- Email: [email protected]
The metallic microstructures and thermal histories of severely reheated chondrites
Geochimica et Cosmochimica Acta, 1977,
Vol.41,pp.1061to 1072.Pergamon Press. Printed inGreatBritain
The metallic microstructures and thermal histories of severely reheated chondrites B. A. SMITH*and J. I. GOLDSTEIN Jones and Laughlin Steel Corp., Aliquippa, Pa. and Dept. of Met. and Mats. Sci., Lehigh University, Bethlehem, Pa. 18015, U.S.A. (Received 5 August 1976; accepted in revised form 1 March 1977)
Abstract-The
metallographic structures of eight severely reheated chondrites (Farmington, Ramsdorf, Orvinio, Wickenburg, Lubbock, Rose City, Arapahoe and Tadjera) have been studied by optical, scanning electron microscope and electron microprobe techniques. Unreheated chondrites and experimentally heat treated chondritic material have also been examined. The following metallographic characteristics can be used to estimate the post-shock residual temperature of reheated chondrites: melted appearance of metal-troilite, presence of martensite, P enrichment of the metal, and the averaging of central metal grain compositions. Metallographic characteristics used to estimate the cooling rates of the severely reheated chondrites are the Ni content of troilite, the Ni gradients in metal grain rims, and the presence of secondary kamacite and phosphides. Farmington, Ramsdorf, Orvinio, Lubbock, Rose City and several of the heat treatment specimens have substantial P in solution in the metal grains (>O.l wt%). P enrichment is apparently caused by reduction of phosphates upon severe reheating and partial melting of metal-troilite areas in chondritic meteorites. The eight severely reheated chondrites studied showed evidence of reheating to temperatures ranging from -950°C to - 1250°C. Ramsdorf has the highest reheating temperature (12OG125o”C) and the fastest cooling rate - lOO”C/day. Wickenberg has the lowest reheating temperature (95Glooo”C) and the slowest cooling rate, - l°C/lOO yrs. Cooling rate estimates correspond to post-reheating burial depths of less than 1 to - lOOOm
INTRODUCTION
cate extensive remelting with subsequent fast cooling (BEGEMANNand WLOTZKA, 1969; TAYLOR and HEYvestigators (BUSECK et al., 1966; HEYMANN, 1967; MANN,1971), (4) Secondary kamacite which nucleates and grows WOOD, 1967; BECEMANNand WLOTZKA, 1969; TAYfrom taenite (y) during cooling (TAYLOR and HEYMR and HEYMANN, 1969, 1971). It is commonly MANN, 1971), believed that the reheating among ordinary chon(5) Relative chemical homogeneity of metal grains drites is related to shock. By studying the mineral which indicates extensive reheating of the original low and metal phases in selected shocked ordinary chonNi kamacite (LX) and high Ni taenite (y) metal phases drites, estimates of the maximum reheating temperatures and cooling rates of these samples have been (BUSECKet al., 1966; HEYMANN, 1967; BEGEMANNand made (WOOD, 1967; BEGEMANNand WLOTZKA, 1969; WLOTZKA, 1969; TAYLOR and HEYMANN,1971), and (6) Phosphorus enrichment of the metal which TAYLOR and HEYMANN, 1969, 1971). Reheating temoccurs on reheating (BEGEMANNand WLOTZKA,1969; peratures range as high as k 1350°C and cooling rates vary widely from _ O.OS”C/secto w 0.01“C/yr. Several TAYLORand HEYMANN,1971) and the growth of phosmicrostructural characteristics of the metal phases of phides during slow cooling. In recent years metallic particles from lunar rocks the shocked chondrites are used to estimate reheating and soils have been studied. Many of these particles temperatures and cooling rates. Among these charachave the same microstructural characteristics as noted teristics are for the metal in shocked chondrites (GOLDSTEINand (1) Martensite which is often called Q and indicates AXON, 1973; GOLDSTEINet al., 1972; MISRAand TAYshock reheating to 2600°C (WOOD, 1967; TAYJ.DR LoR, 1975). Therefore it appears that a better underand HEYMANN, 1971), standing of the metallic microstructure and thermal (2) Melted metal troilite microstructure which inhistories of severely reheated chondrites will also help cludes ovoid metal grains in troilite, rounded metal in the interpretation of lunar samples. This study introilite boundaries and tiny veins of metal-troilite vestigates the thermal history of eight severely within the silicate. The presence of melted meta-troishocked ordinary chondrites. In addition to using the lite structures indicates partial melting at or above metallographic characteristics already noted, equilib950°C (HEYMANN, 1967), ration temperatures of two phase kamacite-taenite, (3) Steep Ni gradients in u2 grain rims which indiand metal-phosphide assemblages (AXONand GOLDREEATED chondrites
have been studied by several in-
* Formerly graduate assistant, Lehigh University.
STEIN,1972) are estimated. 1061
1062
B. A. SETH and J. I. METHOD
Eight very strongly reheated or heavily shocked chondrites were selected for study: Farmington, Ramsdorf, Orvinio, Wickenburg, Lubbock, Rose City, Arapahoe and Tadjera. All the class 5, very heavily shocked samples studied by HEY~AANN (1967) are included in the sample. The samples were mounted as polished sections and prepared for observation using standard metallographic polishing techniques A 1% nital solution was used as an etchant
to reveal metallographic structures. An ARL electron probe microanalyzer
was employed for elemental analysis of the opaque phases. Three types of measurements were made: (1) Ni, Co and P contents of a representative sample of metal grains, (2) Ni and P profiles across zoned metal grains and across phase boundaries with secondary kamacite and phosphide, and (3) Ni contents of the troilite. The X-ray scanning mode was used to study elemental distribution between phases and to identify phosphides and phosphates. An ETEC scanning electron microscope equipped with an energy dispersive detector was used to study the fine structure of the metallic phases and to search for and detect small (< 1 q) phosphides in the metal phases. Several grams of the chondrite Knyahinya were used for the experimental heat treatments. This meteorite is a non-weathered, very lightly reheated, L5 chondrite and its structure and composition should be quite similar to that of the hypersthene chondrites befbre shock. Individual small pieces of Knyahinya were cut from the sample and heated to maximum temperatures varying from 900 to 1250°C and cooled at rates varying from lOO”C/week (2 x 10-40C/sec) to a quench (- 5oo”C/sec). The rapidly cooled pieces were annealed in a vertical molybdenum fur-
nace and were protected from oxidation by flowing argon through the furnace tube. The more slowly cooled samples were vacuum encapsulated in quartz tubes and heated in a horizontal resistance furnace. Cooling rates were controlled by a Trendtrak automatic programmer.
RESULTS AND DISCUSSION Microstructural and ,chemical criteria The following microstructural and chemical features were studied for each of the reheated chondrites:
GOLDSTEIN
martensite, melted metal-troilite microstructure, phosphorus enrichment of metal, metal grain compositions, Ni gradients at the rims of metal grains, Ni content of troilite, the presence or absence of phosphides and secondary kamacite (a), and the compositions of phosphides and secondary kamacite. Table 1 gives a summary of the major microstructural and microprobe results for the eight reheated chondrites. There are two basic groups of criteria: those that reveal information about the maximum reheating temperature caused by the shock process and those that reveal information about low temperature conditions from which cooling rates can be estimated. Martensite, melted metal-troilite microstructure, averaging of central metil grain compositions and P enrichment of the metal indicate extensive reheating and in most cases remelting of the shocked chondrites. The presence of phosphides and secondary kamacite indicates relatively slow cooling rates. Composition and width measurements of these two phases can be used to estimate cooling rates. Ni gradients at the rims of metal grains and the Ni content of troilite indicate substantial reheating, but, more critically, reveal qualitative information on cooling rates. Reheating temperatures Martensite. Martensite or a2 was observed in all eight reheated chondrites and in most of the heat treated samples. Figure 1 shows the a2 structure in Farmington. Various morphologies of martensite were observed among the samples and were apparently due to differences in metal composition. Since a2 occurs upon relatively fast cooling of y-Fe from a temperature 2800°C (TAYLOR and HEYMANN,1971), the presence of a2 in the chondrites indicates that these meteorites experienced temperatures 2 800°C. The presence of martensite apparently reveals no information about the maximum temperature of reheat-
Table 1. Summary of key results from reheated chondrites
u Ramsdorf
OWl”i0
Yes
Rounded met-tr boundaries, blebs
0.13
0.08
Yea
Yes
NUN” 5749
Yes
Extensive; cN0‘d grains, blebs
0.45
0.11
*a
Yes
Yes
Extensive; ovoid grains, blebs
0.49
0.08
No
*es
Yes
burectic met-tr, blebs Rounded met_-tr boundrles, ~o,ne ovoid grains
0.06
0.02
Yea
0.17
0.06
Yes
Yes
>0.39
0.08
Yes
Yes
Yes
No
H6
L5
Tadjera
NEW
BN 1959,1023 BM 1959,887
Yes, Sore Decomposed
(H)
B” 1959,768
Decompo*ed
IS
mm” 2315 B” 1959.827
I.5
Extensive, rounded met-t= bcundaries
Yea
Rounded met-Lr boundaries, (lome maids blebs, veinlets
0.02
< 0.02
Yell
sons ovoid gcsins, sore rounded rat-tr boundaries, blebs, veinlets
0.04
0.31
Classification according to VAN Sc~rms and Wool (1967). $ BM: British Museum, NMNH: National Museum of Natural History.
*
No
No
No
Metallic microstructures
ing or about the cooling rate except that the shocked chondrites have not cooled at an extremely slow rate such as that experienced by the iron meteorites. Melted metal-troilite microstructure. The appearance of melted metal-troilite varied among the reheated chondrites and the heat treated samples. Ovoid metal grains agglomerated within troilite indicate extensive remelting. These structures are common in Ramsdorf (BEGEMANN and WLOTZKA, 1969), Orvinio (TAYLORand HEYMANN,1971), Rose City, and several of the heat treated samples (see Fig. 2), Metal in Lubbock, Farmington, Arapahoe and Tadjera typically occurs in irregularly shaped grains with rounded metal-troilite boundaries (see Fig. 1). Metaltroilite interfaces in Wickenburg are intermixed in a eutectic structure previously observed by TAYLORand HEYMANN(1971), Fig. 3. All of the shocked chondrites contain tiny blebs of metal-troilite dispersed in the silicate. Veinlets are plentiful in Tadjera, Arapahoe, and the moderately remelted samples of Knyahinya (Fig. 4). It appears that veinlets occur most plentifully in samples which were not extensively remelted and/or cooled fairly quickly. Samples of Knyahinya heated to a maximum temperature of 900°C showed very few signs of remelting. Therefore 900°C is probably a lower temperature limit for chondritic metal-troilite melting. On the basis of the observed metal troilite microstructure, Ramsdorf was the most extensively remelted chondrite followed by Orvinio, Rose City, Lubbock, Tadjera, Arapahoe, Farmington and Wickenburg. Phosphorus enrichment. Farmington, Lubbock, Orvinio, Ramsdorf and Rose City contain substantial P (> 0.10 wt%) in the metal phase (Table 1). Wickenburg, Tadjera and Arapahoe have low P contents (co.10 wt%) and compare with unreheated chond&es in this respect. Some of the artificially heated samples contain substantial P in the metal phase (0.154.83 wt%). Since the Knyahinya starting material contained co.01 wt% P in the metal phase, these experiments showed that it is possible to enrich the metal phase in P upon reheating. The reduction of phosphate is dependent upon the oxygen fugacity,fO,, which must be low enough for a reduction reaction to occur (OLSEN and FUCHS, 1967; FRIELand GOLDSTEIN,1976). OLSENand FUCHS (1967) proposed the following reduction reaction 3CaMg(SiO,),
+ 402 + 3Fe + 2(P)F 3 Ca3(PO& + 3MgFe(SiO,),
(1)
where (P,” is the P content in the metal phase. This reaction can explain the relation of P in solution in the metal phase to metal-phosphate equilibrium in iron meteorites. TAYL,ORand HEYMANN(1971) proposed a similar reaction for the reheated chondrites. Data from WILLIAMS(1971) indicate, however, that the f0, of ordinary equilibrated chondrites is too high for reduction via the above reaction to occur. Presumably during the shock process the local f0, in the chondrites is low enough to make reduction
and thermal histories
1063
favorable. This low f0, may be caused by the presence of a reducing agent, such as C or CO (TAYLOR and HEYMANN,1971). The P reduction in the experimental studies using the chondrite Knyahinya, may have been controlled by the presence of tantalum inside the sample capsules. The tantalum was used as an oxygen getter to protect the metal. The Ta-Taz05 equilibrium reaction probably buffered the f0, to well below the value needed for P reduction. Once the f’0, is favorable for P reduction, the amount of P which dissolves in the metal phase is probably most dependent on the volume ratio of phosphate to metal and the residual temperature (TR) or maximum temperature of the chondrite after the shock event. The kinetics of P diffusion and the intimate contact of phosphate and metal would be increased during a partial melting stage. In general, the presence of increased P in the metal of reheated chondrites is consistent with other evidence of severe reheating. On the basis of the observed amount of P in the metal grains, Rose City, Orvinio and Ramsdorf have the highest maximum reheating temperatures followed by Lubbock, Farmington, Wickenburg, Tadjera and Arapahoe. Metal grain center compositions. Previous investigators have noted that the Ni contents of metal grains in reheated chondrites are often similar from grain to grain (HEYMANN,1967; TAYLORand HEYMANN, 1971; Begemann and WLOTZKA,1969). The results of our measurements are illustrated in Fig. 5, where histograms of the number of grains vs central Ni content for an unreheated chondrite, Wellman, and for Orvinio, Farmington and Tadjera are given. Reheating-shock remelting apparently has the effect of averaging out the original compositional differences between grains found in unreheated chondrites (Fig. 5a). BEGEMANNand WLQTZKA (1969) proposed that such averaging out was due to melting, mixing, and redistribution of metal during shock reheatingremelting. This proposal is supported by the results of this study. The degree of compositional similarity of grain centers varies among the reheated chondrites studied (compare Fig. 5b-d) and coincides with the degree of remelting. Chondrites with extensively remelted metal-troilite (Ramsdorf, Orvinio, Rose City) have a narrow range of central Ni contents while the least remelted-reheated samples (Tadjera, Arapahoe, Farmington, Wickenburg) generally show a wide scatter. The veinlets observed in the least reheated chondrites (Fig. 4) indicate that molten metal and sulfide seeped through cracks in the silicate matrix and helped to some degree to mix the metal. For the more extensively remelted chondrites, molten metal and sulfide were able to mix thoroughly allowing a homogenization of the molten metal. Some time must have been available for this mixing to occur. Such a process however must proceed without allowing the metal-sulfide melt to separate from the surrounding silicate. Only minor amounts of solid state
B. A. SMITHand J. 1. GOLIISTEIN
1064
By comparing and integrating results from the above criteria for the eight reheated chondrites, a relative ranking according to maximum reheating temperature can be made. Ramsdorf has the highest reheating temperature followed by Orvinio, Rose City, Lubbock, Farmington. Tadjera, Arapahoe and Wickenberg.
“5
10
15
20
25
30
35
W t
%
40
45
50
55
Ni
14
Orvinio(lt.1
-f 12 .y (3 z
10 8
L 4
6
;E
4 2
0
5
10
15
20 w t %
14
25
35
40
(CL) Farmington
t In
30
Ni
I
’ zt
0
5
IO
15
20
25
30
35
40
wt % Ni
14
E z’
t
Cd)
4 2 0
5
15
20
25
30
35
40
wt % Ni Fig. 5. (a) Wellman histogram of number of grains vs Ni content at grain centers. 23 grains measured; (b) Orvinio histogram of number of grains vs central Ni contents. 20 grains measured; (c) Farmington histogram of number of grains vs central Ni content. 20 grains measured; (d) Tadjera histogram of number of grains YS central Ni contents. 25 grains measured.
diffusion have contributed to the averaging out of the metal compositions even in the very slowly cooled chondrites.
Ni rim yrudienrs. An increase in Ni at metal grain edges or rims of the reheated chondrites was measured (Table 2). The largest gradients are observed in Ramsdorf. Orvinio, Lubbock and Tadjera. Ni and P profiles are plotted in Fig. 6a for typical grain rims in Ramsdorf and in Fig. 6b for heat treated sample I5 (I 250°C max temp, lOO”C/day cooling rate). The Ramsdorf gradients are similar to those observed by BEGEM~NNand WLOTZKA (1969). TAYL~K and HEYMANN (1971) reported similar gradients in Orvinio and Lubbock. BEGEMANN and WLOTZKA (1969) proposed that such gradients are due to Ni segregation during solidification of an Fe-Ni-S melt and this proposal is supported by the experimental heat treatments. Given enough time or a slow cooling rate at high temperatures. the Ni gradients would be expected to be leveled out by solid state diffusion. Ni rim gradients are therefore indicators of relatively fast cooling. in accord with the steep Ni gradients observed in the artificially heated samples where the maximum temperature was > 1050°C and the cooling rates were faster than lS”C/day (Fig. 6b). Ni in troilite. The Ni contents of troilite in the reheated chondrites vary from ~0.02 to 0.31 wt:,, (Table 1) and are higher than those of unreheated chondrites, which contain <0.02 wtO/,,. Studies by KULLEKUD (1963) indicate that the Ni solubility in troilite decreases from
1065
Fig. 1. Typical Farmington metal grain. c(~ martensite (labeled M) is predominant structure. Troilite (T) and secondary c( (1) arc also present. Dark grey matrix is silicate. Nital etch. Marker = 25 pm. Fig. 2. Typical
Ramsdorf surrounds
metal grain. Most of grain is c(~. Light-etching high Ni rim (open I~, Troilite surrounds the metal. Nital etch. Marker = 25 pm.
arrow)
Fig. 3. Metal and troilite of Wickenburg. The metal and troilite are finely intermixed in the central area of photo, indicating a eutectic structure. (Metal is lightest phase, troilite grey.) Major mass of metal is (Ye. No etch. Marker = 1Opm. Fig. 4. Photo of Arapahoe phase). Marker = 25 pm.
Fig. 9. Metal
and
featuring veinlets (arrows) of metal and/or troilite in silicate matrix Metal predominantly G(~ with light-etching (high Ni) rim. Nital Marker = 25 pm. troilite
in heat
treatment
sample
No. 15 (125O”C, lOO”/day).
(grey etch.
Metallic microstructures and thermal histories
1067
Table 2. Ni rim gradients Average Compos
Rim
Ni
P
ition (wt.%)*
Specimen
MO, Grains
Farmington Remsdorf
11 10
Orvinio
9
Wickenburg Lubbock
6 i3
Rose City
6
Arapahde
5
Tadjera
11
Average Center
Maximm
Composition (wt.%) Ni P
Variation of Ni, center to Rim
14.1 0.12
12.4 0.14
-2-49.
21.4
10.3 0.41
-8-20
0.34
12.5 0.33
8.0
0.44
i-8
15.2 0.030
13.4 0.038
-1-Z
15.6 0.17
11.2 0.17
-2-10
8.9 14.2
0.37 0.510
16.5 0.045
8.6 13.5
0.60 0.013
10.9 0.046
* Counts were taken at edges of either a2 st~ctures rims of a2 grains. Secondary a was avoided.
samples contained more than 0.15 wt% Ni. From these results the cooling rates of the shocked chondrites are slower than 25”C/hr. Secondary kamacite (a). Secondary kamacite (a), was
<1 %.1-Z
-2-20
or ii~t-etching
formed from tbe y phase during cooling from maximum shock temperatures. It was observed in Farmington, Wickenburg, Rose City, Lubbock and Arapahoe (see Fig. 1). This phase was observed previously
HT No.15(l2~C,lO~~y)
Groin Rim D Ni *P
&
.
-hm--
Distance Fig. 6. (a) Ni and P profiles across metal grain rim in Ramsdorf, Vertical dashed line is troilite (Tr)metal interface; (b) Ni and P profiles across metal grain rim in heat treatment sample No. 15 (125O”C, lWC/day).
B. A. Sbirm and J. I.
1068
Table 3. Description of two phase
Bulk Composition Sainple
Si
P
(a/y
and phosp~de/me~l) particles in reheated chondrites
interface
Compositions
Precipitating
Phase
Type
GOLDSTEIN
Xi
of ?.-phase
*tmftum Other
P
Type
R05e City
9
oi
6. LM.ZO
10
0.02
c(
6.7t0.20
tubbcch
1,
0.17
a
7.0t0.10
0.24ta.03
Fumington
12
0.13
a
7‘010.20
0.1710.03
Y
.19
Wickenburg
13
0.06
a
6.6M.20
0.09t0.02
Y
>1s
-
-
!&se
9
Y Y
-------
City
Farmington
0.3o*o.clS -0.02
12
0.13
*
-575’C
Fe-Zi-F
-sso*c
Fe-Xi
*
-sso*c
Y
Fe-Xi-P
0.08
-save
Fe-Xi-P
0.04
-SOO%
Fe-XI-P
0.27
-S?SY
Fe-Xi-P
FeJi-P
neg1. *
-_-0.50
Teq
* .I8
Phase Diagrm Employed
P
Ni
Arqpahoe
-0.50
(wt.%)
Phase
____
PII
22t2
IS.5
a
-6.0
ph
-3s
15.5
(I
-6.0
co.2
_
* Interface values too difficult to obtain. 2S’C/hr show Ni contents below 0.3 wt%. Even at a slow cooling rate of 4”C/hr, the experimental in Farmington by WOOD (1967) and BUS&X et al., (1966) and in Wi~enbmg by TAYLORand HEM&ANN (1971). Secondary karnacite typically occurs along the edges or grain boundaries of a2 with silicate, troilite or other metal grains since these locations provide favorable sites for heterogeneous nucleation (Fig. 1). The presence of secondary a indicates that the cooling rate of the metal was slow enough to allow x to nucleate and grow from y during cooling. The presence of P in the metal facilitates bt nucleation and growth, since P is an ELstabilizer (~~~~ and DOAN, 1972), and increases the diffusion coefficient of Ni in Fe-Ni (HOWARD and GOLDSTEIN,1973). The Ni and P ~s~butions at crly interfaces can be used to approximate the final temperature of equilibration (T,,) of the two phase assemblage a-y. The method employed by AXONand GOLDSTEIN(1972) for dete~ning T,, of lunar two-phase particles was applied in the present work. Interface Ni and P compositions of a were fit to the approp~ate Fe-Ni or Fe-Ni-P phase diagram to estimate 7& Point analyses were made using the microprobe along a line crossing the interface between the two phases. Figure 7 shows such a profile in the Fa~~gton chondrite.
interface compositions and ‘I& estimates are listed in Table 3 for those chondrites which contain a. The Ni composition on the y side of the interface near the a/y interface decreases rapidly (Fig 7) due to low diffusion rates in the y. It is therefore difIicult to obtain an accurate interface value for y. Accordingly T_ estimates are based mainly on the a ~m~sitions. The values of 7& can be used to give a relative ranking of chondrite cooling rates; a sample with a lower I& has a slower cooling rate. Cooling rate estimations were made by calculating the times of secondary kamacite growth. The following assumptions were made for the growth calculation: (1) growth is diffusion controlled; (2) chemical ~uilibrium exists at the o/y interface; (3) growth is one dimensional; (4) no imp~ge~nt occurs, and (5) growth can be considered as having occurred at one tempemture, called the optimum growth temperature. With these assumptions, the following analytical solution can be used to solve Fick’s second law for the Ni con~n~ation in CIand y:
C&Gtf = 4 + 4
Farmingt~ ONi
l
, . ~+***= l
lL******..* i
.
*%I=Dirtonce
Fig 7. Ni and P profiles across secondary doin Far~n~on.
erf
(2)
1069
Metallic microstructures and thermal histories
ture at which these effects balance is called the optimum growth temperature. At this temperature the s( phase growth will be a maximum. The optimum temperature for each case was found by performing isothermal calculations over a range of temperatures between the theoretical nucleation temperature and Ter Once the optimum growth temperature was found for each meteorite, calculations were performed to find the time needed to grow c( to a width of the order of that observed in the sample. Due to geometry considerations accurate u. phase widths could not be measured and a best value was estimated. The tem~raturc range between the nucieation temperature and T, was divided by the calculated time needed for growth at the optimum temperature. This result yields an approximate cooling rate for each meteorite. These cooling rates vary from - l”C/yr to l’C/lOO yr. Table 4 lists such cooling rates along with the temperatures and times employed in the calculations. No secondary M was observed in the heat treated samples. This is not surprising as the experimental cooling rates were much faster (>50WC/yr) than those calculated for the shocked chondrites. P~osp~ides. Phosphides were observed in Ramsdo& Orvinio, Rose City, Far~ngton, and Lubbock (all the chondrites with P > 0.1 wty< in the metal). This phase was previously observed in Ramsdorf (BEGEMANN and WLOTZKA,1969) and Orvinio (TAYLORand HEYMANN,1971). The phosphides apparently nucleate mostly at the a-troilite or former y-troilite boundaries during cooling. Rose City and Farmington contain phosphides large enough (72 pm) for compositional measurements across phosphide/metal interfaces. Ni and P profiles across phosphides in these two meteorites are shown in Fig. 8. By applying the method of AXONand Got~srrn*; (1972), 7&‘s for Rose City and Farmington phosphide/me~l were estimated as -575 and <55O”C respectively (Table 3). The phosphides in Lubbock are generally - 2 pm wide and the phosphides in Ramsdorf and Orvinio are <2 pm wide. Rose City, Lubbock and Farmington have the largest phosphides, contain secondary kamacite, and therefore have slow cooling rates,
where A, =
a’ - d
c’ - a’ erf y
1 - erfy ’
B, = ~
1 - erfy’
D is the diffusion coefficient, x the distance, t the growth time, C, the composition in the taenite parent phase, c” the equilibrium Ni content of the precipitate, c’ the equilibrium Ni content of the parent phase, a’ the bulk Ni content of parent phase, and y a dimensionless parameter (JOST,1952). The location of the interface between the precipitate and parent phases, 5 is given by:
5 = 2&/V%
(4)
(JOST, 1952). The y term is found by inserting the general solutions for C, and C,, equations (2) and (3) into the mass balance expression
where
is the concentration 5- and
gradient in j at the &‘a interface, aC,
ax {+ (--_) is the concentration gradient in c( at the ~JOZinterface 4’. After minor rearranging of terms one has: $&eY’)(l
- erfy) = g,
which is solved for y iteratively using a standard Newton-Raphson technique (MCCRACKEN,1965). The interface distance or precipitate width < can then be calculated for desired times using equation (4). The diffusion coefficients used in this calculation were obtained from HEYWARDand GOLDSTEIN (1973). There are two variables that control precipitate growth, difTusion coefficient and driving force, which have opposing effects with respect to temperature. The diffusion coefficient decreases with decreasing temperature and the driving force increases with decreasing temperature (REED-HILL,1964). The tempera-
Table 4. Estimates of cooling rates for reheated chondrites, based on the growth of secondary kamacite T
sampie
Growth
I
ac
nucl.
time Width4
Tmx
Cooling
700-c
575’C
675’C
100
yrs.
10 jmI
.kapsb?
700*
550°
685’
1000
yrs.
8 9”
>lY/lO
Lubbock
650°
5500
600
1000
yrs.
7 w
-1’Cl
Farmingron
6509
500*
boo0
I.000
yrs.
8 P
-1w10
Wickenburg
6759
5000
625’
Rose
city
10,000
yrs.
r Approximate nucleation temperature, no undercooling. * Approximate final equilibrium temperature, Table 3. 3 Calculated optimum growth temperature. 4 Approximate secondary kamacite width.
10$m
Rate
-1wyr. yrs. to
yrs. yrs.
B. A. SMITH and J. 1. GOLDSTEIN
upon resolidification. Also the presence of phosphides in a heat treatment sample 1050°C. lOO”C/day indicates that phosphides can occur in metal which cooled as fast as IOO”C/day.
THERMAL HISTORIES OF REHEATED CHONDRITES
._
Farmington 0 NI .P
40. Tr 35
0
..P
I I Ph I I I no I
1
o(
.
1
.20
@z
I
I. ..I_._.
I,
*lOUrnDistance Fig. X.(a) Ni and P profiles across phosphide (Ph) in Rose City: (b) Ni and P profiles across phosphide (Ph) in Farmington.
0.1~-l”C/yr (Table 4). The presence of phosphides in Ramsdorf and Orvinio, although they are smaller, might indicate slow cooling for these meteorites. However these two meteorites have been extensively remelted, they have the highest reheating temperature 2 1200°C and the highest P content in the metal. It is possible that the phosphides formed during the remelting stage and a small amount of phosphidee metal eutectic nucleated at the edges of the troilite
Ramsdorf has probably experienced the highest reheating temperature of the chondrites studied. The metal troilite assemblages indicate extensive remelting. The central Ni contents of the metal grains are similar to each other and the high P content of the metal also indicate a high shock-reheating temperature. A comparison of the metal structure of Ramsdorf with the experimental heat treatment samples (Fig. 9) indicates a reheating temperature of - 120th I25O”C. Troilite in Ramsdorf contains an average Ni content of 0.1 I wt”,,. which indicates a relatively fast cooling rate. The presence of steep Ni rim gradients and the absence of secondary 3: also indicate a relatively fast cooling rate. A comparison of Ramsdorf with the structure of the heat treated samples (Fig. 9) and the results from cooling rate criteria study lead to an estimate of - IO0 C/day as the cooling rate for Ramsdorf. BEGMANN and WLOTZI(A (1969) propose a cooling rate of - 3 C/min at a reheating temperature 01 > 1200°C. Such a value was based on a calculation of solid state diffusion in the grain rims of Ramsdorf at tcmperaturcs above 1000 C. Evidence from our heat treatment studies shows that metal-troilite melts at a temperature of IOOO‘C and the rims of the metal grains in Ramsdorf did not solidify until they cooled below 1000°C. This result invalidates the solid state diffusion method for T > looO”C. Calculations of BEGEMANN and WLOTZKA (1969) indicate however that a cooling rate of - 100 C/day is possible if solidification occurs below 1000 ‘C. Orvinio has a metal-troilite structure which appears to have been extensively remelted. Results from the study of high temperature criteria in Orvinio and comparison to the heat treated sample lead to a maximum reheating temperature of - 1200°C. The cooling rate of Orvinio must have been fast enough to prevent secondary x growth and also to retain some Ni rim radients. The Ni content of troilite (0.08 wtf, av) is somewhat lower than that of Ramsdorf indicating a slower cooling rate. Orvinio’s cooling rate is estimated as - IOO’C!day to - I S”C/day. Tadjera contains only some metalLtroilite that appears remelted. The metal in this sample contains a relatively low amount of P, and the central Ni content of the metal grains varies greatly (Figure 4). Tadjera contains a high density of veinlets, which again indicates a less severe remelting process. The maximum reheating temperature is estimated as - looO-1050°C. The relatively high Ni content of troilite (0.31 wt”; av), the inhomogeneity of individual metal grains (including some Ni rim gradients)% and
Metallic microstructures
to-3 IO4 IO5 10-6 Icr7
-* 130d0
1071
and thermal hislories
10-E
lo+
10’0
Ra 1200 -
-tt,
,,Rc
Or
TR 1100 YC) 1000 - +Pa
900
1
I
I
1
I
IO
100
1000
Approximate
Depth of Burial
(ml
Fig. IO. Thermal history summary for eight severely reheated chondrites (samples identified by their initials).
the absence of secondary ct indicate a relatively fast cooling rate for Tadjera, probably on the order of LOO-500”C/day. Wickenburg appears to have been the least reheated and slowest cooled of the reheated chondrites studied: The presence of a metal-troilite eutectic (Fig. 3) may well indicate reheating to a temperature just above the Fe-Ni-S eutectic temperature (950-1000°C). In addition the metal has a fairly low P content, 0.55 wt% av. The maximum reheating temperature for Wickenburg is estimated as -9% 1000°C. Fa~ington displays a moderate degree of remelting (evidenced by rounded meal-troilite boundaries), grains with fairly similar central Ni content, and moderate P enrichment (0.13 wt% av). The maximum reheating temperature for Farmington is estimated as _ 1050-1100°C. Rose City has an extensive remelted appearance, a high P content in the metal (0.39 wt% av), and shows little variation of the central Ni content of the metal grains. The maximum reheating temperature is similar to that of Orvinio, + 1200°C. Lubbock has been moderately remelted, its remelt structure and composition falling between Orvinio and Farmington. The ~x~urn reheating temperature is estimated as - 110&l 150°C. The low Ni content of troilite (0.04 wt% av), the presence of secondary CIand the presence of phosphides in the metal indicate a relatively slow cooling rate. Some grains in Lubbock however contain Ni rim gradients. This discrepancy can be at least partially explained by the presence of P; those grains with Ni rim gradients generally contain SO.15 wt% P, while those grains with secondary c( contain 20.20wt% P. Increased P has the effect of increasing the metal phase diffusion coefficients (HEYWARDand GOLDSTEIN,1973). Arapahoe has a large number of veinlets, the P content of metal is very low (0.016 wt”/, av) and the central Ni contents of the metal grains have only a fair degree of similarity. The maximum reheating temperature is estimated as -- IOOO-1050°C.
A sum~ry of the estimated thermal histories of the eight reheated chondrites is presented in Fig. 10. The maximum reheating temperatures and cooling rates of the chondrites are obtained from the preceding discussions of appropriate criteria and the cooling rate data from Table 4. Error bars indicate the degree of uncertainty in the estimated thermal histories. Ramsdorf, Orvinio and Tadjera have relatively fast cooling rates while the remaining chondrites cooled slowly (< 1“C/yr). Previous investigators have related the cooling rates of the reheated chondrites to their depth of burial after the shock event (BEGE~NN and W~TZ~, 1969; TAYLOR and HEY~NN, 1971). Material with a slower cooling rate is buried deeper in the parent body. Correlations between cooling rate curves and burial depths have been made by TAYLOR and HEYMANN(1971) and their data is applied to the cooling rate estimates of the present study. The bottom scale in Fig. 10 indicates the depth of burial of the eight reheated chondrites. These chondrites were apparently buried at depths ranging from -0.5 to 1000 m beneath the surface of their parent bodies. CONCLUSIONS (1) The following me~llo~aphic criteria can be used to estimate the post shock residual temperature of severely reheated chondrites: melted metal-troilite appearance, presence of martensite, phosphorus enrichment of metal and averaging of central metal grain compositions. (2) The presence of phosphides and secondary kamacite are due to slow post-shock cooling rates. These exsolved phases can be used to estimate cooling rates. Ni rim gradients indicate both extensive remelting of metal grains and relatively fast cooling. (3) Ramsdorf had the highest post shock residual tem~rature (12~125O’C) followed by Orvinio and Rose City f- 1200°C), Lubbock ( - 1LO@-1 150°C), Farmington (1050-1 lOO*C), Tadjera and Arapahoe (lC@&1050°C), and Wickenburg (950-1000°C).
B. A. SMITH and J. 1. GOLIISTW
1072
(4) Estimates of cooling rates are: Tadjera - 100500’ C/day, Ramsdorf x IOOLC/day. Orvinio . 100 C/day to - 15’ C/day, Rose City - l’C/yr. Lubbock. Farmington and Arapahoe - l”Cjl0 yr and Wickenburg - I ‘C/100 yr. These chondrites were buried at depths ranging from 0.5 to IOOOm beneath the surface of their parent bodies.
Ack,low/edgemerlts The H~,TWISON of the British
authors
wish to thank Dr. R. Museum. Dr. F. WLOTZKA. Max Planck Institut fiir Chemie, and Dr. R. CLARKE. JR.. of the U.S. National Museum, for their help in selecting meteorite samples for this study. We also wish to thank Dr. S. 0. AC;RLLL for assistance with the optical microscopy, Mr. D. BUSH for assistance with the electron probe work. and Dr. F. BEGEMANN for the careful review of the manuscript. This work was supported by NASA Grant NGR 39-007-043.
REFERENCES AXON H. J. and GOLDSTEIN J. I. (1972) Temperature-time
relationships
from
Earth
Sci. Lett.
Planet.
lunar
two
phase
metallic
particles.
16. 439441.
BEGEMANNF. and WLOTZKAF. (1969) Shock induced thermal metamorphism and mechanical deformations in the Ramsdorf chondrite. Geochim. Cosmochim. Actu 33. 1351.-1370. BLSECK P. R., MASON B. and WIIK H. B. (1966) The Farmington meteorite. Mineralogy and chemistry. Geochim. Cosmochirn. Actu 30. 1 8. FRIEL J. J. and G~LIXTEIN J. I. (1976) An experimental study of phosphate reduction and phosphorus-bearing lunar meial particles. Proc. 7th Lunar Sci. Co@, Geochirn. Cosmochim. Acta Suppl. 7. pp. 791X306. Pergamon Press. GOLDSTEIN J. 1. and AXON H. J. (1973) Composition, structure and thermal history of metallic particles from 3 Apollo I6 soils. 65701,685Ol and 63501. Proc. 4th Lunur
Cor$, Geochim. Cosmochim. .4ctu Suppl. 4, pp. 751 775. Pergamon Press. GOLIISTEIN J. I., AXON H. J. and YE& C. F. (1972) Metallic particles in the Apollo I4 lunar soil. Proc. 3rd Lunar Sci. Cor$. Geochim Cosmochim. kta Suppl. 3, pp. 1037-~1064. Pergamon Press. GOLDSTEIN J. 1. and DOAN A. S.. (1972) The effect of phosphorus on the formation of the Widmanstltten pattern in iron meteorites. Geochim. Cosmochim. Acta 36, 51-69. HEYMANN D. (1967) On the origin of hypersthene chondritcs: ages and shock effects of black chondrites. Icarus 6. 189-221. H~~WARI) T. R. and GOLIXTEIN J. I. (1973) Ternary diffusion in the z and ;‘ phases of the Fe-Ni- P System. Mrttrllurgical Trunsu( tion.\ 4. 2335 2342. JOST W. (1952)DifSusior~. Academic Press. KULLERI:~ G. (I 963) The Fe--Ni -S system. Cwnegie Institute Yearbook 62, I75 I X9. MCCRACKEN D. D. (1965) ,4 Guide to FORTRAN IV’ Programming. Wiley. MISRA K. C. and TA’~I.OK L. A. (1975) Characteristics of metal particles in Apollo I6 rocks. Proc. 6th Lunur Sci. Conf. Grochim. Cosmochim. .4ctu Suppl. 6. pp. 615-639. Pergamon Press. OLSF-N I!. and FC’CHS L. H. (1967) The state of oxidation of some iron meteorites. Icarus 6. 242-253. REEI)-HILL R. E. (I 964) Ph!micu/ Metullurg~~ Principles. Van Nostrand-Reinhold. SWALSN R. A. (I 972) Thermod~numics of’ Solids, Wiley. TAYLOR G. J. and HI:~MANN D. (1969) Shock. reheating and the gas retention ages of chondritcs. Earth Planet. %I. Lrtt. 7. 151 ~161. TAYLOR G. J. and H~YMANN D. (1971) Postshock thermal histories of reheated chondrites. J. Geophys. Rex 76. 1879 1x93. VAN SWMLIS W. R. and WOOI, J. A. (1967) A chemical petrologic classification for the chondritic meteorites. Sci.
krochirn.
Cosmochiin
Acta 31. 747-765.
WILLIAMS R. J. (1971) Equilibrium temperatures. pressures and oxygen fugacities of the equilibrated chondrites. &ochitn Cosmochim. .Acfcc 35. 407.-4 I I. WOOII J. A. (1967) Chondrites: their metallic mmerals. thermal histories, and parent planets. lcurus 6. I -49.
Vol.41,pp.1061to 1072.Pergamon Press. Printed inGreatBritain
The metallic microstructures and thermal histories of severely reheated chondrites B. A. SMITH*and J. I. GOLDSTEIN Jones and Laughlin Steel Corp., Aliquippa, Pa. and Dept. of Met. and Mats. Sci., Lehigh University, Bethlehem, Pa. 18015, U.S.A. (Received 5 August 1976; accepted in revised form 1 March 1977)
Abstract-The
metallographic structures of eight severely reheated chondrites (Farmington, Ramsdorf, Orvinio, Wickenburg, Lubbock, Rose City, Arapahoe and Tadjera) have been studied by optical, scanning electron microscope and electron microprobe techniques. Unreheated chondrites and experimentally heat treated chondritic material have also been examined. The following metallographic characteristics can be used to estimate the post-shock residual temperature of reheated chondrites: melted appearance of metal-troilite, presence of martensite, P enrichment of the metal, and the averaging of central metal grain compositions. Metallographic characteristics used to estimate the cooling rates of the severely reheated chondrites are the Ni content of troilite, the Ni gradients in metal grain rims, and the presence of secondary kamacite and phosphides. Farmington, Ramsdorf, Orvinio, Lubbock, Rose City and several of the heat treatment specimens have substantial P in solution in the metal grains (>O.l wt%). P enrichment is apparently caused by reduction of phosphates upon severe reheating and partial melting of metal-troilite areas in chondritic meteorites. The eight severely reheated chondrites studied showed evidence of reheating to temperatures ranging from -950°C to - 1250°C. Ramsdorf has the highest reheating temperature (12OG125o”C) and the fastest cooling rate - lOO”C/day. Wickenberg has the lowest reheating temperature (95Glooo”C) and the slowest cooling rate, - l°C/lOO yrs. Cooling rate estimates correspond to post-reheating burial depths of less than 1 to - lOOOm
INTRODUCTION
cate extensive remelting with subsequent fast cooling (BEGEMANNand WLOTZKA, 1969; TAYLOR and HEYvestigators (BUSECK et al., 1966; HEYMANN, 1967; MANN,1971), (4) Secondary kamacite which nucleates and grows WOOD, 1967; BECEMANNand WLOTZKA, 1969; TAYfrom taenite (y) during cooling (TAYLOR and HEYMR and HEYMANN, 1969, 1971). It is commonly MANN, 1971), believed that the reheating among ordinary chon(5) Relative chemical homogeneity of metal grains drites is related to shock. By studying the mineral which indicates extensive reheating of the original low and metal phases in selected shocked ordinary chonNi kamacite (LX) and high Ni taenite (y) metal phases drites, estimates of the maximum reheating temperatures and cooling rates of these samples have been (BUSECKet al., 1966; HEYMANN, 1967; BEGEMANNand made (WOOD, 1967; BEGEMANNand WLOTZKA, 1969; WLOTZKA, 1969; TAYLOR and HEYMANN,1971), and (6) Phosphorus enrichment of the metal which TAYLOR and HEYMANN, 1969, 1971). Reheating temoccurs on reheating (BEGEMANNand WLOTZKA,1969; peratures range as high as k 1350°C and cooling rates vary widely from _ O.OS”C/secto w 0.01“C/yr. Several TAYLORand HEYMANN,1971) and the growth of phosmicrostructural characteristics of the metal phases of phides during slow cooling. In recent years metallic particles from lunar rocks the shocked chondrites are used to estimate reheating and soils have been studied. Many of these particles temperatures and cooling rates. Among these charachave the same microstructural characteristics as noted teristics are for the metal in shocked chondrites (GOLDSTEINand (1) Martensite which is often called Q and indicates AXON, 1973; GOLDSTEINet al., 1972; MISRAand TAYshock reheating to 2600°C (WOOD, 1967; TAYJ.DR LoR, 1975). Therefore it appears that a better underand HEYMANN, 1971), standing of the metallic microstructure and thermal (2) Melted metal troilite microstructure which inhistories of severely reheated chondrites will also help cludes ovoid metal grains in troilite, rounded metal in the interpretation of lunar samples. This study introilite boundaries and tiny veins of metal-troilite vestigates the thermal history of eight severely within the silicate. The presence of melted meta-troishocked ordinary chondrites. In addition to using the lite structures indicates partial melting at or above metallographic characteristics already noted, equilib950°C (HEYMANN, 1967), ration temperatures of two phase kamacite-taenite, (3) Steep Ni gradients in u2 grain rims which indiand metal-phosphide assemblages (AXONand GOLDREEATED chondrites
have been studied by several in-
* Formerly graduate assistant, Lehigh University.
STEIN,1972) are estimated. 1061
1062
B. A. SETH and J. I. METHOD
Eight very strongly reheated or heavily shocked chondrites were selected for study: Farmington, Ramsdorf, Orvinio, Wickenburg, Lubbock, Rose City, Arapahoe and Tadjera. All the class 5, very heavily shocked samples studied by HEY~AANN (1967) are included in the sample. The samples were mounted as polished sections and prepared for observation using standard metallographic polishing techniques A 1% nital solution was used as an etchant
to reveal metallographic structures. An ARL electron probe microanalyzer
was employed for elemental analysis of the opaque phases. Three types of measurements were made: (1) Ni, Co and P contents of a representative sample of metal grains, (2) Ni and P profiles across zoned metal grains and across phase boundaries with secondary kamacite and phosphide, and (3) Ni contents of the troilite. The X-ray scanning mode was used to study elemental distribution between phases and to identify phosphides and phosphates. An ETEC scanning electron microscope equipped with an energy dispersive detector was used to study the fine structure of the metallic phases and to search for and detect small (< 1 q) phosphides in the metal phases. Several grams of the chondrite Knyahinya were used for the experimental heat treatments. This meteorite is a non-weathered, very lightly reheated, L5 chondrite and its structure and composition should be quite similar to that of the hypersthene chondrites befbre shock. Individual small pieces of Knyahinya were cut from the sample and heated to maximum temperatures varying from 900 to 1250°C and cooled at rates varying from lOO”C/week (2 x 10-40C/sec) to a quench (- 5oo”C/sec). The rapidly cooled pieces were annealed in a vertical molybdenum fur-
nace and were protected from oxidation by flowing argon through the furnace tube. The more slowly cooled samples were vacuum encapsulated in quartz tubes and heated in a horizontal resistance furnace. Cooling rates were controlled by a Trendtrak automatic programmer.
RESULTS AND DISCUSSION Microstructural and ,chemical criteria The following microstructural and chemical features were studied for each of the reheated chondrites:
GOLDSTEIN
martensite, melted metal-troilite microstructure, phosphorus enrichment of metal, metal grain compositions, Ni gradients at the rims of metal grains, Ni content of troilite, the presence or absence of phosphides and secondary kamacite (a), and the compositions of phosphides and secondary kamacite. Table 1 gives a summary of the major microstructural and microprobe results for the eight reheated chondrites. There are two basic groups of criteria: those that reveal information about the maximum reheating temperature caused by the shock process and those that reveal information about low temperature conditions from which cooling rates can be estimated. Martensite, melted metal-troilite microstructure, averaging of central metil grain compositions and P enrichment of the metal indicate extensive reheating and in most cases remelting of the shocked chondrites. The presence of phosphides and secondary kamacite indicates relatively slow cooling rates. Composition and width measurements of these two phases can be used to estimate cooling rates. Ni gradients at the rims of metal grains and the Ni content of troilite indicate substantial reheating, but, more critically, reveal qualitative information on cooling rates. Reheating temperatures Martensite. Martensite or a2 was observed in all eight reheated chondrites and in most of the heat treated samples. Figure 1 shows the a2 structure in Farmington. Various morphologies of martensite were observed among the samples and were apparently due to differences in metal composition. Since a2 occurs upon relatively fast cooling of y-Fe from a temperature 2800°C (TAYLOR and HEYMANN,1971), the presence of a2 in the chondrites indicates that these meteorites experienced temperatures 2 800°C. The presence of martensite apparently reveals no information about the maximum temperature of reheat-
Table 1. Summary of key results from reheated chondrites
u Ramsdorf
OWl”i0
Yes
Rounded met-tr boundaries, blebs
0.13
0.08
Yea
Yes
NUN” 5749
Yes
Extensive; cN0‘d grains, blebs
0.45
0.11
*a
Yes
Yes
Extensive; ovoid grains, blebs
0.49
0.08
No
*es
Yes
burectic met-tr, blebs Rounded met_-tr boundrles, ~o,ne ovoid grains
0.06
0.02
Yea
0.17
0.06
Yes
Yes
>0.39
0.08
Yes
Yes
Yes
No
H6
L5
Tadjera
NEW
BN 1959,1023 BM 1959,887
Yes, Sore Decomposed
(H)
B” 1959,768
Decompo*ed
IS
mm” 2315 B” 1959.827
I.5
Extensive, rounded met-t= bcundaries
Yea
Rounded met-Lr boundaries, (lome maids blebs, veinlets
0.02
< 0.02
Yell
sons ovoid gcsins, sore rounded rat-tr boundaries, blebs, veinlets
0.04
0.31
Classification according to VAN Sc~rms and Wool (1967). $ BM: British Museum, NMNH: National Museum of Natural History.
*
No
No
No
Metallic microstructures
ing or about the cooling rate except that the shocked chondrites have not cooled at an extremely slow rate such as that experienced by the iron meteorites. Melted metal-troilite microstructure. The appearance of melted metal-troilite varied among the reheated chondrites and the heat treated samples. Ovoid metal grains agglomerated within troilite indicate extensive remelting. These structures are common in Ramsdorf (BEGEMANN and WLOTZKA, 1969), Orvinio (TAYLORand HEYMANN,1971), Rose City, and several of the heat treated samples (see Fig. 2), Metal in Lubbock, Farmington, Arapahoe and Tadjera typically occurs in irregularly shaped grains with rounded metal-troilite boundaries (see Fig. 1). Metaltroilite interfaces in Wickenburg are intermixed in a eutectic structure previously observed by TAYLORand HEYMANN(1971), Fig. 3. All of the shocked chondrites contain tiny blebs of metal-troilite dispersed in the silicate. Veinlets are plentiful in Tadjera, Arapahoe, and the moderately remelted samples of Knyahinya (Fig. 4). It appears that veinlets occur most plentifully in samples which were not extensively remelted and/or cooled fairly quickly. Samples of Knyahinya heated to a maximum temperature of 900°C showed very few signs of remelting. Therefore 900°C is probably a lower temperature limit for chondritic metal-troilite melting. On the basis of the observed metal troilite microstructure, Ramsdorf was the most extensively remelted chondrite followed by Orvinio, Rose City, Lubbock, Tadjera, Arapahoe, Farmington and Wickenburg. Phosphorus enrichment. Farmington, Lubbock, Orvinio, Ramsdorf and Rose City contain substantial P (> 0.10 wt%) in the metal phase (Table 1). Wickenburg, Tadjera and Arapahoe have low P contents (co.10 wt%) and compare with unreheated chond&es in this respect. Some of the artificially heated samples contain substantial P in the metal phase (0.154.83 wt%). Since the Knyahinya starting material contained co.01 wt% P in the metal phase, these experiments showed that it is possible to enrich the metal phase in P upon reheating. The reduction of phosphate is dependent upon the oxygen fugacity,fO,, which must be low enough for a reduction reaction to occur (OLSEN and FUCHS, 1967; FRIELand GOLDSTEIN,1976). OLSENand FUCHS (1967) proposed the following reduction reaction 3CaMg(SiO,),
+ 402 + 3Fe + 2(P)F 3 Ca3(PO& + 3MgFe(SiO,),
(1)
where (P,” is the P content in the metal phase. This reaction can explain the relation of P in solution in the metal phase to metal-phosphate equilibrium in iron meteorites. TAYL,ORand HEYMANN(1971) proposed a similar reaction for the reheated chondrites. Data from WILLIAMS(1971) indicate, however, that the f0, of ordinary equilibrated chondrites is too high for reduction via the above reaction to occur. Presumably during the shock process the local f0, in the chondrites is low enough to make reduction
and thermal histories
1063
favorable. This low f0, may be caused by the presence of a reducing agent, such as C or CO (TAYLOR and HEYMANN,1971). The P reduction in the experimental studies using the chondrite Knyahinya, may have been controlled by the presence of tantalum inside the sample capsules. The tantalum was used as an oxygen getter to protect the metal. The Ta-Taz05 equilibrium reaction probably buffered the f0, to well below the value needed for P reduction. Once the f’0, is favorable for P reduction, the amount of P which dissolves in the metal phase is probably most dependent on the volume ratio of phosphate to metal and the residual temperature (TR) or maximum temperature of the chondrite after the shock event. The kinetics of P diffusion and the intimate contact of phosphate and metal would be increased during a partial melting stage. In general, the presence of increased P in the metal of reheated chondrites is consistent with other evidence of severe reheating. On the basis of the observed amount of P in the metal grains, Rose City, Orvinio and Ramsdorf have the highest maximum reheating temperatures followed by Lubbock, Farmington, Wickenburg, Tadjera and Arapahoe. Metal grain center compositions. Previous investigators have noted that the Ni contents of metal grains in reheated chondrites are often similar from grain to grain (HEYMANN,1967; TAYLORand HEYMANN, 1971; Begemann and WLOTZKA,1969). The results of our measurements are illustrated in Fig. 5, where histograms of the number of grains vs central Ni content for an unreheated chondrite, Wellman, and for Orvinio, Farmington and Tadjera are given. Reheating-shock remelting apparently has the effect of averaging out the original compositional differences between grains found in unreheated chondrites (Fig. 5a). BEGEMANNand WLQTZKA (1969) proposed that such averaging out was due to melting, mixing, and redistribution of metal during shock reheatingremelting. This proposal is supported by the results of this study. The degree of compositional similarity of grain centers varies among the reheated chondrites studied (compare Fig. 5b-d) and coincides with the degree of remelting. Chondrites with extensively remelted metal-troilite (Ramsdorf, Orvinio, Rose City) have a narrow range of central Ni contents while the least remelted-reheated samples (Tadjera, Arapahoe, Farmington, Wickenburg) generally show a wide scatter. The veinlets observed in the least reheated chondrites (Fig. 4) indicate that molten metal and sulfide seeped through cracks in the silicate matrix and helped to some degree to mix the metal. For the more extensively remelted chondrites, molten metal and sulfide were able to mix thoroughly allowing a homogenization of the molten metal. Some time must have been available for this mixing to occur. Such a process however must proceed without allowing the metal-sulfide melt to separate from the surrounding silicate. Only minor amounts of solid state
B. A. SMITHand J. 1. GOLIISTEIN
1064
By comparing and integrating results from the above criteria for the eight reheated chondrites, a relative ranking according to maximum reheating temperature can be made. Ramsdorf has the highest reheating temperature followed by Orvinio, Rose City, Lubbock, Farmington. Tadjera, Arapahoe and Wickenberg.
“5
10
15
20
25
30
35
W t
%
40
45
50
55
Ni
14
Orvinio(lt.1
-f 12 .y (3 z
10 8
L 4
6
;E
4 2
0
5
10
15
20 w t %
14
25
35
40
(CL) Farmington
t In
30
Ni
I
’ zt
0
5
IO
15
20
25
30
35
40
wt % Ni
14
E z’
t
Cd)
4 2 0
5
15
20
25
30
35
40
wt % Ni Fig. 5. (a) Wellman histogram of number of grains vs Ni content at grain centers. 23 grains measured; (b) Orvinio histogram of number of grains vs central Ni contents. 20 grains measured; (c) Farmington histogram of number of grains vs central Ni content. 20 grains measured; (d) Tadjera histogram of number of grains YS central Ni contents. 25 grains measured.
diffusion have contributed to the averaging out of the metal compositions even in the very slowly cooled chondrites.
Ni rim yrudienrs. An increase in Ni at metal grain edges or rims of the reheated chondrites was measured (Table 2). The largest gradients are observed in Ramsdorf. Orvinio, Lubbock and Tadjera. Ni and P profiles are plotted in Fig. 6a for typical grain rims in Ramsdorf and in Fig. 6b for heat treated sample I5 (I 250°C max temp, lOO”C/day cooling rate). The Ramsdorf gradients are similar to those observed by BEGEM~NNand WLOTZKA (1969). TAYL~K and HEYMANN (1971) reported similar gradients in Orvinio and Lubbock. BEGEMANN and WLOTZKA (1969) proposed that such gradients are due to Ni segregation during solidification of an Fe-Ni-S melt and this proposal is supported by the experimental heat treatments. Given enough time or a slow cooling rate at high temperatures. the Ni gradients would be expected to be leveled out by solid state diffusion. Ni rim gradients are therefore indicators of relatively fast cooling. in accord with the steep Ni gradients observed in the artificially heated samples where the maximum temperature was > 1050°C and the cooling rates were faster than lS”C/day (Fig. 6b). Ni in troilite. The Ni contents of troilite in the reheated chondrites vary from ~0.02 to 0.31 wt:,, (Table 1) and are higher than those of unreheated chondrites, which contain <0.02 wtO/,,. Studies by KULLEKUD (1963) indicate that the Ni solubility in troilite decreases from
1065
Fig. 1. Typical Farmington metal grain. c(~ martensite (labeled M) is predominant structure. Troilite (T) and secondary c( (1) arc also present. Dark grey matrix is silicate. Nital etch. Marker = 25 pm. Fig. 2. Typical
Ramsdorf surrounds
metal grain. Most of grain is c(~. Light-etching high Ni rim (open I~, Troilite surrounds the metal. Nital etch. Marker = 25 pm.
arrow)
Fig. 3. Metal and troilite of Wickenburg. The metal and troilite are finely intermixed in the central area of photo, indicating a eutectic structure. (Metal is lightest phase, troilite grey.) Major mass of metal is (Ye. No etch. Marker = 1Opm. Fig. 4. Photo of Arapahoe phase). Marker = 25 pm.
Fig. 9. Metal
and
featuring veinlets (arrows) of metal and/or troilite in silicate matrix Metal predominantly G(~ with light-etching (high Ni) rim. Nital Marker = 25 pm. troilite
in heat
treatment
sample
No. 15 (125O”C, lOO”/day).
(grey etch.
Metallic microstructures and thermal histories
1067
Table 2. Ni rim gradients Average Compos
Rim
Ni
P
ition (wt.%)*
Specimen
MO, Grains
Farmington Remsdorf
11 10
Orvinio
9
Wickenburg Lubbock
6 i3
Rose City
6
Arapahde
5
Tadjera
11
Average Center
Maximm
Composition (wt.%) Ni P
Variation of Ni, center to Rim
14.1 0.12
12.4 0.14
-2-49.
21.4
10.3 0.41
-8-20
0.34
12.5 0.33
8.0
0.44
i-8
15.2 0.030
13.4 0.038
-1-Z
15.6 0.17
11.2 0.17
-2-10
8.9 14.2
0.37 0.510
16.5 0.045
8.6 13.5
0.60 0.013
10.9 0.046
* Counts were taken at edges of either a2 st~ctures rims of a2 grains. Secondary a was avoided.
samples contained more than 0.15 wt% Ni. From these results the cooling rates of the shocked chondrites are slower than 25”C/hr. Secondary kamacite (a). Secondary kamacite (a), was
<1 %.1-Z
-2-20
or ii~t-etching
formed from tbe y phase during cooling from maximum shock temperatures. It was observed in Farmington, Wickenburg, Rose City, Lubbock and Arapahoe (see Fig. 1). This phase was observed previously
HT No.15(l2~C,lO~~y)
Groin Rim D Ni *P
&
.
-hm--
Distance Fig. 6. (a) Ni and P profiles across metal grain rim in Ramsdorf, Vertical dashed line is troilite (Tr)metal interface; (b) Ni and P profiles across metal grain rim in heat treatment sample No. 15 (125O”C, lWC/day).
B. A. Sbirm and J. I.
1068
Table 3. Description of two phase
Bulk Composition Sainple
Si
P
(a/y
and phosp~de/me~l) particles in reheated chondrites
interface
Compositions
Precipitating
Phase
Type
GOLDSTEIN
Xi
of ?.-phase
*tmftum Other
P
Type
R05e City
9
oi
6. LM.ZO
10
0.02
c(
6.7t0.20
tubbcch
1,
0.17
a
7.0t0.10
0.24ta.03
Fumington
12
0.13
a
7‘010.20
0.1710.03
Y
.19
Wickenburg
13
0.06
a
6.6M.20
0.09t0.02
Y
>1s
-
-
!&se
9
Y Y
-------
City
Farmington
0.3o*o.clS -0.02
12
0.13
*
-575’C
Fe-Zi-F
-sso*c
Fe-Xi
*
-sso*c
Y
Fe-Xi-P
0.08
-save
Fe-Xi-P
0.04
-SOO%
Fe-XI-P
0.27
-S?SY
Fe-Xi-P
FeJi-P
neg1. *
-_-0.50
Teq
* .I8
Phase Diagrm Employed
P
Ni
Arqpahoe
-0.50
(wt.%)
Phase
____
PII
22t2
IS.5
a
-6.0
ph
-3s
15.5
(I
-6.0
co.2
_
* Interface values too difficult to obtain. 2S’C/hr show Ni contents below 0.3 wt%. Even at a slow cooling rate of 4”C/hr, the experimental in Farmington by WOOD (1967) and BUS&X et al., (1966) and in Wi~enbmg by TAYLORand HEM&ANN (1971). Secondary karnacite typically occurs along the edges or grain boundaries of a2 with silicate, troilite or other metal grains since these locations provide favorable sites for heterogeneous nucleation (Fig. 1). The presence of secondary a indicates that the cooling rate of the metal was slow enough to allow x to nucleate and grow from y during cooling. The presence of P in the metal facilitates bt nucleation and growth, since P is an ELstabilizer (~~~~ and DOAN, 1972), and increases the diffusion coefficient of Ni in Fe-Ni (HOWARD and GOLDSTEIN,1973). The Ni and P ~s~butions at crly interfaces can be used to approximate the final temperature of equilibration (T,,) of the two phase assemblage a-y. The method employed by AXONand GOLDSTEIN(1972) for dete~ning T,, of lunar two-phase particles was applied in the present work. Interface Ni and P compositions of a were fit to the approp~ate Fe-Ni or Fe-Ni-P phase diagram to estimate 7& Point analyses were made using the microprobe along a line crossing the interface between the two phases. Figure 7 shows such a profile in the Fa~~gton chondrite.
interface compositions and ‘I& estimates are listed in Table 3 for those chondrites which contain a. The Ni composition on the y side of the interface near the a/y interface decreases rapidly (Fig 7) due to low diffusion rates in the y. It is therefore difIicult to obtain an accurate interface value for y. Accordingly T_ estimates are based mainly on the a ~m~sitions. The values of 7& can be used to give a relative ranking of chondrite cooling rates; a sample with a lower I& has a slower cooling rate. Cooling rate estimations were made by calculating the times of secondary kamacite growth. The following assumptions were made for the growth calculation: (1) growth is diffusion controlled; (2) chemical ~uilibrium exists at the o/y interface; (3) growth is one dimensional; (4) no imp~ge~nt occurs, and (5) growth can be considered as having occurred at one tempemture, called the optimum growth temperature. With these assumptions, the following analytical solution can be used to solve Fick’s second law for the Ni con~n~ation in CIand y:
C&Gtf = 4 + 4
Farmingt~ ONi
l
, . ~+***= l
lL******..* i
.
*%I=Dirtonce
Fig 7. Ni and P profiles across secondary doin Far~n~on.
erf
(2)
1069
Metallic microstructures and thermal histories
ture at which these effects balance is called the optimum growth temperature. At this temperature the s( phase growth will be a maximum. The optimum temperature for each case was found by performing isothermal calculations over a range of temperatures between the theoretical nucleation temperature and Ter Once the optimum growth temperature was found for each meteorite, calculations were performed to find the time needed to grow c( to a width of the order of that observed in the sample. Due to geometry considerations accurate u. phase widths could not be measured and a best value was estimated. The tem~raturc range between the nucieation temperature and T, was divided by the calculated time needed for growth at the optimum temperature. This result yields an approximate cooling rate for each meteorite. These cooling rates vary from - l”C/yr to l’C/lOO yr. Table 4 lists such cooling rates along with the temperatures and times employed in the calculations. No secondary M was observed in the heat treated samples. This is not surprising as the experimental cooling rates were much faster (>50WC/yr) than those calculated for the shocked chondrites. P~osp~ides. Phosphides were observed in Ramsdo& Orvinio, Rose City, Far~ngton, and Lubbock (all the chondrites with P > 0.1 wty< in the metal). This phase was previously observed in Ramsdorf (BEGEMANN and WLOTZKA,1969) and Orvinio (TAYLORand HEYMANN,1971). The phosphides apparently nucleate mostly at the a-troilite or former y-troilite boundaries during cooling. Rose City and Farmington contain phosphides large enough (72 pm) for compositional measurements across phosphide/metal interfaces. Ni and P profiles across phosphides in these two meteorites are shown in Fig. 8. By applying the method of AXONand Got~srrn*; (1972), 7&‘s for Rose City and Farmington phosphide/me~l were estimated as -575 and <55O”C respectively (Table 3). The phosphides in Lubbock are generally - 2 pm wide and the phosphides in Ramsdorf and Orvinio are <2 pm wide. Rose City, Lubbock and Farmington have the largest phosphides, contain secondary kamacite, and therefore have slow cooling rates,
where A, =
a’ - d
c’ - a’ erf y
1 - erfy ’
B, = ~
1 - erfy’
D is the diffusion coefficient, x the distance, t the growth time, C, the composition in the taenite parent phase, c” the equilibrium Ni content of the precipitate, c’ the equilibrium Ni content of the parent phase, a’ the bulk Ni content of parent phase, and y a dimensionless parameter (JOST,1952). The location of the interface between the precipitate and parent phases, 5 is given by:
5 = 2&/V%
(4)
(JOST, 1952). The y term is found by inserting the general solutions for C, and C,, equations (2) and (3) into the mass balance expression
where
is the concentration 5- and
gradient in j at the &‘a interface, aC,
ax {+ (--_) is the concentration gradient in c( at the ~JOZinterface 4’. After minor rearranging of terms one has: $&eY’)(l
- erfy) = g,
which is solved for y iteratively using a standard Newton-Raphson technique (MCCRACKEN,1965). The interface distance or precipitate width < can then be calculated for desired times using equation (4). The diffusion coefficients used in this calculation were obtained from HEYWARDand GOLDSTEIN (1973). There are two variables that control precipitate growth, difTusion coefficient and driving force, which have opposing effects with respect to temperature. The diffusion coefficient decreases with decreasing temperature and the driving force increases with decreasing temperature (REED-HILL,1964). The tempera-
Table 4. Estimates of cooling rates for reheated chondrites, based on the growth of secondary kamacite T
sampie
Growth
I
ac
nucl.
time Width4
Tmx
Cooling
700-c
575’C
675’C
100
yrs.
10 jmI
.kapsb?
700*
550°
685’
1000
yrs.
8 9”
>lY/lO
Lubbock
650°
5500
600
1000
yrs.
7 w
-1’Cl
Farmingron
6509
500*
boo0
I.000
yrs.
8 P
-1w10
Wickenburg
6759
5000
625’
Rose
city
10,000
yrs.
r Approximate nucleation temperature, no undercooling. * Approximate final equilibrium temperature, Table 3. 3 Calculated optimum growth temperature. 4 Approximate secondary kamacite width.
10$m
Rate
-1wyr. yrs. to
yrs. yrs.
B. A. SMITH and J. 1. GOLDSTEIN
upon resolidification. Also the presence of phosphides in a heat treatment sample 1050°C. lOO”C/day indicates that phosphides can occur in metal which cooled as fast as IOO”C/day.
THERMAL HISTORIES OF REHEATED CHONDRITES
._
Farmington 0 NI .P
40. Tr 35
0
..P
I I Ph I I I no I
1
o(
.
1
.20
@z
I
I. ..I_._.
I,
*lOUrnDistance Fig. X.(a) Ni and P profiles across phosphide (Ph) in Rose City: (b) Ni and P profiles across phosphide (Ph) in Farmington.
0.1~-l”C/yr (Table 4). The presence of phosphides in Ramsdorf and Orvinio, although they are smaller, might indicate slow cooling for these meteorites. However these two meteorites have been extensively remelted, they have the highest reheating temperature 2 1200°C and the highest P content in the metal. It is possible that the phosphides formed during the remelting stage and a small amount of phosphidee metal eutectic nucleated at the edges of the troilite
Ramsdorf has probably experienced the highest reheating temperature of the chondrites studied. The metal troilite assemblages indicate extensive remelting. The central Ni contents of the metal grains are similar to each other and the high P content of the metal also indicate a high shock-reheating temperature. A comparison of the metal structure of Ramsdorf with the experimental heat treatment samples (Fig. 9) indicates a reheating temperature of - 120th I25O”C. Troilite in Ramsdorf contains an average Ni content of 0.1 I wt”,,. which indicates a relatively fast cooling rate. The presence of steep Ni rim gradients and the absence of secondary 3: also indicate a relatively fast cooling rate. A comparison of Ramsdorf with the structure of the heat treated samples (Fig. 9) and the results from cooling rate criteria study lead to an estimate of - IO0 C/day as the cooling rate for Ramsdorf. BEGMANN and WLOTZI(A (1969) propose a cooling rate of - 3 C/min at a reheating temperature 01 > 1200°C. Such a value was based on a calculation of solid state diffusion in the grain rims of Ramsdorf at tcmperaturcs above 1000 C. Evidence from our heat treatment studies shows that metal-troilite melts at a temperature of IOOO‘C and the rims of the metal grains in Ramsdorf did not solidify until they cooled below 1000°C. This result invalidates the solid state diffusion method for T > looO”C. Calculations of BEGEMANN and WLOTZKA (1969) indicate however that a cooling rate of - 100 C/day is possible if solidification occurs below 1000 ‘C. Orvinio has a metal-troilite structure which appears to have been extensively remelted. Results from the study of high temperature criteria in Orvinio and comparison to the heat treated sample lead to a maximum reheating temperature of - 1200°C. The cooling rate of Orvinio must have been fast enough to prevent secondary x growth and also to retain some Ni rim radients. The Ni content of troilite (0.08 wtf, av) is somewhat lower than that of Ramsdorf indicating a slower cooling rate. Orvinio’s cooling rate is estimated as - IOO’C!day to - I S”C/day. Tadjera contains only some metalLtroilite that appears remelted. The metal in this sample contains a relatively low amount of P, and the central Ni content of the metal grains varies greatly (Figure 4). Tadjera contains a high density of veinlets, which again indicates a less severe remelting process. The maximum reheating temperature is estimated as - looO-1050°C. The relatively high Ni content of troilite (0.31 wt”; av), the inhomogeneity of individual metal grains (including some Ni rim gradients)% and
Metallic microstructures
to-3 IO4 IO5 10-6 Icr7
-* 130d0
1071
and thermal hislories
10-E
lo+
10’0
Ra 1200 -
-tt,
,,Rc
Or
TR 1100 YC) 1000 - +Pa
900
1
I
I
1
I
IO
100
1000
Approximate
Depth of Burial
(ml
Fig. IO. Thermal history summary for eight severely reheated chondrites (samples identified by their initials).
the absence of secondary ct indicate a relatively fast cooling rate for Tadjera, probably on the order of LOO-500”C/day. Wickenburg appears to have been the least reheated and slowest cooled of the reheated chondrites studied: The presence of a metal-troilite eutectic (Fig. 3) may well indicate reheating to a temperature just above the Fe-Ni-S eutectic temperature (950-1000°C). In addition the metal has a fairly low P content, 0.55 wt% av. The maximum reheating temperature for Wickenburg is estimated as -9% 1000°C. Fa~ington displays a moderate degree of remelting (evidenced by rounded meal-troilite boundaries), grains with fairly similar central Ni content, and moderate P enrichment (0.13 wt% av). The maximum reheating temperature for Farmington is estimated as _ 1050-1100°C. Rose City has an extensive remelted appearance, a high P content in the metal (0.39 wt% av), and shows little variation of the central Ni content of the metal grains. The maximum reheating temperature is similar to that of Orvinio, + 1200°C. Lubbock has been moderately remelted, its remelt structure and composition falling between Orvinio and Farmington. The ~x~urn reheating temperature is estimated as - 110&l 150°C. The low Ni content of troilite (0.04 wt% av), the presence of secondary CIand the presence of phosphides in the metal indicate a relatively slow cooling rate. Some grains in Lubbock however contain Ni rim gradients. This discrepancy can be at least partially explained by the presence of P; those grains with Ni rim gradients generally contain SO.15 wt% P, while those grains with secondary c( contain 20.20wt% P. Increased P has the effect of increasing the metal phase diffusion coefficients (HEYWARDand GOLDSTEIN,1973). Arapahoe has a large number of veinlets, the P content of metal is very low (0.016 wt”/, av) and the central Ni contents of the metal grains have only a fair degree of similarity. The maximum reheating temperature is estimated as -- IOOO-1050°C.
A sum~ry of the estimated thermal histories of the eight reheated chondrites is presented in Fig. 10. The maximum reheating temperatures and cooling rates of the chondrites are obtained from the preceding discussions of appropriate criteria and the cooling rate data from Table 4. Error bars indicate the degree of uncertainty in the estimated thermal histories. Ramsdorf, Orvinio and Tadjera have relatively fast cooling rates while the remaining chondrites cooled slowly (< 1“C/yr). Previous investigators have related the cooling rates of the reheated chondrites to their depth of burial after the shock event (BEGE~NN and W~TZ~, 1969; TAYLOR and HEY~NN, 1971). Material with a slower cooling rate is buried deeper in the parent body. Correlations between cooling rate curves and burial depths have been made by TAYLOR and HEYMANN(1971) and their data is applied to the cooling rate estimates of the present study. The bottom scale in Fig. 10 indicates the depth of burial of the eight reheated chondrites. These chondrites were apparently buried at depths ranging from -0.5 to 1000 m beneath the surface of their parent bodies. CONCLUSIONS (1) The following me~llo~aphic criteria can be used to estimate the post shock residual temperature of severely reheated chondrites: melted metal-troilite appearance, presence of martensite, phosphorus enrichment of metal and averaging of central metal grain compositions. (2) The presence of phosphides and secondary kamacite are due to slow post-shock cooling rates. These exsolved phases can be used to estimate cooling rates. Ni rim gradients indicate both extensive remelting of metal grains and relatively fast cooling. (3) Ramsdorf had the highest post shock residual tem~rature (12~125O’C) followed by Orvinio and Rose City f- 1200°C), Lubbock ( - 1LO@-1 150°C), Farmington (1050-1 lOO*C), Tadjera and Arapahoe (lC@&1050°C), and Wickenburg (950-1000°C).
B. A. SMITH and J. 1. GOLIISTW
1072
(4) Estimates of cooling rates are: Tadjera - 100500’ C/day, Ramsdorf x IOOLC/day. Orvinio . 100 C/day to - 15’ C/day, Rose City - l’C/yr. Lubbock. Farmington and Arapahoe - l”Cjl0 yr and Wickenburg - I ‘C/100 yr. These chondrites were buried at depths ranging from 0.5 to IOOOm beneath the surface of their parent bodies.
Ack,low/edgemerlts The H~,TWISON of the British
authors
wish to thank Dr. R. Museum. Dr. F. WLOTZKA. Max Planck Institut fiir Chemie, and Dr. R. CLARKE. JR.. of the U.S. National Museum, for their help in selecting meteorite samples for this study. We also wish to thank Dr. S. 0. AC;RLLL for assistance with the optical microscopy, Mr. D. BUSH for assistance with the electron probe work. and Dr. F. BEGEMANN for the careful review of the manuscript. This work was supported by NASA Grant NGR 39-007-043.
REFERENCES AXON H. J. and GOLDSTEIN J. I. (1972) Temperature-time
relationships
from
Earth
Sci. Lett.
Planet.
lunar
two
phase
metallic
particles.
16. 439441.
BEGEMANNF. and WLOTZKAF. (1969) Shock induced thermal metamorphism and mechanical deformations in the Ramsdorf chondrite. Geochim. Cosmochim. Actu 33. 1351.-1370. BLSECK P. R., MASON B. and WIIK H. B. (1966) The Farmington meteorite. Mineralogy and chemistry. Geochim. Cosmochirn. Actu 30. 1 8. FRIEL J. J. and G~LIXTEIN J. I. (1976) An experimental study of phosphate reduction and phosphorus-bearing lunar meial particles. Proc. 7th Lunar Sci. Co@, Geochirn. Cosmochim. Acta Suppl. 7. pp. 791X306. Pergamon Press. GOLDSTEIN J. 1. and AXON H. J. (1973) Composition, structure and thermal history of metallic particles from 3 Apollo I6 soils. 65701,685Ol and 63501. Proc. 4th Lunur
Cor$, Geochim. Cosmochim. .4ctu Suppl. 4, pp. 751 775. Pergamon Press. GOLIISTEIN J. I., AXON H. J. and YE& C. F. (1972) Metallic particles in the Apollo I4 lunar soil. Proc. 3rd Lunar Sci. Cor$. Geochim Cosmochim. kta Suppl. 3, pp. 1037-~1064. Pergamon Press. GOLDSTEIN J. 1. and DOAN A. S.. (1972) The effect of phosphorus on the formation of the Widmanstltten pattern in iron meteorites. Geochim. Cosmochim. Acta 36, 51-69. HEYMANN D. (1967) On the origin of hypersthene chondritcs: ages and shock effects of black chondrites. Icarus 6. 189-221. H~~WARI) T. R. and GOLIXTEIN J. I. (1973) Ternary diffusion in the z and ;‘ phases of the Fe-Ni- P System. Mrttrllurgical Trunsu( tion.\ 4. 2335 2342. JOST W. (1952)DifSusior~. Academic Press. KULLERI:~ G. (I 963) The Fe--Ni -S system. Cwnegie Institute Yearbook 62, I75 I X9. MCCRACKEN D. D. (1965) ,4 Guide to FORTRAN IV’ Programming. Wiley. MISRA K. C. and TA’~I.OK L. A. (1975) Characteristics of metal particles in Apollo I6 rocks. Proc. 6th Lunur Sci. Conf. Grochim. Cosmochim. .4ctu Suppl. 6. pp. 615-639. Pergamon Press. OLSF-N I!. and FC’CHS L. H. (1967) The state of oxidation of some iron meteorites. Icarus 6. 242-253. REEI)-HILL R. E. (I 964) Ph!micu/ Metullurg~~ Principles. Van Nostrand-Reinhold. SWALSN R. A. (I 972) Thermod~numics of’ Solids, Wiley. TAYLOR G. J. and HI:~MANN D. (1969) Shock. reheating and the gas retention ages of chondritcs. Earth Planet. %I. Lrtt. 7. 151 ~161. TAYLOR G. J. and H~YMANN D. (1971) Postshock thermal histories of reheated chondrites. J. Geophys. Rex 76. 1879 1x93. VAN SWMLIS W. R. and WOOI, J. A. (1967) A chemical petrologic classification for the chondritic meteorites. Sci.
krochirn.
Cosmochiin
Acta 31. 747-765.
WILLIAMS R. J. (1971) Equilibrium temperatures. pressures and oxygen fugacities of the equilibrated chondrites. &ochitn Cosmochim. .Acfcc 35. 407.-4 I I. WOOII J. A. (1967) Chondrites: their metallic mmerals. thermal histories, and parent planets. lcurus 6. I -49.