlO"4)

7.797.898.018.358.718.909.7810.310.811.512.1

Atonic % Pa

4.84.1

2.9

1.71.30.990.590.45

Atomic % Pa

4.13.73.02.81.81.20.90.570.47

Atomic * PY

1.91.61.31.20.680.490.270.220.15

* Doan and Goldstein (1970), table IV.** Doan and Goldstein (1970), table I I I .

A', B', and C represent corners of compositionfields, and these particular compositions are verydilute solutions of P in Fe or in Fe-Ni. Data forthese points at a number of temperatures withinthe experimental range are given by Doan andGoldstein (1970), and their values converted toatomic % P are given in Table 17. This type ofdata is frequently used to extrapolate concentra-tion-temperature relationships to lower tempera-tures and is plotted accordingly in Figures 21 and22. Atomic % P on a logarithmic scale is plottedagainst 1/T. Figure 21 uses P saturated kamacitevalues from the Fe-P system, and Figure 22 usesboth P saturated kamacite and taenite values fromthe Fe-Ni-P system. Straight lines were drawn withease through the three sets of data, suggestingthat the extrapolations may give useful low tem-perature composition estimates. The thermody-namic rationale for this procedure is given in"Appendix."

A particularly interesting observation based onFigures 21 and 22 is that the kamacite line in theplot of the Fe-Ni-P system data is identical to thekamacite line in the plot of Fe-P system. There-fore , both sets of data yield the same extrapolatedP-temperature relationship for kamacite. Thismeans that line A'B' is parallel to the Fe-Ni axis,and that the saturation value of P in kamacite isindependent of Ni concentration. Should this ac-tually prove to be the case, it would have interest-ing consequences for the interpretation of mete-orite structures. Kamacite-schreibersite interfacevalues would not only represent tie lines in an a +Ph field of the ternary diagram, but the P value inkamacite at a specific interface would representthe temperature of the isothermal section to which

T°C

540527508488473465460454449444436431424417410403394383372362348333315298252

Atomic % Pa

0.400.350.300.250.220.200.190.180.170.160.150.140.130.120.110.100.0900.0800.0700.0600.0500.0400.0300.0200.010

Atomic X PY

0.140.120.100.0800.0700.0620.0580.0540.0520.0480.0460.0410.0380.0340.0320.0280.0250.0220.0190.0150.0130.0090.0070.0050.002

Weight % Pa

Q.ZZ0.190.170.140.120.110.110.100.090.090.080.080.070.070.060.060.050.040.040.030.030.020.020.010.006

that tie line belonged. The implication of thiswould be that schreibersite growth effectivelystopped over a sequence of temperatures within agiven meteorite. Numerical values for P concentra-tion variation with temperature for both kamaciteand taenite were read from Figure 22 and arelisted in Table 18.

Although a flat line A'B' has several interestingimplications, it is necessary to consider as an alter-nate possibility that it actually slopes downwardfrom left to right. The experimental techniqueemployed by Doan and Goldstein (1970) wasstretched to its limit in obtaining the lower temper-ature data, resulting in the possibility of significanterror for the problem at hand. There is alsoambiguity in their paper concerning one criticalvalue. The weight % P in kamacite value for theFe-P system at 550° C taken from their table IV is0.25 ±0.03. The value used on the 550° C isother-mal section (their figure 8) is 0.4 weight % P. Thisdifference probably represents confusion as towhich value is best rather than a simple misprintin the diagram, a difficulty that can best be re-solved by more careful experimental work. If this0.4 weight % P value had been used in the extrap-olation, the line in Figure 21 would have lain tothe right of the kamacite line in Figure 22. Thiswould mean a sloping line A'B' with P decreasingas Ni increases. This situation could imply that allschreibersite within a given meteorite grew downto the same final temperature. Before attempting

Authors Copy

52 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

.0017 8 9 10 II

FIGURE 21. —Extrapolation of P saturation values in kamacite in Fe-P system to lower temperatures.

12 13 14 15 16 17 18 19 20 21I/T ( ° K x l O ~ 4 )

Authors Copy

NUMBER 21 53

10 II 12 13 14 15 16 17 18 19 20 21001

FIGURE 22. —Extrapolation of P saturation values in kamacite and taenite in Fe-Ni-P system tolower temperatures.

Authors Copy

54 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

to resoive this and related problems in terms ofmeteorite measurements, however, it will be help-ful if we first look at the diffusion process as itapplies to schreibersite growth.

DIFFUSION-CONTROLLED SCHREIBERSITE GROWTH

The data presented above on Ni and P gradientsdemonstrates that schreibersite grew by drawingthese elements from increasingly restricted vol-umes of kamacite with decreasing temperature.Petrographic considerations combined with an un-derstanding of the Fe-Ni-P system suggest thebroad outlines of the nucleation and growth proc-ess. With this background in mind, a set of con-straints and assumptions can be developed thatallow the qualitative application of the nonisother-mal diffusion-controlled growth theory for ternarysystems recently developed by Randich and Gold-stein (1975) to be applied to schreibersite growthin these meteorites (see also Randich, 1975).

Figure 23 may be used to relate schreibersitegrowth both to the Fe-Ni-P diagram and to growthequations. The top part of the diagram is a sche-matic isothermal section at some temperature be-low 550° C. The scale for P is broken so that verylow levels of P can be indicated. The tie linerepresents schreibersite growing within kamacite.Its ends give interface P and Ni compositions forboth kamacite (CP, CNi) and schreibersite (Cp,C^i). The theory of Randich and Goldstein (1975)requires that equilibrium be maintained to theextent that Ni and P concentrations in the twophases are related by a tie line in the ternarysystem at the temperature that pertains. For dif-fusion-controlled conditions, this tie line need notbe the equilibrium tie line (ETL), the one thatpasses through the bulk composition of the system.

The lower part of Figure 23 indicates kineticfactors that are important in the growth process.A schreibersite (Ph) of half width £p" equal to £m isgrowing within kamacite in the X direction. The Pand Ni values in schreibersite at the interface areCP and CNi, and the equivalent values in kamaciteare CP and CNi . The P and Ni fluxes from kamaciteinto the growing schreibersite are J£ f

+ and J$U(+ >

and the Ni flux from the interface into the schrei-bersite is Sm.f • The concentration gradients in Pand Ni are dCP/dX and aCNi/aX.

The diffusion controlled growth process may be

described by the mathematical treatment of Ran-dich and Goldstein (1975). Diffusion in the Fe-Ni-P system may be described using an extension ofFick's first law. The P and Ni fluxes in terms ofconcentration gradients in kamacite and schreiber-site are as follows.

T a = —JP

T a — _

pp dXPNi

NiNidX

dX

ax

TPh= _r>Fe,Ph

NiNiax

,ph aCp

ax

Jp"h = o

(la)

(lb)

(lc)

(Id)

Separate sets of equations are needed for eachphase. The P flux in schreibersite is zero due tothe constant atomic proportion of this element inschreibersite. The symbols D£pa and D£fN? arediffusion coefficients, measures of the influenceof the concentration gradients of these two ele-ments on their own flux, where as D£Nia and D iP

a

reflect cross effects usually referred to as theternary diffusional interaction. The latter coeffi-cients have been found to be so small in thissystem that the terms containing them may bedropped from the above equations (Heyward andGoldstein, 1973). These equations may then besimplified by neglecting the diffusional interactionterms and the P flux in schreibersite.

J a —P -

Hx

ax

(2a)

(2b)

(2c)

Fick's second law is applied in order to considerthe time dependence of the fluxes. It is assumedthat diffusion coefficients are independent of com-position.

at ax ppax2 (3a)

Authors Copy

NUMBER 21 55

ATOMIC %Ni

30

20

o

o

10

Ph

Ni

J P t+ aCp

x ^ HFIGURE 23. —Relation of schreibersite growth both to Fe-Ni-P diagram and to growth equations:top, schematic low temperature ternary with tie line indicating schreibersite growing in a + Phfield; bottom, a + Ph interface compositions, composition gradients, and associated P and Nifluxes.

Authors Copy

56 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

dt

at

axJNl

ax

NiNi

NiNiax2

(3b)

(3c)

The resulting mass balance equations for P andNi at a schreibersite-kamacite interface are as fol-lows.

(4a)dt

(4b)

The two mass balances are related to each otheras the rate of movement of the interface, d£/dt,must be the same for both elements.

(5)dt dt dt

A schreibersite growth rate may then be expressedin terms of interface Ni and P concentrations, theflux of P in kamacite, and the fluxes of Ni in bothkamacite and schreibersite.

1dt Cp — Cp

7(-JPV)

dt

(6a)

(6b)

It is beyond the scope of this study to attemptthe difficult numerical analysis required to solvethese growth equations. Randich (1975) has devel-oped a model for this purpose and successfullyapplied it to the growth of schreibersite in hexa-hedrites. He compiled a computer program thattreats diffusion controlled phase growth in ternarysystems when the requisite portions of the phasediagram are known. The model accomodates ter-nary interactions, nonisothermal transformations(variable cooling rates), and impingement (over-lapping diffusion fields). It is based on a one-dimensional space grid and is rigorous only forlamellar schreibersite growing within kamacite.The concepts that this model employs, however,are valid for meteorites more complex than thehexahedrites, and they will be used here in aqualitative interpretation of the data.

Equations 6a and 6b relate P and Ni interfaceconcentrations to their gradients in kamacite and

schreibersite for the final stages of growth. Cool-ing is assumed to be uninterrupted, with schreiber-site growth continuing down to very low tempera-tures. Measurements with the electron micro-probe, however, do not reveal changes that takeplace within less than 1 /urn of an interface. There-fore, as a consequence of this limitation of themeasurement technique, growth apparently stopswhen subsequent changes become unresolvable.Randich's (1975) calculations indicate a lower limitof 250° C for detectable change.

Equation 6a is a key to the understanding ofschreibersite growth under the conditions thatpertained in the meteorites under study. It dem-onstrates that the P flux controlled the schreiber-site growth rate. The quantity Cp in the equationis the P content of schreibersite at the interface(Figure 23), a constant at 25 atomic %. The quan-tity Cp is the interface P content in kamacite. It isa small number, 0.2 to 0.05 atomic %, in thetemperature range of formation of the observedinterface relationships. The quantity (CP — CP),therefore, may be considered to be independentof temperature even though Cp is not. With thissimplifying assumption, the rate of growth of agiven interface, d£/dt, is seen to be dependentonly on the P flux, J£ f

+ . The P flux, however,will be dependent upon CP at any given tempera-ture. The value of Cp combines with the P level inthe surrounding kamacite to establish the P gra-dient, aCp/aX, the variable that determines theflux (equation 2a).

The system responds to the constraint of the Pflux determining schreibersite growth rates byadjusting individual interface Ni concentrations sothat the growth rate established by P (equation 6a)is matched by the growth rate due to Ni (equation6b). This permits individual schreibersites contain-ing markedly different Ni levels to continue togrow simultaneously within the cooling meteorite.Kamacite and schreibersite interface Ni and Pconcentrations must continue to be related by a tieline relationship, a requirement that is met by tielines shifting from the equilibrium tie line. Thenature of these tie line shifts and their effects ongrowth rates of individual schreibersites is illus-trated in Figure 24.

Tie line shifts and their effects are illustratedusing hypothetical ternary diagrams in Figure 24.The symbols for Ni and P are retained as a matterof convenience, although scale problems preclude

Authors Copy

NUMBER 21 57

1

11

^ 1LJ 1

1

11

\ V

liHOJu3

<Ua.EV

ese

a

ao

d -10 IN % DIW01V !N % OIW01V

2 "°C ca

- "I3

C a,•2 E

"rt OHu'3 +•S «

8.S

!i• * . y

" .S

o a a

Authors Copy

58 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

use of actual values. The tie line indicated at thehigher temperature Ti, ETL(TX), is assumed torepresent bulk equilibrium in the a + Ph regionof the diagram. It passes through the bulk com-position indicated by the small circle (5.5 atomic %Ni and 6.0 atomic % P). Upon cooling to tempera-ture T2 the a + Ph field expands along the a +Ph/Ph boundary, and the a/a + Ph boundarymoves closer to the Fe-Ni axis. It is assumed thatin this lower temperature range mass transfer isinsufficient to achieve bulk equilibrium. Growthbecomes diffusion controlled and gradients de-velop within the phases. Tie lines may thereforeshift from the equilibrium tie line so that the Nigrowth rate can match the P growth rate. Threepossible tie lines are indicated at T2, the equilib-rium tie line for the same composition used above,ETL(T2), and tie lines shifted to the right and tothe left of ETL(T2),

The interface relationships for these four tielines are illustrated at the right in Figure 24. Theupper diagram shows interface Ni and P composi-tions for ETL(Tj). The bulk Ni value for thesystem is indicated by Co1- The bulk P value is 0.5atomic % higher and was omitted for clarity. Bulkequilibrium has been achieved at temperature Tjas is indicated by the absence of compositionalgradients in both kamacite and schreibersite.

Interface relationships for diffusion controlledgrowth at temperature T2 are illustrated in thelower right side of Figure 24. The P interfacevalues are assumed to be identical in the threeinterface situations shown. The consequence issimilar P fluxes feeding schreibersite growth,]P,(+, and the establishment of similar growthrates, d£/dt, for the interfaces through equation6a. Tie line shifts from ETL(T2) induced by con-ditions of limited mass transfer produce a contin-uously varying range of growth rates that permitsNi flow to match that required by P.

The equilibrium tie line at the lower tempera-ture, ETL(T2), passes through the bulk composi-tion. Its interface Ni and P values are those thatwould be observed if the system were at completeequilibrium. The presence of compositional gra-dients in both kamacite and schreibersite, how-ever, show that complete equilibrium has not beenreached.

Shifting the tie line to the left results in anincrease in the growth rate due to Ni (equation6b). The quantity 1/(CNI — CNI) and the Ni flux in

kamacite, ]xU(+, become larger and the Ni flux in

schreibersite, J^,f~. becomes smaller. All three ofthese changes increase the growth rate over thatof ETL(T2). Ni is supplied to the growing schrei-bersite in a way that size is increased at the expenseof Ni enrichment.

Moving the tie line to the right of ETL(T2) hasthe effect of slowing the growth rate due to Ni.The quantity 1/(CNI ~ C i) and the Ni flux inkamacite, ]yiti

+, become smaller, and the Ni fluxin schreibersite, J^.f » becomes larger. All threeof these changes decrease the growth rate overthat for ETL(T2). Ni is supplied to the growingschreibersite in a way that it becomes enriched inNi rather than growing in size. A theoreticallyinfinite range of growth rates may be obtained bythese tie line shifts. One of these possible tie lineswill give a unique growth rate that satisfies bothequations 6a and 6b.

COOLING RATE VARIATIONS

Cooling rates of meteorite parent bodies areknown to vary over several orders of magnitude(Goldstein and Short, 1967b). These variations willaffect the growth of schreibersite, particularly inthe lower temperature ranges. Compositions andassociations we observe in meteorite specimenswill depend in part on specific cooling histories.Measurements of schreibersite, therefore, wouldbe expected to be interpretable in terms of coolingrates. Randich (1975) has recently demonstratedthe validity of this assumption by calculating cool-ing rates for hexahedrites. In the course of hiswork important observations were made that willhelp in the interpretation of more complex mete-orite structures. Several aspects of this work willbe reviewed here.

Calculations using the Randich and Goldstein(1975) schreibersite growth model were made us-ing a bulk composition of 2.1 weight % P, 4.1weight % Ni, and 93.8 weight % Fe. Cooling ratesof 5 X 10~3, 5 x 10~4, and 5 X 10~50 C per secondwere employed over a cooling interval of 900° C to685° C. Decreasing the cooling rate producedschreibersites of predicted greater widths andhigher Ni contents. P gradients were predicted inkamacite and Ni gradients were predicted for bothkamacite and schreibersite. The calculated Ni in-terface values were significantly lower than thosefor the equilibrium tie line for this composition.

Authors Copy

NUMBER 21 59

Tie lines had shifted to the left to increase thegrowth rate for Ni. An experimental validation ofthese calculations was carried out using this samecomposition and the 5 x 10~4° C per second (43.2°C per day) cooling rate. Calculated and observedcompositions agreed within the limits of electronmicroprobe measurements.

Additional cooling rate calculations were madeby Randich (1975) in a study of schreibersite nu-cleation temperatures. Cooling rates in the mete-orite range of 1°, 10°, and 100° C per million yearswere used with the Coahuila bulk composition of5.49 weight % Ni and 0.49 weight % P. Theresults were similar to those given above, theslower cooling rates producing slightly largerschreibersite containing slightly higher Ni values.The calculated Ni values were within the range ofobserved Ni values in hexahedrites, establishingthe validity of the model at cooling rates withinthe meteoritic range. Ni gradients were predictedfor the schreibersites, contrary to observations onmeteorites. This is probably explained by the factthat the ternary diffusion constant for Ni in schrei-bersite is not known in this temperature range.

These calculations led to a conclusion that issignificant for our purpose. It was shown that ifnucleation occurs above 500° C the amount ofundercooling prior to nucleation in unimportant.The kamacite-schreibersite assemblage at 500° Cwill be in equilibrium, assuming a diffusion fieldof 1000 [im or less. Randich (1975) also points outthat this situation leads to the conclusion thatReed's (1965b) theory that lamellar schreibersiteand rhabdite nucleated simultaneously must becorrect. Had the plate schreibersite grown signifi-cantly prior to nucleation of rhabdite, the rhabditewould have been prevented from nucleating. Bothtypes are observed together over very short dis-tances, requiring that nucleation temperatureswere very close. Under these circumstances, im-pingment of P and cooling rate are the two factorsthat control schreibersite growth for a given bulkcomposition.

INTERFACE DATA AND SCHREIBERSITE

DISTRIBUTION

In a previous section we have seen that themain petrographic features of coarse-structurediron meteorites can be understood in terms ofequilibrium cooling of specific compositions in the

Fe-Ni-P system. This approach can be extendedto much lower temperatures with the aid of con-cepts developed by Randich (1975) in his analysisof schreibersite growth in hexahedrites. The appli-cation of diffusion theory permits the interpreta-tion of schreibersite-kamacite interface data in away that yields a more detailed understanding ofschreibersite distribution and composition thanhas been possible previously.

Phases and compositions resulting from equilib-rium cooling of the meteorite compositions ofinterest have been summarized in Tables 10through 15, and these same compositions havebeen plotted on the iron-rich corner of isothermalsections in Figure 19. At the top of Figure 25these compositions have also been plotted on the550°C isothermal section with their respective tielines. This is the lowest temperature for which anexperimentally based ternary section is available.Table 19 lists the equilibrium percentages of kam-acite and schreibersite with their Ni contents astaken from Figure 25. This plot shows that hadcomplete equilibrium been reached at this temper-ature for these compositions, taenite would havedisappeared and the observed structures wouldconsist of kamacite containing inclusions of schrei-bersite.

The bottom section of Figure 25 contains anestimated isothermal section at approximately 300°C. The schreibersite corner of the three phasefield is plotted at 45 atomic % Ni, a value that isprobably close to the real value. The kamacitecorner of the three phase field is plotted at 5.5atomic % Ni, and this value is considerably lesscertain. The compositions used above are againplotted with their tie lines. Although the diagrammay not be accurate, it does tell us something ofwhat would be expected if equilibrium coolingextended to this very low temperature. Propor-tions of phases and the expected compositionstaken from this diagram are also listed in Table19.

The bulk compositions plotted in the lowersection of Figure 25 would also consist of kamaciteand schreibersite only, and these two phases wouldbe uniform in composition. The Bellsbank schrei-bersite would contain much more Ni than wasobserved for its massive schreibersite, and its kam-acite would be considerably poorer in Ni than wasobserved. The massive schreibersite in Santa Luziawould be greatly enriched in Ni, its small schrei-

Authors Copy

60 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

550° C

y +• P h

too %FE 20 30ATOMIC % Ni

40

ABELLSBANK

• COAHUILA

• SANTA LUZIA

• BALLINGER

• LEXINGTON COUNTY- BAHJOI

A GOOSE LAKE- BALFOUR DOWNS

50

10 20 30ATOMIC % Ni

40 50

FIGURE 25. —Schreibersite compositions resulting from equilibrium cooling of meteorite composi-tions: top, 550° C isothermal section with meteorite compositions and their respective equilibriumtie lines; bottom, similar plot at approximately 300° C.

bersite would contain less Ni, and its kamacitewould be poorer in Ni. The Coahulia, Ballinger,and Goose Lake-Balfour Downs compositions alllie on the same tie line, but would consist ofdifferent proportions of kamacite and schreiber-

site. The Lexington County-Bahjoi compositionwould produce the most Ni rich schreibersite. Thestructures suggested by this diagram, therefore,are markedly different than those actually ob-served in the meteorites under study. At what

Authors Copy

NUMBER 21 61

TABLE 19. —Equilibrium compositions at 550° C

and -300° C taken from Figure 25

Meteorite

BeilsbankCoahuilaSanta Luzia

Lexington County &Bahjoi

Goose Lake &Balfour Downs

Meteorite

BellsbankCoahuilaSanta Luzia

Lexington County &Bahjoi

Goose Lake &Balfour Downs • • • •

550°C

% a % Ph % Nia % Niph

87.0 13.0 4.1 11.999.3 0.7 5.2 13.391.3 8.7 5.4 14.096.3 3.7 5.8 15.0

94.8 5.2 6.6 16.2

91.2 8.8 6.9 17.0

-300°C

% a. % Ph % Nia % Niph

86.4 13.6 2.5 20.897.2 2.8 4.3 38.090.2 9.8 3.6 30.594.5 5.5 4.3 38.0

93.3 6.7 4.7 41.2

90.0 10.0 4.3 38.0

temperatures do these observed differences be-come introduced into the structures?

Taenite is retained in the form of the Widman-statten pattern in the six octahedrites studied.This shows that mass transfer required by thetaenite-kamacite transformation is insufficient atsomewhat above 550° C to achieve bulk equilibriumin the cooling rate range pertaining. The presenceof the Widmanstatten pattern may also affect sub-sequent P movement through the structure. Ran-dich's (1975) calculations show that mass transferthrough 1000 /xm diffusion fields in kamacite issufficient for kamacite-schreibersite equilibrium tobe maintained down to 500° C in the same rangeof cooling rates. Kamacite-taenite transformationsbecome restricted at higher temperatures thankamacite-schreibersite transformations. There-fore, it is reasonable to assume that as tempera-tures drop below 600° C, localized compositionalchanges become increasingly more important andbulk compositional effects become increasingly lessimportant. As mass flow becomes more restricted,individual structural units maintain interface equi-librium and respond to compositional gradients inincreasingly small volumes of material. Theseprocesses are responsible for the interface andgradient data that have been reported above.

Tie lines representing interface data from theCoahuila, Santa Luzia, and Goose Lake meteorites

(Tables 2, 5, and 8) are plotted on isothermalsections in Figures 26. Average values were usedwhere more than one set of interface measure-ments gave similar values. Any given tie line,therefore, may represent more than one set ofmeasurements. The scale used made it impossibleto indicate the differing low P values in kamacite,so all of these compositions are indicated by pointsresting on the Fe-Ni axis. This treatment of thedata may well be an over simplification as varioustie lines within a given meteorite may actuallybelong on different isothermal sections. A rangeof temperatures may be involved, and this pointwill be returned to in a later section. The threemeteorites selected are representative of the groupof eight meteorites studied.

Typical tie lines for the Coahuila meteorite aregiven in the top section of Figure 26. They fanout through a narrow composition range of 3.2 to4.7 atomic % Ni in kamacite and 20 to 30 atomic% Ni in schreibersite. Kamacite P values at theinterfaces range from 0.14 to 0.09 atomic %, asignificant point that will be dealt with later. Thestructural associations related to compositions havebeen discussed above, but it should be noted herethat the tie line on the left represents early-formedstructures (schreibersite bordering troilite-dau-breelite inclusions and a large schreibersite), andthe three on the right are later-formed structures(large rhabdites). The estimated bulk compositionof Coahuila is indicated on the diagram as a largedot. The point apparently falls on the right-handtie line. All the measured schreibersite composi-tions in Coahuila then appear to be to the left ofthe ETL or on it, as was to be expected on thebasis of Randich's (1975) analysis. Reed (1965a)has reported Coahuila schreibersite with 34 atomic% Ni. It is not clear how this value would plot onFigure 26 as the necessary Ni value in kamacite isnot available.

The relationship of the equilibrium tie line toobserved meteorite tie lines is actually somewhatmore complex then Randich (1975) has implied.This can be concluded from consideration of Coa-huila measurements, and it will become clearerwhen measurements on the octahedrites are con-sidered. At temperatures above the final growthstages represented by the Coahuila tie lines inFigure 26, mass transfer had become inhibitedexcept for very short-range interactions. Schrei-bersite continues to grow as long as there is a P

Authors Copy

62 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

COAHUILA5.3%Ni, 08%P

SANTA LUZIA6.2 %Ni, 2.7%P

1 S TATOMIC % Ni

FIGURE 26. —Representative tie lines from Coahuila, Santa Luzia, and Goose Lake meteoritesplotted on isothermal sections.

Authors Copy

NUMBER 21 63

flow to the interface. Growth stops when P imping-ment reduces the P flux to zero, or, for ourpurpose, when we can no longer measure interfacerelationship changes. Although a theoretical pos-sibility, there is no evidence that any of the schrei-bersites studied were in the process of dissolving.As the P flux decreases with decreasing tempera-ture, a given interface tie line would be expectedto shift to the right to conform to the change inshape of the a + Ph field of the phase diagram. Itwould also be expected to move farther to theright in order to slow the Ni growth rate as the Psupply becomes more limited. The largest schrei-bersites undoubtedly nucleated first and havegrown at comparatively high rates, drawing Pfrom large volumes of metal. They have the lowestNi concentrations. The smaller schreibersites thatgrew at slower rates due to a more limited Psupply have higher Ni levels. The smallest schrei-bersites would be expected to approach a Ni con-tent given by the tie line whose Ni value in kama-cite equals the Ni value of the kamacite in whichthey are growing. Kamacite in the Coahuila mete-orite has a higher Ni value than it would have hadif complete equilibrium had been reached. It istherefore possible that microrhabdites in Coahuilacontain more Ni than would be expected from thelow temperature equilibrium tie line. Unfortu-nately, the necessary measurements needed toconfirm this were beyond the capability of thetechniques available.

Tie lines for the Santa Luzia meteorite areshown on a single temperature plot in the centersection of Figure 26. Kamacite Ni values rangefrom 2.2 to 6.2 atomic % with the correspondingNi values in schreibersite of 14 to 42 atomic %covering a much broader range when comparedto Coahuila. Kamacite P values at the interfacesrange from 0.13 to 0.05 atomic %. The three tielines at the left represent the early formed massiveschreibersites, the three to their right are fromgrain boundary schreibersite, and the one on theextreme right is from a late-stage structure, asmall schreibersite partially embedded in taenite.The Santa Luzia bulk composition lies in aboutthe center of the spread of tie lines, demonstratingthat interface tie lines depend upon local structureand may fall on either side of the ETL.

Had complete equilibrium been achieved forSanta Luzia at 550° C, kamacite would have con-tained 5.4 atomic % Ni and schreibersite 14 atomic

% Ni. Equilibrium extending to the 300° C temper-ature range would have resulted in kamacite of3.6 atomic % Ni and schreibersite of 31 atomic %Ni (Table 19, Figure 25). A very different situationis indicated by the tie lines in Figure 26. Theserepresent interfaces from a range of schreibersitemorphologies from massive schreibersite sur-rounded by large volumes of swathing kamacite totaenite border schreibersite.

The explanation of these differences lies in thefact that under conditions of diffusion-controlledgrowth, P can be drawn from much greater dis-tances than Ni within a given time period. Com-plete equilibrium was not reached in the 600° to500° C temperature range. Massive schreibersiteshad nucleated at much higher temperatures andgrown to near their final size. This process hadreduced the level of P in the complete volume ofmetal to near the equilibrium level. Even thevolumes of metal most remote from the largeschreibersites probably only contained about 0.6atomic % P at 550° C (the equilibrium value is 0.4atomic % P). Massive schreibersite growth may beviewed as a process that equalizes P levels inmeteoritic metal in the 500° to 600° C temperaturerange regardless of initial P concentrations. Differ-ences in initial P level are reflected in the amountof schreibersite produced.

While P diffusion in this temperature rangeresulted in a meteorite with a modest P gradient,the slower Ni diffusion resulted in an importantsegregation of this element. Ni could not be movedfrom a sufficiently large distance to maintain bulkequilibrium in the times available. The massiveschreibersites had grown to near their final size,but their Ni contents had to have been significantlybelow the equilibrium value. Their iriterface tielines, therefore, must have been to the left of theequilibrium tie line. The Ni gradient that probablyalready existed within the swathing kamacite zonebecame more pronounced. Beyond the swathingzone, however, sufficient Ni was maintained toyield the observed Wildmanstatten pattern. Theseareas of the meteorite have a higher average Nicontent than the initial bulk Ni content of themeteorite. As a matter of fact, some areas ofkamacite within the Widmanstatten pattern con-tain more Ni than the accepted bulk Ni value.

The consequence of massive schreibersitegrowth under these circumstances of compositionand cooling is to establish two distinctly different

Authors Copy

64 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

structural areas within the meteorite, swathingkamacite areas containing massive schreibersitesand areas of Widmanstatten pattern (Figure 7).Both types of areas continue to change with de-creasing temperature, but distances involved be-come increasingly more restricted. The massiveschreibersites continue to grow, their growth ratesbeing determined by the flux of P from the swath-ing kamacite. Interface Ni values shift to producea tie line with a matching growth rate. The shift isundoubtedly to the right, resulting in a slowergrowth rate due to Ni and Ni enrichment of theschreibersite. The final Ni values of these schrei-bersites approximates the equilibrium value at 550°C.

The outer edge of the swathing zone developsas a kamacite-taenite boundary that recedes fromthe massive schreibersite with decreasing temper-ature. The final stage in this progression is thereplacement of the boundary in part by grainboundary schreibersite. P is supplied to the bound-ary initially by the swathing kamacite, and thetaenite boundary supplies the initial Ni. Nuclea-tion of this type results in concentrations that arerequired by the three-phase field of the ternarydiagram. The initial schreibersite-kamacite tie linewill be the a + Ph/a + y + Ph boundary for thegiven temperature. With decreasing temperaturethe taenite border is replaced with a grain bound-ary that contains grain boundary schreibersite overabout half of its distance. Nucleation may takeplace over a range of temperatures as the taeniterecedes along the forming grain boundary, result-ing in different initial Ni contents in the schreiber-site and different initial growth rates. As growthcontinues at the individual interfaces, Ni concen-trations adjust to match the P controlled growthrate. The general tendency will be for tie lines tomove to the right as the P supply dwindles.

Simultaneously with continued growth of themassive schreibersites and development of theexterior swathing zone grain boundary, Widman-statten pattern development is continuing in theremaining areas of the meteorite. The Ni level ishigh enough in these areas so that a coarse Wid-manstatten pattern forms, containing large kama-cite lamellae and occasional plessite areas andgrain boundary taenite. Grain boundaries containoccasional grain boundary schreibersite, and thekamacite lamellae contain rhabdites and micro-

rhabdites. The rhabdites and grain boundaryschreibersites develop analogously to the similarmorphologies described above for Coahuila, react-ing to specific local conditions to achieve appropri-ate interface values. Nucleation of grain boundaryschreibersites under these circumstances may takeplace at lower temperatures, undoubtedly leadingto smaller schreibersites with higher initial Nicontent that continues to increase upon furthercooling. Small rhabdites would also be expected toreach high Ni values because they are growing ina relatively Ni rich kamacite.

A final type of schreibersite morphology ob-served in Santa Luzia is represented by the tie lineat the far right in Figure 26. It represents thetaenite border schreibersite illustrated in Figures10 and 11. Schreibersite of this type appears tonucleate within taenite at a taenite-kamacite inter-face at a very late stage in the structural develop-ment process. P is undoubtedly supplied mainlyfrom the kamacite, with most of the Ni comingfrom the taenite. Associations of this type yieldthe highest Ni schreibersites to be observed incoarse octahedrites. The schreibersites may alsocontain a slight Ni gradient, undoubtedly due tothe fact that they are associated with kamacite onone side and taenite on the other. If this type ofschreibersite nucleates earlier or grows more rap-idly due to a particularly high P gradient, it isolatesitself from the enclosing taenite by converting it tokamacite.

Goose Lake meteorite tie lines are shown in thebottom isothermal section in Figure 26. Differ-ences from the Santa Luzia pattern result fromhigher bulk concentrations of both carbon andNi. Only the upper portions of the two tie lines onthe left can be indicated. At some point in thedevelopment of these schreibersites, cohenite pre-cipitated in the adjacent metal, transforming theinterface to a cohenite-schreibersite interface andpreventing further schreibersite growth. The par-tial range of Ni values in kamacite, therefore, is3.2 to 5.7 atomic %, with Ni values in schreibersiteranging from 16 to 45 atomic %. The value of 45atomic % Ni is the highest observed for a schrei-bersite in contact with kamacite. Kamacite P valuesat the interfaces cover the same range as observedin Santa Luzia, 0.13 to 0.05 atomic %. This rangein P values may indicate that these tie lines do notall belong on the same isothermal section, a point

Authors Copy

NUMBER 21 65

that will be discussed in a later section. The bulkcomposition again lies in the center of the spreadof tie lines.

The general outline of the schreibersite growthprocess may be summarized on the basis of thisdiscussion and that given in previous sections. Inthe meteorites under study schreibersite petrogra-phy has been strongly affected by sequential nu-cleation. Initial schreibersite nucleation is a func-tion of temperature and bulk composition. It tookplace at greater than 850° C for all of the meteor-ites containing massive schreibersite. The low PCoahuila meteorite is the only exception. Thesemassive schreibersites grew rapidly under equilib-rium or near equilibrium conditions, at least asfar as P is concerned, for most of their growthperiod. Randich (1975) has shown that meteoritecooling rates are sufficiently slow to allow this. Itis important to note that massive schreibersite hasachieved the major portion of its growth by thetime it has cooled to 600° C. On subsequent cool-ing, Ni is enriched but size does not increaseappreciably. The massive schreibersite growthprocess, therefore, has the effect of lowering theP level in the bulk metal and isolating it, asschreibersite enclosed in zones of swathing kama-cite, from areas of developing Widmanstatten pat-tern.

By 600° C the Widmanstatten pattern is wellestablished in these meteorites (Table 16). Kama-cite lamellae have grown to near their final size,and taenite bands have shrunk to near theirs.Both taenite and kamacite will continue to increasein Ni with cooling, but compositional changes willtake place over increasingly short distances. It isin this environment, one that contains P in kama-cite at the 0.5 to 0.8 atomic % level, that rhabditenucleation must take place. They nucleate homo-geneously in kamacite within a narrow range oftemperatures and grow under conditions of equi-librium combined with severe P impingment. It isthe limited amount of P and the competition for itfrom closely spaced schreibersites that controlstheir growth rates and consequently their eventualsize. The larger the volume of metal from which aschreibersite draws P, the faster it grows and thelarger it becomes. Tie line shifts permit Ni levelsto vary depending upon growth rates. The largerfaster-grown rhabdites achieve lower Ni levels thanthe smaller slower-grown schreibersites.

A third major type of schreibersite is grainboundary schreibersite. Its petrography suggeststhat it replaces taenite at late stages in structuraldevelopment. Heterogeneous nucleation probablytakes place at kamacite-taenite interfaces, resultingin initial concentrations being controlled by thethree-phase a + y + Ph field of the ternarydiagram. P is supplied to the nucleation site mainlyfrom the bordering kamacite and Ni comes mainlyfrom the taenite. Equation 6 does not describeinitial growth rates but may well apply to kamacite-schreibersite interfaces after they become isolatedfrom the influence of taenite. Nucleation temper-atures probably depend upon specific local condi-tions and may extend to considerably lower tem-peratures than those appropriate for rhabdite for-mation.

There are three types of associations that maybe included under the term grain boundary schrei-bersite. They appear to represent a sequence ofstages in structural development. The first type isthe isolated grain boundary schreibersite. Theseare elongated schreibersites that lie along kama-cite-kamacite grain boundaries. They frequentlyoccupy grain boundary junctions and branch ac-cordingly. Their morphology suggests that theyreplaced residual taenite as it receded along thegrain boundary. Their Ni values decrease withdistance away from the nearest taenite ribbon ortaenite-plessite area, suggesting that the lowest Nigrain boundary schreibersite nucleated first. Analternate possibility is essentially simultaneous nu-cleation with faster growth accounting for lowerNi. These schreibersites are frequently somewhatlarger than the higher Ni ones closer to taenite.There is no obvious reason, however, why P im-pingment should be more critical at one site thanat another.

The second type of grain boundary schreibersiteis found in close association with taenite and tae-nite-plessite areas. It appears to have formed sim-ilarly to the isolated grain boundary schreibersites,but it either nucleated later or grew more slowly.It separated itself from taenite but was not able tocompletely absorb it. The final stage in this se-quence is the formation of taenite boundaryschreibersites. These are the small, highest Nischreibersites that appear to be embedded in tae-nite at taenite-kamacite borders. These wouldseem to be the final schreibersites to form.

Authors Copy

66 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

INTERFACE DATA AND THE a/a + PH BOUNDARY

In the earlier discussion of the extrapolation ofexperimental data to low temperatures the slopeof the a/a + Ph boundary was mentioned. Theextrapolation shows that the boundary moves to-ward the Fe-Ni axis with decreasing temperatureand suggests that it is parallel to it (Figures 21,22). This led to the simplifying assumption that Psaturation values in kamacite are independent ofNi and Fe concentrations and may be equated tospecific temperatures (Table 18). This in turnleads to a possible interpretation of the measuredP in kamacite values at kamacite-schreibersite in-terfaces as being equivalent to a final growthtemperature, a temperature below which subse-quent interface changes would be undetectable bythe technique employed.

Observed P interface values cover a range ofconcentrations from 0.16 to 0.05 atomic %. Afactor of three is involved, indicating that smalldifferences in measured concentrations may havemore than trivial significance. The higher P valuesare associated with massive schreibersite at inter-faces with low Ni kamacite, and the lower P valuesare associated with taenite border schreibersite atinterfaces with high Ni kamacite. Assuming thatthese values represent solubility limits, we canequate them to a temperature range of 445° to 350°C (Table 18). Measurements on the Santa Luziameteorite alone covered most of this range, 0.14to 0.05 atomic % P, equivalent to a temperaturedifference of 430° to 350° C. Is it reasonable toassume that large schreibersites stopped detectablegrowth before the later formed schreibersites, ordoes the a/a + Ph boundary really have a down-ward slope with increasing Ni? Are there otherfactors that may make a clear choice between thesetwo alternatives impossible on the basis of theavailable data?

The P and Ni kamacite interface data is summa-rized in Figure 27. The points represent individualinterface Ni and P concentrations in kamacite andare plotted using the axes at the left and at thebottom of the diagram. The symbols indicate themeteorite from which the data were taken. Thereis a general trend of decreasing P concentration inkamacite with increasing Ni concentration. Therelationship of these values to the estimated a/a+ y boundary of the Fe-Ni binary diagram ofGoldstein and Olgivie (1965) is also indicated. The

scale to the right of the diagram is the temperatureequivalent to the P concentration at the left, usingthe data from Table 18. The dashed line is the a/a + y boundary plotted in relation to temperatureand Ni concentration. It is obvious that interfacevalues for individual meteorites, as is the case forthe group as a whole, cover a range of both P andNi levels. A meteorite has interface values at sev-eral P levels, and a range of Ni values may becovered at any given P level. All of the interfacevalues are to the left of the a/a + y boundary, thehigher Ni values being closest to it.

There are at least three factors that must beconsidered as possible contributors to the observeddistribution in Figure 27. These are (1) coolingrates, (2) the possibility of a slopping a/a + Phboundary, and (3) the possibility of low tempera-ture nucleation at a/y boundaries and subsequentgrain boundary diffusion playing a role in thegrowth of grain boundary and taenite borderschreibersite. The available data is insufficient tosort out these influences in a quantitative way, buttheir expected effects can be outlined and sugges-tive observations mentioned.

Cooling rates have only been reported for fourof the meteorites under consideration here. Valuesof 2° C per million years for Lexington Countyand Goose Lake and 3.5° C per million years forBahjoi were given by Goldstein and Short (1967b).The recent measurements of Randich (1975) gavea somewhat higher value of 5° C per million yearsfor Coahuila, and there is no reported value forBellsbank. Kamacite band widths for the otherthree meteorites were taken from Buchwald(1976), and cooling rates were estimated using theGoldstein and Short (1967b) cooling rate curves.Ballinger and Balfour Downs gave cooling ratesof 2° C per million years and Santa Luzia gave anestimated 0.8° C per million years. These numbersindicate that cooling rate values for this group ofmeteorites cluster in a rather narrow range center-ing around 2° C per million years.

The effect of cooling rate variations on schrei-bersite-kamacite interface values has been dis-cussed above. It is to be expected that slowercooling rates would produce lower detectable in-terface values of P in kamacite and lower levels ofP in kamacite generally. This would be equivalentto detectable growth continuing to lower tempera-tures. This type of variation may be present on aminor scale in the data considered here. Coahuila

Authors Copy

NUMBER 21 67

0.30

O25

0.22

0.200.190.180.170.16

0.15

0.14

a* o.i2SJO.II

| 0.10

< 0.09

0.08

0.07

0.06

0.05-

0.04

0.03

0.02

A

•

•

•

O

A

D

a,BELLSBANK

b, COAHUILA

C, SANTA LUZIA

d, BALLIN6ER

e, LEXINGTON CO.

f, BAHJOI

Q, 600SE LAKE

h, BALFOUR DOWNSa + y

AA

(A CCE A A | » ^ Jft.

*co

500

490

480

470

460

450

440

430

420

410

400

390

380

370

360

350

340

330

320

310

300

29O

280

I 2 3 4 5 6 7 8 9

ATOMIC % N i a

FIGURE 27. —P in kamacite at kamacite-schreibersite interfaces plotted against Ni in kamacite(temperature scale at right relates P saturation values to a/a + y boundary of Fe-Ni system).

Authors Copy

68 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

does have the highest reported cooling rate of 5°C per million years, and its interface P values andP levels in kamacite do seem to be somewhathigher than for some of the other meteoritesstudied. Bellsbank also seems to have unusuallyhigh levels of P in kamacite, and this could be theconsequence of a comparatively high cooling rate.The points representing both Coahuila and Bells-bank interfaces are in the upper part of Figure27, perhaps meaning that detectable change ceasedat higher temperatures for these two meteoritesdue to higher cooling rates.

An a/a + Ph boundary (AB in Figure 20) thatslopes downward from the Fe-P axis toward theFe-Ni axis would contribute to the type of datadistribution observed in Figure 27. At any giventemperature lower Ni schreibersites would have ahigher P interface value in kamacite than higherNi schreibersites. The range of the data, however,seems to require much too steep a slope to bereasonable in terms of the extrapolation givenabove. An Fe-P axis intercept value of approxi-mately 0.2 atomic % would be indicated. Thiswould mean either an unreasonably high P satu-ration temperature in the neighborhood of 465° C(Table 18), or that the actual curve bends sharplyto the right of the indicated extrapolation (Figure21). It also seems unlikely that error in locatingthe data points is a major factor. It was pointedout above that reproducibility of P determinationswas ±0.005 weight % and for Ni was ±0.1 weight%. This means that the P levels shown are distin-guishable from their nearest neighbors with rea-sonable certainty and that the Ni variations aremeaningful. Ni variations at a given P level rangefrom 0.8 to 3.0 atomic %. Variations for a specificmeteorite at a given P level are not so wide, butthey do range up to 1.5 atomic %. This range inNi values for a single meteorite at a given P levelsuggests that factors other than simply a slopinga/a + Ph boundary are involved. The boundarymay well have a gentle slope, but other factorsmust also contribute to the observed distribution.

Tie lines representing three of the P levels inFigure 27 are plotted on isothermal sections inFigure 28. The symbols used in Figure 27 areused at the schreibersite ends of the tie lines toidentify the meteorite represented. Tie lines ac-tually belonging to a given isothermal section willfan out across the field without overlap. Whenoverlap occurs it implies either error in the data

or plotting on the wrong isothermal section. Wesaw in Figure 26 that all of the measured tie linesfor three of the meteorites can be plotted withoutobvious cases of overlap. This could mean they allbelong to the same isothermal section, but it couldalso be fortuitous. If interface data for all of themeteorites could be plotted as tie lines on a singleintelligible drawing, several cases of overlap wouldbe observed. When tie lines are separated on thebasis of P level as in Figure 28, no serious cases ofoverlap are encountered. The two tie lines at theleft on the 0.09 atomic % P section do overlapslightly, but a small shift in Ni values could doaway with it. The surprising point is how littleoverlap is seen when the data are taken directlyfrom the tables given earlier and plotted in thisway.

Had all of the tie lines in Figure 28 been plottedon a single isothermal section, there would havebeen several cases of overlap. The most seriousoffender would be a Bellsbank tie line, the secondone from the left on the 0.09 atomic % P isother-mal section. If it were plotted on the top sectionof Figure 28, there would have been obviousoverlap with another Bellsbank tie line. All threeof the tie lines in this top isothermal section arefrom the Bellsbank meteorite, and there is addi-tional evidence to be discussed below that suggeststhat two temperatures are involved.

Plotting tie lines as has been done in Figure 28supports the idea that a temperature effect is real.That is to say that late formed schreibersites con-tinued significant growth to temperatures belowthose of early formed schreibersites. Unfortu-nately, these meteorites do not supply us with afull range of tie lines at any one temperature.More work will be required before the slope ofthe a/a + Ph boundary within this low tempera-ture range can be established with certainty.

A third possibly complicating process that couldbe of importance here has been discussed byComerford (1969). He pointed out that grainboundary diffusion might play an important rolein the development of grain boundary schreiber-sites. He suggested that grain boundary diffusionwould supply Ni to schreibersites growing at lowtemperatures more readily than lattice diffusion.The discussion above has pointed out that P is thecontrolling element rather than Ni, but his sugges-tion would be equally valid for P.

The measurements on grain boundary schrei-

Authors Copy

NUMBER 21 69

0.09 ATOMIC %P395° C

4.4

0100 % Fe

10 20 30ATOMIC %Ni

40

FIGURE 28. —Tie lines from various meteorites separated according to interface P saturation valuesin kamacite for plotting on isothermal sections.

Authors Copy

70 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

bersite in the Santa Luzia meteorite are represent-ative of many such measurements reported above(Figures 10, 11). Petrographic relationships sug-gest that this type of schreibersite nucleated atkamacite-taenite interfaces absorbing residual tae-nite. Nucleation temperatures were probably lowcompared to those for homogeneously nucleatedrhabdite, but the data does not allow a closertemperature estimate. It would also appear thatnucleation occurred over a range of temperature,nucleation of the lowest Ni schreibersite farthestalong the grain boundary away from taenite hav-ing taken place first (Figure 11, traverse 11). TheNi concentration of schreibersite nucleated at akamacite-taenite border is that of the schreibersitecorner of the three-phase field at the nucleationtemperature. The presence of the phase boundaryprobably serves both as a site for nucleation andas a source of P through the process of grainboundary diffusion. As the schreibersite becomeslarger, however, the contribution of grain bound-ary diffusion becomes increasingly less important.The schreibersite grows rapidly, with its Ni levelincreasing only slightly. Its interface with kamaciterepresents a tie line that probably tends to shift tothe left with decreasing temperature, as P is drawnfrom a relatively large diffusion field.

Along the grain boundary at a location closer tothe final residual taenite a second grain boundaryschreibersite nucleates at a lower temperature (Fig-ure 11, traverse 12). Its initial Ni is higher, andthe P diffusion rate is lower. Grain boundarydiffusion again makes an important contributionduring initial growth, but it probably becomesincreasingly less important as the schreibersiteincreases in size.

The observed taenite border schreibersite un-doubtedly nucleated at a still lower temperature(Figure 11, traverse 13). It formed at the taenite-kamacite border and had a high initial Ni concen-tration. Growth was insufficient to separate theschreibersite completely from taenite. A smallschreibersite that has a Ni gradient due to thefact that it is in contact with kamacite at one endand taenite at the other resulted. This type ofschreibersite has both the highest observed Niconcentrations and the lowest observed P interfacevalues in kamacite. A reasonable explanation ofthese observations is that grain boundary diffusionplays an increasingly important role both with

small schreibersite size and lower temperature ofgrowth.

Grain boundary diffusion may also help explainthe Bellsbank interface measurements summa-rized in Figure 5. Four of the seven interfacesshown have a P level in kamacite of 0.09 weight %P (0.16 atomic %), and the other three are at 0.05weight % P (0.09 atomic %). The schreibersiteinterface illustrated at the top of the diagram is atypical one for massive schreibersite. The schrei-bersite illustrated at the lower left is a large rhab-dite (Table 3, traverse 6) surrounded by clearkamacite within an area of microrhabdites. Therewere no grain boundaries or subgrain boundariesobserved. The schreibersite illustrated at the lowerright is another large elongated rhabdite (Table 3,traverses 4 and 5) that was observed to be along adefinite subgrain boundary. Both of these rhab-dites have extremely large P gradients, and theirP interface values differ by 0.04 weight % P (0.07atomic %). The center schreibersite in the lowersection of the diagram is a large rhabdite (Table3, traverses 2 and 3) along a subgrain boundarywith different P and Ni interface values in kama-cite on its opposite sides. Tie lines representingthese interfaces are given in Figure 28. The threehigh P tie lines are in the top isothermal sectionand the fourth, representing the low P interfaces,is at the left in the middle section. A possiblerationale for these observations is that Bellsbankcooled more rapidly than the other meteoritesstudied. As a result, growth apparently stopped athigher temperatures than has been the case forother rhabdite-containing meteorites. The pres-ence of grain boundaries, however, permitted Pflow to continue to significantly lower tempera-tures for the two schreibersites with the lowerinterface levels. A faster type of diffusion permit-ted detectable interface changes to extend to lowertemperatures. It would be interesting to find simi-lar situations in other meteorites.

SCHREIBERSITE WITH COHENITE BORDERS

A prominent feature of the meteorites selectedfor study is the presence of 100-to-200-/Lim-widecohenite borders separating massive schreibersitefrom kamacite. It is a feature that can be recog-nized with the unaided eye on many well-preparedexhibit sections of Group I meteorites, but it has

Authors Copy

NUMBER 21 71

been little noted in the literature until relativelyrecently. Cohenite borders may partially or com-pletely enclose schreibersite bordering troilite in-clusions or massive schreibersite isolated withinkamacite. This type of association has been re-ported above for the Ballinger (Figure 6), Lexing-ton County (Figure 12), Bahjoi (Figure 13), GooseLake, and Balfour Downs (Figures 15, 16, 17)meteorites. The Ni concentrations in these cohen-ite enclosed schreibersites range from 16 to 21atomic %, comparable to those observed for simi-lar unenclosed schreibersites within the same me-teorite or within similar meteorites. The coheniteborders contain small amounts of Ni and no de-tected P. Their Ni profiles show a gradient varyingfrom as much as 1.7 weight % Ni at the cohenite-kamacite interface to as low as 0.8 weight % Ni atthe cohenite-schreibersite interface (Figures 13,17).

Similar Ni gradients in cohenite borders havebeen reported previously by Drake (1970) andScott (1971a). Drake (1970) noted that Ni diffusesthrough cohenite, increasing the Ni content ofenclosed kamacite and taenite grains with decreas-ing temperature, but he felt that cohenite shellsaround schreibersite prevented equilibration ofschreibersite with the enclosing metal. Scott(1971a) interpreted the presence of Ni gradientsin cohenite borders as implying equilibration ofNi to low temperatures. The observed Ni concen-trations in cohenite-enclosed schreibersite sup-ports Scott's interpretation.

Massive schreibersite growth is apparentlyblocked by enclosure in cohenite. The absence ofdetectable P in, these cohenites indicates that Psolubility is very low. It seems reasonable to as-sume, therefore, that cohenite borders isolateschreibersite from a source of additional P. Thesituation with Ni, however, is different. The co-henite borders contain appreciable Ni, and themeasured Ni gradients suggest a flow of Ni to thecohenite-schreibersite interfaces. Under these cir-cumstances the size of the schreibersite indicatesthe maximum temperature at which cohenitecould have nucleated. The temperature must havebeen low enough to permit schreibersite to reachits full size. At this maximum temperature, or atsome lower temperature, cohenite encases theschreibersite. Growth is stopped but Ni enrich-ment continues with decreasing temperature. The

minimum temperature at which cohenite couldhave formed is at or below the temperature atwhich the schreibersite achieved its final Ni con-centration. According to the Fe-Ni-P diagram thiswould be at 550° C or below. In this low tempera-ture range schreibersite would be surrounded bylarge volumes of kamacite. Is kamacite a reasona-ble source for the large amount of carbon requiredfor cohenite growth? The work of Brett (1967)and A. D. Romig, Jr. (1977, pers. comm.) on theFe-Ni-C system seems to require an alternate ex-planation.

Consideration of the Goose Lake meteoritestructure in terms of both its cooling history withinthe Fe-Ni-P system (Table 14) and Romig's Fe-Ni-C diagram may be instructive at this point. GooseLake is a meteorite that has large amounts ofmassive schreibersite. Surface areas totaling morethan 1000 cm2 were examined and found to con-tain more than 20 areas of clustered massive schrei-bersites. All of these schreibersites appeared to beenclosed in cohenite, although poor condition ofsurfaces left some ambiguity in one or two cases.In any event, the ubiquitous nature of cohenitebordering massive schreibersite in Goose Lake wasestablished. This suggested that Goose Lake mightbe expected to yield a more straight-forward inter-pretation of cohenite border growth than wouldbe possible using meteorites where the associationwas more sporadic.

Equilibrium compositions of phases derivedfrom the Fe-Ni-P system for a range of decreasingtemperatures are given in the lower section ofTable 14 for the assumed Goose Lake meteoritebulk composition. At 925° C schreibersite is inequilibrium with 96% taenite. On cooling to 750°C the schreibersite doubles in amount to 8% andincreases in Ni from 5.7 to 6.6 atomic %. Taeniteis slightly reduced in volume during this process,but its Ni concentration remains essentially con-stant at 7.9 atomic %. Cooling through this 175° Crange has resulted primarily in the localization ofP. Massive schreibersites have grown within taeniteand have achieved most of their eventual size.

In the 750° to 650° C temperature range majorstructural changes take place. Taenite increases inNi from 7.9 to 11 atomic % and decreases involume by one-half. Schreibersite increases in Nifrom 6.6 to 9.0 atomic % and increases in volumeby approximately 10%. The shrinking taenite is

Authors Copy

72 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

converted to kamacite of 5.0 atomic % Ni, whichmakes up nearly half the structure at 650° C. Withfurther decreasing temperature, kamacite ex-pands in volume, taenite shrinks, and schreibersiteholds its own. All three phases continue to increasein Ni.

What happens if carbon is also present in thesystem? Romig's Fe-Ni-C phase diagram gives car-bon saturation values for taenite in equilibriumwith cohenite. These are given in Table 20 forselected taenite compositions for the range oftemperatures covered. The taenite Ni values givenare those of taenite in equilibrium with schreiber-site at the indicated temperature and the GooseLake bulk composition using the Fe-Ni-P system(Table 14). Carbon has appreciable solubility intaenite throughout this temperature range, but itssolubility in kamacite is negligible. This solubilityof carbon in taenite means that a meteorite suchas Goose Lake with a structure consisting of 92%taenite at 750° C might well contain enough dis-solved carbon to produce significant amounts ofcohenite on subsequent cooling. The importantfactor is that the amount of taenite present isdrastically reduced with decreasing temperature.This results in carbon concentration increasing inthe ever decreasing amount of taenite. Supersatu-ration must result at some temperature if sufficientcarbon were present to start with.

Brett (1967) assumed that P could be neglectedin his discussion of cohenite formation in meteor-ites. That may be true for some types of coheniteassociations, but it may well not hold for cohenitebordering massive schreibersite. It is difficult tounderstand these associations using the Fe-Ni-Csystem alone. Brett (1967) shows that in this systemcohenite grows by the following reaction:

y = cohenite + a.

He also points out that when the Ni content ishigh, kamacite and graphite form with fallingtemperature. Cohenite is what we see, however,not significant amounts of graphite in kamacite.

If we assume that carbon is present in taenite,then we are actually dealing with the Fe-Ni-P-Csystem. The more complex system permits thepostulation of the following sequence of events toaccount for massive schreibersite with coheniteborders. Massive schreibersite nucleates at hightemperature and grows upon cooling to near its

TABLE 20. —Carbon saturation values for taenite inequilibrium with cohenite from the Fe-Ni-C dia-gram of A. D. Romig, Jr. (taenite Ni values areequilibrium values using the Goose Lake bulk com-position, Table 14).

°c

730

650

600

500

Weight % C

0.75

0.45

0.36

0.18

Atomic % Ni

n14

•v25

final size, perhaps in the 650° to 700° C tempera-ture range, assuming the Goose Lake bulk com-position. These schreibersites are in contact withtaenite that has become saturated in carbon. Withfurther cooling both taenite and schreibersite re-quire more Ni, and taenite requires a repositoryfor its carbon. Under these circumstances it isreasonable to assume the following nucleation re-action taking place at taenite-schreibersite inter-faces:

OKQat) + Ph = y + cohenite + Ph.

A cohenite border at the preexisting taenite-schrei-bersite interface results. Cohenite becomes a re-pository for carbon and the Ni that previouslyresided there is absorbed by taenite and/or schrei-bersite. As the structure develops, a dynamic equi-librium involving two interfaces is established; car-bon saturated taenite-cohenite and cohenite-schreibersite. Cohenite grows as long as carbonsaturated taenite is present. Taenite shrinks involume, cohenite fills in behind the receding tae-nite, and schreibersite becomes enriched in Ni. Atsome point kamacite precipitates at the cohenite-taenite interface and then kamacite becomes thesource of Ni for transport to schreibersite. Thecohenite border is established and subsequentchanges involve diffusion of Ni and Fe only.

The postulated reaction sequence would permitcohenite to grow over a broader temperaturerange than was suggested by Brett (1967). Themaximum temperature can be estimated from thesize of the cohenite-enclosed schreibersite. Whatis the highest temperature that could have pro-duced schreibersite of the observed size startingwith a given bulk composition? In a meteorite likeGoose Lake the temperature might be as high as750° C. The minimum temperature of nucleation

Authors Copy

NUMBER 21 73

would be established by the disappearance of suf-ficient taenite to supply the required carbon super-saturation. It seems unlikely for the meteoritesthat have been studied here that this would bebelow 600° C. This postulated growth sequencecould extend Brett's (1967) temperature range of610° to 650° C upward, but it is unlikely that itwould be lowered appreciably. If Brett's (1967)estimated rates of cohenite decomposition withtemperature are correct, nucleation within thelower part of the possible temperature rangewould favor preservation of cohenite. Confirma-tion of this proposed sequence of events will haveto await the results of detailed experimentation.

Summary and Conclusions

The role that schreibersite growth played in thestructural development process in coarse-struc-tured iron meteorites has been examined. Theavailability to the writer of many large meteoritesurfaces and an extensive collection of metallo-graphic sections made it possible to undertake acomprehensive survey of schreibersite petrogra-phy. This study was the basis for the selection ofsamples for detailed electron microprobe analysis.Samples containing representative structures fromeight chemical Groups I and IIAB meteorites wereselected. All of these meteorites contain schreiber-site, ranging in amount from easily detectable toabundant, and they contain additional P in solu-tion in kamacite and taenite. Their detailed metal-lography indicates that they are comparatively freeof secondary effects such as shock and reheatingthat might mask or distort relationships of interest.The range of bulk compositions covered is onewhere small changes in P-Ni concentration rela-tionships may be more important to the structuraldevelopment process than would be the case athigher Ni levels.

Lengthy electron microprobe traverses weremade across structures representative of the ob-served range of schreibersite associations, with theexception of microrhabdites. Particular emphasiswas placed on schreibersite-kamacite interfacecompositions. Ni and P interface concentrationsin kamacite and Ni interface values in schreibersitewere determined, along with associated Ni and Pconcentration gradients in kamacite and Ni gra-dients in schreibersite. An analysis of these data

has led to a more comprehensive description ofthe structural development process in these indi-vidual meteorites than has been available previ-ously.

Massive schreibersite, one of the four majortypes of schreibersite encountered, may be ac-counted for by equilibrium considerations. TheFe-Ni-P equilibrium diagram and estimated bulkcompositions for individual meteorites were usedto explain the basic structures present. Subsolidusnucleation and growth with slow cooling fromtemperatures at least as high as 850° C, and prob-ably much higher, explain the phase relationshipsthat one sees in meteorite specimens. No needwas found to invoke eutectic liquids to account forany of the observed massive schreibersite mor-phologies. After nucleation, schreibersite andkamacite grew in volume and increased in Nicontent while taenite receded and increased in Nicontent under equilibrium or near equilibriumconditions into the 600° C temperature range. Theretention of taenite in the octahedrites establishesthat bulk equilibrium did not extend as low as 550°C. Schreibersite undoubtedly continued in equilib-rium with its enclosing kamacite to lower temper-atures. The growth of massive schreibersite fromhigh temperatures into the 600° C temperaturerange served both to localize P by reducing itslevel throughout the bulk metal and to producethe large swathing kamacite zones that are ob-served.

The second type of schreibersite to form ishomogeneously nucleated rhabdite. It nucleatedin kamacite in the 600° C temperature range,either as a consequence of low initial P level orafter local P supersaturation developed followingmassive schreibersite growth. Growth of theseschreibersites was also under equilibrium condi-tions to below 500° C. They drew their source of Pfrom limited diffusion fields when compared tothe massive schreibersites.

The third type of schreibersite is grain boundaryand taenite border schreibersite. It formed atkamacite-taenite interfaces, absorbing residual tae-nite. Nucleation took place successively along grainboundaries over a range of temperatures startingas high as 500° C or perhaps slightly higher. Withdecreasing temperature taenite receded and be-came richer in Ni and the schreibersite that nu-cleated at these interfaces had higher Ni. The

Authors Copy

74 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

highest Ni schreibersites are the small ones thatnucleated at low temperatures and are observedstill in contact with taenite. Grain boundary diffu-sion probably became an increasingly importantfactor in the growth of these schreibersites withdecreasing temperature.

The fourth type of schreibersite is microrhab-dite. These schreibersites nucleated homogene-ously in supersaturated kamacite at temperaturesin the 400° C range or below.

P diffusion controlled the growth rate of schrei-bersite. The Ni flux to a growing interface had toproduce a growth rate equal to that established bythe P flux. This was accomplished by tie line shiftsthat permitted a broad range of Ni growth ratesand accounts for the observed range of Ni concen-trations in schreibersite. Interface measurements

on meteoritic schreibersites, therefore, yield arange of tie lines in the low temperature Fe-Ni-Psystem. Equilibrium conditions pertained atgrowth interfaces to temperatures far below thoseavailable experimentally. Kinetic factors, however,restricted mass transfer to increasingly small vol-umes of material with decreasing temperature.

Schreibersite with cohenite borders is a commonoccurrence in many meteorites. It is suggestedthat it formed as a reaction in the Fe-Ni-P-Csystem in the 750° to 600° C temperature range.Carbon saturated taenite reacted with schreibersiteto form cohenite. Cohenite then grew in equilib-rium with both taenite and schreibersite.

Berzelius' prescient comment of 1832 on therole of P in the development of the Widmanstattenpattern has been confirmed and expanded upon.

Authors Copy

Appendix

In the discussion of the low-temperature ternarydiagram, experimental values for P saturation con-centrations in kamacite and taenite were extrapo-lated to lower temperatures (Table 17, Figures 21,22). This is a commonly used procedure whendealing with dilute solutions and is a variation ona method used for the determination of partialmolal properties of substances. It is applicable tosolid state transformations in dilute metallic sys-tems and may be justified by the following ther-modynamic reasoning adapted from Swalin(1972:168-171).

At a given temperature (Tx) the end points of akamacite-schreibersite tie line in the kamacite-schreibersite field of the equilibrium diagram rep-resent equilibrium concentrations of the twophases. The relationship between P-saturatedkamacite (ap(Sat)) and schreibersite (Ph) may beexpressed as follows:

ttp(sat) ^ Ph + a AG = 0.

Since the reaction represents equilibrium thechange in the Gibbs free energy (AG) is zero.Therefore, the chemical potentials (fi) of P in thetwo phases are equal,

f^P M'P(sat)

as are their activities (a),•Ph _ „<*

aP — aP ( s a t ) .

Since schreibersite is a pure substance with respecttoP

requiring that

- •Ph _

aP —

—aP(sat) — 1 •

If the saturation concentration of P in kamacite issufficiently small, P may be assumed to followHenry's law and the activity coefficient of P inkamacite, y?, may be evaluated.

aP(sat) — 7P AP(sat)

1

In order to derive a relationship for the solvusline (kamacite saturated with P) as a function oftemperature (T), we must consider a situationwhere AG for the reaction is not zero. This maybe done by assuming a P concentration in kama-cite, Xg, such that Xp" is within the kamacite fieldof the equilibrium diagram

X aP -P(sat)-

Kamacite of this concentration would react withschreibersite to form P-saturated kamacite,

aP + Ph «* «P(Sat) AGX.

The following equation may be written relatingthe Gibbs free energy ( AG2) for the above reaction,the chemical potentials of the reacting species,and the activities of these species,

AGX = n% - At°Pph = RT In - ^ •

for a pure substance,

aPh = 1

for a dilute solution following Henry's law

a? = yAX®

and from equation given above

A P

-m) =•(sat) -PCsat)

Substituting these values of aPh and aP in theexpression for AGX given above we obtain

fi$ - n°Pph = RT In Xg - RT In Xp\sat) - RT In (1).

We know that the enthalpy of a component i, H,,is related to the temperature and chemical poten-tial by

= Hi.

75

Authors Copy

76 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

Substituting, we obtain

- Hp h

d(RT In Xg/T) a(RT In

3(1/1)

Xp and (1) are constants, dropping out these twoterms,

HPa - HPh = - Rd(lnXHsat))

a(VT)

3(1/'T)R

= - 3(ln

In Xp(Sat) =- A H F

R

Therefore, when Xp\sat) is plotted on a log scaleagainst 1/T, a straight line with slope — AHP/Rshould result. Straight lines with negative slopesare observed in our case, and it is reasonable toassume on the basis of experience in other systemsthat AHPa is independent of temperature over areasonable temperature range. It would seem thenthat an extrapolation over a reasonable tempera-ture range is justified. A similar treatment wouldapply to P-saturated taenite data.

Authors Copy

Literature Cited

Adler, I., and E. J. Dwornik1961. Electronprobe Analysis of Schreibersite (Rhabdite) in

the Canyon Diablo Meteorite. U.S. Geological SurveyProfessional Papers, 424B(112):B263-B265.

Agrell, S. O., J.V.P. Long, and R. E. Ogilvie1963. Nickel Content of Kamacite Near the Interface with

Taenite in Iron Meteorites. Nature, 198:749-750.Anders, Edward

1962. Meteorite Ages. Reviews of Modern Physics, 34:287-325.

1971. Meteorites and the Early Solar System. Annual Reviewof Astronomy and Astrophysics, 9:1-34.

Axon, H. J., and j . Boustead1967. Kamacite-Taenite Relationships in Iron Meteorites.

Nature, 213:166-167.Axon, H. J., and D. Faulkner

1970. A Metallographic and Microprobe Study of the Bar-ranca Blanca Meteorite. Mineralogical Magazine,37:898-904.

Axon, H. J., and j . I. Goldstein1972. Temperature-Time Relationships from Lunar Two

Phase Metallic Particles (14310, 14163, 14003). Earthand Planetary Science Letters, 16:439-447.

1973. Metallic Particles of High Cobalt Content in Apollo15 Soil Samples. Earth and Planetary Science Letters,18:173-190.

Axon, H. J., and P. L. Smith1972. A Metallographic Study of Some Iron Meteorites of

High Nickel Content. Mineralogical Magazine, 38:736-755.

Axon, H.J . , andC. V. Waine1971. A Metallographic Study of the Angra dos Reis (Iron)

Meteorite. Mineralogical Magazine, 38:94-101.1972. A Metallographic Study of Some Hexahedrites. Min-

eralogical Magazine, 38:725-735.Belaiew, N. T.

1924. On the Genesis of Widmanstatten Structure in Mete-orites and in Iron-Nickel and Iron-Carbon Alloys.Mineralogical Magazine, 20:173-185.

Benedicks, Carl1910. Synthese du fer meteorique. Nova Ada Reglae Societa-

tis Scientiarum Upsaliensis, series IV, 2(10): 1-26.Berwerth, Frederick

1907. Steel and Meteoritic Iron. Journal of the Iron and SteelInstitute, 75(3):37-51.

Berzelius, J. J.1832a. Undersokning af en vid Bohumiliz i Bohmen funnen

jernmassa. Kongelige Svenska Vetenskaps-AcademiensHandlingar, pages 106-119.

1832b. Untersuchung der Bohumulitzer Eisenmasse. Zeit-schriftfur Physik und Mathematik, Wein, 1:289-297.

1833. Untersuchung einer bie Bohumiliz in Bohmengefundenen Masse. Annalen der Physik und Chemie,

27:118-132.1834a. Om Meteorstenar. Kongelige Svenska Vetenskaps-Acade-

miens Handlingar, pages 115-183.1834b. Ueber Meteorsteinen, V: Pallas-Eisen und Pallas-Oli-

vin, VI: Meteoreisen von Elbogen. Annalen der Physikund Chemie, 33:123-137.

Bild, Richard W.1974. New Occurrences of Phosphates in Iron Meteorites.

Contributions to Mineralogy and Petrology, 45:91-98.Brett, Robin

1967. Cohenite: Its Occurrence and a Proposed Origin.Geochimica et Cosmochimica Ada, 31:143-159.

Brezina, Aristides1906. Zur Frage der Bildungsweise Eutropischer Gemenge.

Denkschriften der Kaiserlichen Akademie der Wissenschaf-ten, Wein, 78:635-643.

Brown, G. M., A. Peckett, R. Phillips, and C. H. Emeleus1973. Mineral-Chemical Variations in the Apollo 16 Mag-

nesio-Feldspathic Highland Rocks. In Proceedings ofthe Fourth Lunar Science Conference. Geochimica etCosmochimica Ada, supplement 4, 1:505-518.

Buchwald, V. F.1966. The Iron-Nickel-Phosphorous System and the Struc-

ture of Iron Meteorites. Ada Polytechnica Scandinavica,51: 45 pages.

1976. Handbook of Iron Meteorites. Volumes I—III, 1418pages. Tempe: Center for Meteorite Studies, ArizonaState University; Berkeley: University of CaliforniaPress.

Buchwald, V. F., and S. Munck1965. Catalogue of Meteorite in the Mineralogical Museum

of the University, Copenhagen. Analecta Geologica, 1:81 pages.

Buseck, Peter R.1969. Phosphide from Meteorites: Barringerite, a New

Iron-Nickel Mineral. Science, 165:169-171.Carter, J. L., and E. Padovani

1973. Genetic Implications of Some Unusual Particles inApollo 16 Less Than 1 mm Fines 68841,11 and69941,13. In Proceedings of the Fourth Lunar ScienceConference. Geochimica et Cosmochimica Acta, supple-ment 4, 1:323-333.

Clarke, Roy S., Jr.1969. Comments on Cohenite and Schreibersite in Iron

Meteorites (Abstract). Meteoritics, 4:161.Cohen E.

1894. Meteoreisen-Studien III: Rhabdit. Annalen des K.K.Naturhistorischen Hofmuseums, IX:97-118.

Comerford, M. F.1969. Phosphide and Carbide Inclusions in Iron Meteorites.

Pages 780-795 in P. M. Millman, editor, MeteoriteResearch. 940 pages. Dordrecht: D. Reidel PublishingCompany.

77

Authors Copy

78 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

Daubree, M. A.1868. Experiences synthetiques relatives aux meteorites. 65 pages.

Paris: Dunod.De Laeter, J. R., G.J.H. McCall, and S.J.B. Reed

1973. The Redfields Meteorite —A Unique Iron from West-ern Australia. Mineralogical Magazine, 39:30-35.

Derge, G., and A. R. Kommel1937. The Structure of Meteoric Irons. American Journal of

Science, 34:203-214.Doan, A. S., Jr., and J. I. Goldstein

1969. The Formation of Phosphides in Iron Meteorites.Pages 763-779 in P. M. Millman, editor, MeteoriteResearch. 940 pages. Dordrecht: D. Reidel PublishingCompany.

1970. The Ternary Phase Diagram, Fe-Ni-P. MetallurgicalTransactions ,1:17 59-1767.

Drake, John C.1970. Mineralogical Relationships in Seymour, a Course

Octahedrite. Meteoritics, 5:19-31.El Goresy, Ahmed

1965. Mineralbestand und Strukturen der Graphit- undSulfideinschliisse in Eisenmeteoriten. Geochimica etCosmochimica A eta, 29:1131-1151.

El Gorsey, A., P. Ramdohr, and O. Medenback1973. Lunar Samples from Descartes Site: Opaque Mineral-

ogy and Geochemistry. In Proceedings of the FourthLunar Science Conference. Geochimica et CosmochimicaAda, supplement 4, 1:733-750.

Farrington, Oliver C.1915. Meteorites, Their Structure, Composition, and Terrestrial

Relations. 225 pages. Chicago: Published by the au-thor.

Fricker, P. E., J. I. Goldstein, and A. L. Summers1970. Cooling Rates and Thermal Histories of Iron and

Stony-iron Meteorites. Geochimica et Cosmochimica Acta,34:475-491.

Fuchs, Louis H.1969. The Phosphate Mineralogy of Meteorites. Pages 681-

695 in P. M. Millman, editor, Meteorite Research. 940pages. Dordrecht: D. Reidel Publishing Company.

Goldberg, E., A. Uchiyama, and H. Brown1951. The Distribution of Nickel, Cobalt, Gallium, Palla-

dium, and Gold in Iron Meteorites. Geochimica etCosmochimica Acta, 2:1-25.

Goldstein, J. I.1965. The Formation of the Kamacite Phase in Metallic

Meteorites. Journal of Geophysical Research, 70:6223-6232.

1969. The Classification of Iron Meteorites. Pages 721-737in P. M. Millman, editor, Meteorite Research. 940pages. Dordrecht: D. Reidel Publishing Company.

Goldstein, J. I., and H. J. Axon1973. The Widmanstatten Figure in Iron Meteorites.

Naturwissenschaften, 60:313-321.Goldstein, J. I., H. J. Axon, and C. F. Yen

1972. Metallic Particles in the Apollo 14 Lunar Soil. InProceedings of the Third Lunar Science Conference.Geochimica et Cosmochimica Acta, supplement 3,1:1037-1064.

Goldstein, J. I., and A. S. Doan, Jr.1972. The Effects of Phosphorus on the Formation of the

Widmanstatten Pattern of Iron Meteorites. Geochimicaet Cosmochimica Acta, 36:51-69.

Goldstein, J. I., E. P., Henderson, and H. Yakowitz1970. Investigation of Lunar Metal Particles. In Proceedings

of the Apollo 11 Lunar Science Conference. Geochim-ica et Cosmichimica Acta, supplement 1, 1:499-512.

Goldstein, J. I., and R. E. Ogilvie1963. Electron Microanalysis of Metallic Meteorites, Part 1:

Phosphides and Sulfides. Geochimica et CosmochimicaActa, 27:623-637.

1965a. A Re-evaluation of the Iron-Rich Portion of the Fe-Ni System. Transactions of the Metallurgical Society ofAIME, 233:2083-2087.

1965b. The Growth of the Widmanstatten Pattern in MetallicMeteorites. Geochimica et Cosmochimica Acta, 29:893-920.

Goldstein, J. I., and J. M. Short1967a. Cooling Rates of 27 Iron and Stony-iron Meteorites.

Geochimica et Cosmochimica Acta, 31:1001-1023.1967b. The Iron Meteorites, Their Thermal History and

Parent Bodies. Geochimica et Cosmochimica Acta,31:1733-1770.

Goldstein, J. I., and H. Yakowitz1971. Metallic Inclusions and Metal Particles in the Apollo

12 Lunar Soil. In Proceedings of the Second LunarScience Conference. Geochimica et Cosmochimica Acta,supplement 2, 1:177-191.

Goode, George Brown1877. The Smithsonian Institution 1846-1896. 856 pages.

Washington: The Smithsonian Institution.Gooley, R. C , R. Brett, and J. L. Warner

1973. Crystallization History of Metal Particles in Apollo 16Rake Samples. In Proceedings of the Fourth LunarScience Conference. Geochimica et Cosmochimica Acta,supplement 4, 1:799-810.

Grossman, L., and J. W. Larimer1974. Early Chemical History of the Solar System. Reviews

of Geophysics and Space Physics, 12:71-101.Haidinger, Wilhelm

1847. Versammlung, am 16, Juli. Berichte uber die Mitteil-ungen von Freunden der Naturwissenschaften in Wien,3:69-71.

Hey ward, T. R., and J. I. Goldstein1973. Ternary Diffusion in the a and y Phases of the Fe-

Ni-P System. Metallurgical Transactions, 4:2335-2342.Henderson, Edward P.

1965. Hexahedrites. Smithsonian Miscellaneous Collections,148(5): 41 pages.

Henderson, E. P., and S. H. Perry1958. Studies of Seven Siderites. Proceedings of the United

States National Museum, 107:339-403.Hornbogen, E., and H. Kreye

1970. The Microstructure of Two Iron Meteorites. Zeit-schriftfur Metallkunde ,61:914-923.

Lovering, J. F., W. Nichiporuk, A. Chodos, and H. Brown1957. The Distribution of Gallium, Germanium, Cobalt,

Chromium, and Copper in Iron and Stony-Iron Me-teorites in Relation to Nickel Content and Structure.Geochimica et Cosmochimica Acta, 11:263-278.

Malissa, H. jun1974. Electron Microprobe Analysis of Some Minor- and

Authors Copy

NUMBER 21 79

Accessory Components in Mesosiderites. TMPMTschermaks Mineralogische und Petrographische Mitteil-ungen, 21:233-245.

Massalski, T. B., and F. R. Park1962. A Quantitative Metallographic Study of Five Octa-

hedrite Meteorites. Journal of Geophysical Research,67:2925-2934.

McKay, G. A., S. J. Kridelbaugh, and D. F. Weill1973. The Occurrence and Origin of Schreibersite-Kama-

cite Intergrowths in Microbreccia 66055. In Proceed-ings of the Fourth Lunar Science Conference. Geo-chimica et Cosmochimica Ada, supplement 4, 1:811-818.

Mehl, R. F., and G. Derge1937. Studies Upon the Widmanstatten Structure, VIII:

The Gamma-Alpha Transformation in Iron-NickelAlloys. Transactions of the American Institute of Miningand Metallurgical Engineers, Iron and Steel Division,125:482-500.

Meunier, Stanislas1880. Imitation synthetique des fers nickeles meteoritiques.

Bulletin de la Societe Mineralogique de France, 3:153-155.

Moore, C. B., C. F. Lewis, and D. Nava1969. Superior Analyses of Iron Meteorites. Pages 738-748

in P. M. Millman, editor, Meteorite Research. 940pages. Dordrecht: D. Reidel Publishing Company.

Olsen, E., and K. Fredriksson1966. Phosphates in Iron and Pallasite Meteorites. Geochim-

ica et Cosmochimica Ada, 30:459-470.Olsen, E., and L. H. Fuchs

1967. The State of Oxidation of Some Iron Meteorites.Icarus, 6:242-253.

Osmond, F., and G. Cartaud1904. Sur les fers meteoriques. Revue de Metallurgie (Paris),

1:69-79.Owen, E. A.

1938. Formation of Widmanstatten Figures in Meteorites.Nature, 142:999-1000.

Owen, E. A., and Y. H. Liu1949. Further X-Ray Study of the Equilibrium Diagram of

the Iron-Nickel System. Journal of the Iron and SteelInstitute, 163:132-137.

Owen, E. A., and A. H. Sully1939. The Equilibrium Diagram of Iron-Nickel Alloys. Phil-

osophical Magazine, 7th series, 27:614-636.Palache, C , H. Berman, and C. Frondel

1944. The System of Mineralogy. 7th edition, volume I, 834pages. New York: John Wiley & Sons, Inc.

Paneth, F. A.1960. The Discovery and Earliest Reproduction of the Wid-

manstatten Figures. Geochimica et Cosmochimica Ada,18:176-182.

Pauly, Hans1969. White Cast Iron with Cohenite, Schreibersite, and

Sulfides from Tertiary Basalts on Disko, Greenland.Bulletin of the Geological Society of Denmark, 19:8-26.

Perry, Stuart H.1944. The Metallography of Meteoric Iron. United States

National Museum Bulletin, 184: 206 pages.Phillips, J. P.

1965. J. Lawrence Smith, Pioneer of American Analytical

Chemistry. Journal of Chemical Education, 42:390-391.Powell, Benjamin N.

1971. Petrology and Chemistry of Mesosiderites, II: SilicateTextures and Compositions and Metal-Silicate Rela-tionships. Geochimica et Cosmochimica Ada, 35:5-34.

Ramdohr, Paul1973. The Opaque Minerals in Stony Meteorites. 245 pages.

New York: Elsevier Publishing Company.Randich, E.

1975. Non-Isothermal Diffusion Controlled Phase Growth in Ter-nary Systems and Its Application to Iron Meteorites. 125pages. Ph. D. dissentation, Lehigh University.

Randich, E., and J. I. Goldstein1975. Non-Isothermal Finite Diffusion-Controlled Growth

in Ternary Systems. Metallurgical Transactions,6A: 1553-1560.

Reed.S.J.B.1965a. Electron-Probe Microanalysis of Schreibersite and

Rhabdite in Iron Meteorites. Geochimica et Cosmochim-ica Ada, 29:513-534.

1965b. Electron-Probe Microanalysis of the Metallic Phasesin Iron Meteorites. Geochimica et Cosmochimica Ada,29:535-549.

1967. The Distribution of Phosphorus in the Mount EdithOctahedrite. Geochimica et Cosmochimica Acta, 31:1969-1974.

1969. Phosphorus in Meteoritic Nickel-Iron. Pages 749-762in P. M. Millman, editor, Meteorite Research. 940pages. Dordrecht: D. Reidel Publishing Company.

1972. The Oktibbeha County Iron Meteorite. MineralogicalMagazine, 38:623-626.

Rose, Gustav1865. Systematische Eintheilung der Meteoriten. Annalen

der Physik und Chemie ,124:193-213.Schreibers, C. Von

1820. Beytrdge zur Geschichte und Kenntniss meteorischer Steinund Metallmassen, und der Erscheinungen, welche der enNeiderfall zu begleiten pflegen (als Nachtrag zu HerrnChladni's neuestem Werke iiber Feuer-Meteore). 97 pages.Vienna: J. G. Heubner.

Scott, Edward R. D.1971a. Studies of the Structure and Composition of Iron Meteorites.

179 pages. Ph. D. dissertation, University of Cam-bridge, Cambridge, England.

1971b. New Carbide, (Fe,Ni)23C6, Found in Iron Meteorites.Nature Physical Science, 229:61-62.

Scott, E.R.D., and J. T. Wasson1975. Classification and Properties of Iron Meteorites. Re-

views of Geophysics and Space Physics, 13:527-546.1976. Chemical Classification of Iron Meteorites, VIII:

Groups IC, HE, IIIF, and 97 Other Irons. Geochimicaet Cosmochimica Acta, 40:103-115.

Shepard, Charles Upham1846. Report on Meteorites. American Journal of Science,

series 2, 2:377-392.1853. Notice of the Meteoric Iron Found Near Seneca

River, Cayuga County, N.Y. American Journal of Sci-ence, series 2, 15:363-366.

Short, J. M., and C. A. Anderson1965. Electron Microprobe Analyses of the Widmanstatten

Structure of Nine Iron Meteorites.yowrraa/ of Geophys-

Authors Copy

80 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES

icalResearch, 70:3745-3759.Short, J. M., and J. I. Goldstein

1967. Rapid Method for Determining Cooling Rates ofIron and Stony-iron Meteorites. Science, 156:59-61.

Silliman, Benjamin1886. Memoir of John Lawrence Smith 1818-1883. National

Academy of Science Biographical Memoirs, 2:217-248.Smith, Cyril Stanley

1960. A History of Metallography. 293 pages. Chicago: Univer-sity of Chicago Press.

1962. Note on the History of the Widmanstatten Structure.Geochimica et Cosmochimica Ada, 26:971-972.

Smith, J. Lawrence1855. Meteoritic Iron from Tazewell County, East Tennes-

see. American Journal of Science, series 2, 19:153-159.Smith, S.W.J., and J. Young

1938. Formation of Widmanstatten Figures in Meteorites.Nature, 142:1162.

1939. The Widmanstatten Structure of Octahedral Mete-oric Iron. Nature, 143:384-385.

Sorby, H. C.1877. On the Structure and Origin of Meteorites. Nature,

15:495-498.1887. On the Microscopical Structure of Iron and Steel.

Journal of the Iron and Steel Institute, 1:255-288.Spencer, L.J.

1916. Crystals of Iron Phosphide (Rhabdite) from a Blast-furnace. Mineralogical Magazine, 17:340-343.

Stodart, J., and M. Faraday1820. Experiments on the Alloys of Steel, Made with a

View to Its Improvement. Quarterly Journal of Science,9(18):319-330.

Swalin, R. A.1972. Thermodynamics of Solids. 2nd edition, 387 pages. New

York: John Wiley and Sons.Thomson, G.

1804. On the Malleable Iron, etc: Essai sur le fer malleabletrouve en Siberie par le Prof. Pallas. BibliothequeBritannique, 27(2, October 1804): 135-154, plate oppo-site page 184; (3, November 1804): 209-229.

1808. Sur ferro malleabile trovato da Pallas in Siberie. AttiAccademia delle Scienze di Siena, 9:37-57.

Vogel, Rudolf1927. Uber die Strukturformen des Meteoreisens und Ihre

Spezielle Beeinflussung durch Umwendlung und Bei-gemengten Phosphor. Abhandlunger der Gesellschaft

der Wissenschaften zu Gottingen, Mathematisch-physikal-ischeKlasse, 12(2): 51 pages.

1928. Ueber die Strukturformen des Meteoreisen. Archivfur das Eisenhuttenwesen, 1:605-611.

1932. Eine umfassendere Deutung der Gefugeerscheinun-gen des Meteoreisens durch das Zustandsdiagrammdes ternaren Systems Eisen-Nickel-Phosphor. Ab-handlungen der Gesellschaft der Wissenschaften zu Gottin-gen, Mathematisch-physikalische Klasse, 3(6): 31 pages.

1951. Die Gefiigeformen des Meteoreisen und ihre Er-klarung auf Grund des Zustandsdiagrammes desSystems Eisen-Nickel-Phosphor. Neues Jahrbuch furMineralogie, Abhandlungun, 83:23-52.

1952. Uber Rhabdit und Schreibersite. Neues Jahrbuch furMineralogie, Abhandlungen, 84:327-349.

1957. Uber Metabolite und ihre Kennzeichnung durch dasGefuge. Chemie derErde, 19:147-169.

1964. Uber die Entstehung des Wickelkamazit. Neues Jahr-buch fur Mineralogie, Monatshefte, 2:57-63.

Vogel, R., and H. Baur1931. Ueber das ternare System Eisen-Nickel-Phosphor.

Archiv fur das Eisenhuttenwesen, 5:269-278.von Reichenbach, C.

1861. Ueber die nahern Bestandtheile des Meteoreisens.Annalen der Physik und Chemie, 114:477-491.

Wai, C. M.1974. Geochemical Affinities of Cobalt and Germanium

toward Metal, Silicate, and Sulfide Phases at HighTemperature. Geochimica et Cosmochimica Ada,38:1821-1825.

Wasson, J. T.1974. Meteorites. 316 pages. New York: Springer-Verlag.

Waterston, Charles D.1965. William Thomson (1761-1806) a Forgotten Benefac-

tor. University of Edinburgh Journal (autumn 1965),pages 122-134.

Wood, John A.1964. The Cooling Rates and Parent Planets of Several

Iron Meteorites.Icarus, 3:429-459.Young, J.

1926. The Crystal Structure of Meteoric Iron as Deter-mined by X-Ray Analysis. Proceedings of the RoyalSociety (London), Al 12:630-641.

Ziebold, Thomas O.1967. Precision and Sensitivity in Electron Microprobe

Analysis. Analytical Chemistry, 39:858-861.

Authors Copy