- Email: [email protected]
87Rb-86Sr age of rocks from the Apollo 15 landing site and significance of internal isochrons
Earth and Planetary Science Letters, 26 (1975) 29-35
F~
LLJ
© Elevier Scientific Publishing Comapny, Amsterdam - Printed in The Netherlands
87Rb-86Sr AGE OF ROCKS FROM THE APOLLO 15 LANDING SITE AND SIGNIFICANCE OF INTERNAL ISOCHRONS JEAN LOUIS BIRCK, SERGE FOURCADE and CLAUDE J. ALLEGRE Laboratoire de GOochimie et Cosmochimie, Institut de Physique du Globe, Universit~ de Paris V1
and D~partement des Sciences de la Terre, Universitd de Paris VII, Paris (France)
Revised version received February 12, 1975
Internal isochrons for two Apollo 15 rocks give an age of (3.34 -+ 0.09) and (3.46 -+ 0.04) × 109 years with an identical 87Sr/86Sr initial ratio of 0.69928. Considering the possibility for the line obtained in a 87Sr/86Sr, 87Rb/86Sr diagram to be a mixing line, the significance of these results are discussed. 1. Introduction a7Rb-87Sr ages of Apollo 15 samples have already been published by several groups of investigators [ 1 4]. However, discrepancies can be noted among these results. As is usual in science, these ages had to be measured independently. The two rocks we measured have been sampled near the site of the LEM in the basaltic flow area. One of these (15555) has already been measured by other groups, the other one (15058) has not been dated yet. The 15555 rock is an olivine basalt. The study of the zoning in the pyroxenes seems to show that it is coming from a partially differentiated liquid [5]. The 15058 rock is a pyroxene-rich dolerite. The olivine can be seen only as fayalite in the mesostatis [5]. With the study of the pyroxene, Papike et al. [6] came to the conclusion that these two rocks belong to the center of two different lava flows.
2. Experimental techniques After rock crushing, the minerals are separated by use of heavy liquids and further purified by handpicking under the microscope. This procedure is done
in a clean room equipped with fdtered air. To avoid any contamination during this process, we systematically measured procedure blanks (Table 1). Using less than 10 cm 3 of heavy liquids for one separation and washing the separate thoroughly with acetone we estimate the contamination during mineral separation at less than 50 pg for Sr and Rb. Then, each mineral concentrate is spiked with 87Rb and 84Sr TABLE 1 Potz:ssium, rubidium, and st*ontium concentration (ng/cm 3) of the reagents Reagent
K
Rb
Sr
Bromoform Methylene iodine Acetone Methanol
0.95 1.17 0.45 0.46
0.0019 0.0028 0.0004 0.0021
0.024 0.031 0.0036 0.0042
HNO3 65% HC1 (6N) H20
0.32 0.0015 0.22 0.0007 0.2 0.00008 0.32 0.00017
0.0021 0.0078 0.0003 0.0025
HF 48%
0.88
0.0014
0.0042
0.020
0.035
Column blank
3O
and dissolved in HF + HNO3. After column separation, Sr and Rb are run on two separate mass spectrometers. Each of them is equipped with a programmable magnetic field and digital output. All the details of our technique were described previously [7]. The only change in analytical technique is the use of a new 99.7% 84 Sr enriched spike. During the course of this work, we carefully monitored the blanks and the mass spectrometry measurements of 87Sr/86Sr. (1) Blanks are repetitively measured. Our total blanks are K = 10 ng, Rb = 0.04 ng, Sr = 0.15 ng. These blanks can be grossly calculated with the K, Rb, and Sr values of the components quoted in Table 2. The sample/blank ratio is always superior to 800 for Sr and superior to 100 for Rb. Therefore, we have not applied any blank correction. (2) Mass spectrometer runs for strontium are always done on a Faraday cup using currents of about 0.5 to 1.10 -11 amps for the 8SSr isotope. During the course of this work, we have respectively run the CIT seawater, and the NBS 987 standards. However, this kind of measurement without 84Sr spiking, can-
TABLE 2 87Sr/86Sr measurements of the standards 1
(1) NBS 98 7 standard (comparison between 84Sr spiked and unspiked run) Spiked:
Unspiked:
0.71014 (12) 0.71025 (15)
(84Sr/86Sr)~ 3
0.71016 (9) 0.71015 (6)
(84Sr/86Sr)~.0.3
0.71014 0.71013 0.71022 0.71017 0.71020
(7) (10) (8) (9) (10)
(2) CIT water standard 2 (Ur/spiked) 0.70905 (15) 0.70912 (9) 1 Each value corresponds to a separate measurement. All measurements are normalized to an 86Sr/aSSr ratio of 0.1194. 2 The number in parentheses after each value of 87Sr/86Sr corresponds to the ±2a/x/N on the last digit. For example, 0.71014 (12) = 0.71014 ± 0.00012.
TABLE 3 Potassium, rubidium and strontium concentrations and isotopic comp o sitio n of lunar rocks 15058,85 and 15555,147.
Samples
Weight (mg)
K (ppm)
87Rb (ppm)
86Sr (ppm)
87Rb/86Sr 1
87Sr/86Sr2,3
26.7 38.3 31.2 12.9 2.7 3.2 1.5
332 545 605 143 2275 367 3360
0.2254 0.1197 0.297 0.1364 1.361 0.366 1.416
9.85 35.9 40.44 1.800 7.715 2.005 1.416
0.02262 0.00320 0.0073 0.0749 0.1743 0.1804 8.223
0.70040 (6) 0.699507 (60) 0.69659 (7) 0.70288 (9) 0.70789 (5) 0.70818 (7) 0.70767 (8)
31.4 9.2 2.9 23.2 2.3 18.9
325 580 483 176 904 312
0.2395 0.132 0.4072 0.1271 0.5543 0.245
9.044 29.85 3.048 2.654 8.413 3.147
0.0262 0.00437 0.1321 0.0473 0.06513 0.0769
0.70046 0.69951 0.70567 0.70148 0.70237 0.70286
15058,85
Total rock Plagioclase Plagioclase Pyroxene Cristobalite Ilmenite Cristobalite 15555,147
Total rock Plagioclase Ilmenite Pyroxene Cristobalite Olivine
1 Errors on the 87Rb/86Sr approximately ratio is 1%. 2 The numbers in parentheses correspond to the standard error +- 2ox/N on the last digit. 3 Values of 87Sr/86Sr are normalized to 86 St/88 Sr = 0.1194.
(15) (8) (20) (10) (10) (15)
31 not lead to perfect comparison with the actual measurements. We, therefore ran the NBS 987 standard with different proportions of 84Sr spike. The results quoted in Table 2 show clearly that the spiking procedure does not introduce any bias in our measurements and is also in very good agreement with the measurements done by Moore et al. [12] and Papanastassiou et al. [4] on the same standards. The analytical results quoted in Table 3 are given with the corresponding errors.
STsJS%
=
#k,~
0.708
1 ~ ,
0.7070.706
llAgdkkT- I I ~ D L ~
0705 0.704
f /
T: 3.46 ..0.04 BY. /= 0.69928
Q703 /
2J-
t-"
3. Results and discussion
2t°~to~eff/
"":--
0.700
~rP/ag/oc/ase/8-//
For both rocks, the experimental results (Table 3) plotted in a 87Sr/S6Sr, S7Rb/S6Sr diagram define a straight line (Figs. 1 and 2). A linear least-square fit to the data points yields the slope and intercept on the 87Sr/86Sr axis [8]. (1) The classical and most conservative way to interpret such a pattern is to admit that the slope is related to the age of the last isotopic rehomogeneization. The line is then an isochron. If we adopt this point of view, the ages for the two rocks are (3.34 -+ 0.09) × 109 years for 15555 and (3.46 + 0.07) X 109 years for 15058, using ~ = 1.39 X 10 -1~ yr -~ as the
5~,
!
0.706
OLIVINE
141 BASALT/
/
0.705
/#men/te
/
/ \T:3.34"_-OO9BY I:Q69927
0.704
/
0.703
/ 0.699
~:;I
I
lJ_~_J 0 " -
01
,~. . . . . . . . . . .
I
87Rb/86Sr 0.2 Fig. 1. 87Sr/86Sr, 87Rb/86Sr diagram for Apollo 15555,147 rock. The inset corresponds to ~, 87Rb/87Sr diagram. ~(87Sr/86Sr) measured - (87Sr/86Sr) best fit line) (
(87Sr]86Sr)best fit line
I is the (87Sr/86Sr) initial ratio.
)
1000
0.699
_
&l
;2
87Rb/~Sr Fig. 2. 87Sr/86Sr, S7Rb/S6Sr diagram for Apollo 15058,85. Symbols are similar as for Fig. 1. 87Rb decay constant. We can remark that these ages are in the range defined for the Apollo 15 samples by 87Rb-87Sr [1-4]. A quite good general agreement is also obtained by comparison with 39A-4°Aages [9] despite that fact that 87Rb-87Srages appear a little older in some cases (for instance Husain gives a 39A-4°Aage of 3.36 × 109 yr to 15058). (2) As pointed out and discussed extensively by All~gre and Dars [10] a straight line in the 87Sr/S6Sr, 87Rb/86Sr diagram can be obtained by mixing of the two end-members. The line is then not an isochron but a mixing line. This possibility should be seriously checked before saying anything about lunar chronology. A priori, two reasons seem to invite such an inspection. From the common knowledge of mineral chemistry, we can presume that ilmenite or olivine or even orthopyroxene are almost free of Rb and Sr. Calcium-rich plagioclase is, of course, rich in Sr but should be poor in Rb. A K-rich interstitial phase (close to cristobalite in density) is clearly rich in both Rb and Sr and is the dominant source of radiogenic 875r. Inspection of these rocks with a Cameca ion analyzer [11] (as reported at the third Lunar Science Conference) indicates clearly that several minerals are intimately mixed with the finegrained (5/1 or smaller) K-rich phase and, thus, that pure separates must be difficult to obtain (Fig. 3). Straight lines ha the Rb-Sr diagram may simply be mixing lines between plagioclase and the K-rich phase.
~ii~
ii
i il !~i
iili! II~
i~i ~
:i ~i!~!~iii~ ii~i~ii~!ili~i!~ii!iii~i~ i!i~i~iiiii~iii!!i~ii:i i!ili ~
Fig. 3. A. Secondary ion emission pictures for Apollo 15555,147 basalt. The sample area consists of a large ilmenite grain enclosing fine-grained K-rich phases, composite pyroxene grains, and small amounts of cristobalite and plagioclase. The area is ap-
48-t., /|
i i¸¸
Fig. 3. B. Secondary ion emission pictures for Apollo 15058,85. The sample area consists of large cristobalite grains spotted with fine-grained K-rich phases plagioclase, clinopyroxene and titano-magnetite. The area is approximately 200 ttm in diameter.
34
H~ghKp~
///
i/O01~m" y
Plog~cla~
i
~
10
,J
~o $r ~ m
Fig. 4. Rb-Sr correlation diagram for rocks 15555 and 15058 (see the text for comments). To check this possibility we plotted the Rb versus Sr content in Fig. 4. Plagioclase plots almost on the Sr axis. But the other minerals plot on a straight line. This straight line can be easily interpreted as a
,,',, /
/ /
Rb
/
\
• \
+
\ \
\
\
\
\ \
/
/
Pho~ x
minerals.
\\ \
i
oJ~,s~'ved
e, \
\
II~ioclases
Sr
High K phase
£ ~oT, e x
87Rb/ 86Sr
Fig. 5. Correspondence between Rb-Sr correlation diagram and 87Sr/86Sr, 87Rb/86Sr diagram.
mixing line between a high-K phase for one endmember and a point on the Sr axis (Fig. 5) for the other. With these results, nothing disproves the possibility that the straight line in the 87Sr/86Sr, 87Rb/86Sr diagram is a mixing line between one end-member with high Rb/Sr ratio and another almost free of Rb. Note that even if the Sr content is variable, several phases free of Rb have the same ordinate in the 87Sr/86Sr, SVRb/86Sr diagram. Before saying anything about the significance of such phenomenon, we have tried to plot the results of other workers in the same way [ I - 4 ] . All of the data we have studied this way are compatible with a three-phase model. One rich in Rb/Sr, one clearly rich in Sr (plagioclase), and one poor in both Rb and Sr but much poorer in Rb than in Sr. Then, the last two phases almost degenerate in a commbn point on the 87Sr/S6Sr, 87Rb/86Sr evolution diagram. Nevertheless, in some cases the K-rich phase has a variable R b - S r ratio and can then, by itself, produce the necessary spread to establish the line as an isochron. It is not yet clear, however, whether this effect was created by plagioclase contamination in the high-K mineral separate or whether it is a real effect. We are not, therefore, claiming that all of the 87Rb-87Sr mineral data should be interpreted as mixing lines but that it is a serious possibility at least for the Apollo 15 samples. It is now necessary to discuss the consequences of these observations. If the line is a mixing line, we should examine the significance of ages. In the strict sense, we are calculating the apparent age of the high-K phase using an externally imposed 87Sr/S6Sr initial ratio. Since the ages obtained agree remarkably with 39A-4°Aages, it seems reasonable to assume that the initial 87Sr/a6Sr ratio of this high-K phase was not too different from the measured one. The only thing that can occur is that slight differences can create the small spread in the calculated ages observed for the Apollo 15 rocks. In conclusion, it may be said that straight lines in the R b - S r evolution diagram may result from mixing of a phase rich in both rubidium and potassium with phases poor in rubidium. Slight contamination (L5%) by the former may even obscure 39A-4°A release patterns. When the lines are interpreted, as is usual, as isochrons, these may only represent the time of K-rich
35 phase "mixing". In the case of the Apollo 15 rocks several hypotheses arise, which remain to be tested: -Segregation of a K-rich phase as the ultimate product of volcanic differentiation and/or the result of impact events subsequent to the crystallization of the rocks. - C o n t a m i n a t i o n of the m a g n a by subcrustal potassic material during ascent to the lunar surface, and/or by furmadic activity subsequent to emplacement. -Because the rocks studied show generally concordant ages and the K-rich phases seem to have had initial 87Sr/86Sr ratios in equilibrium with the other phases, we believe that basalts are emplaced around 3.35 × 10 -9 yr but the small variations around these values like 3 . 2 5 - 3 . 4 8 X 10~9 yr are not. Acknowledgements Michel Semet and Nobu Shimizu have reviewed the manuscript. References 1 G.J. Wasserburgand D.A. Papanastassiou, Age of Apollo 15 mare basalts. Lunar crust and mantle evolution~Earth Planet. Sci. Lett. 13 (1971) 97-104. 2 B.W. Chapell, W. Compston, D.H. Green and N.G. Ware, Chemistry, geochronology and petrogenesis of lunar sample 15555, Science 175 (1972) 415.
3 V.R. Murthy, N.M. Evensen, B.M. John, M.R. Gioscio, J.C. Dragon and R.O. Pepin, Rubidium-strontium and potassium argon age of lunar sample 15555, Science 175 (1972) 419. 4 D.A. Papanastassiou and G.J. Wasserburg, Rb-Sr ages and initial strontium in basalts from Apollo 15, Earth Planet. Sci. Lett. (1973) 17 324-337. 5 G.M. Brown, C.H. Emelens, J.G. Holland, A. Peckett and R. Phillips, Petrology, mineralogy and classification of Apollo 15 mare basalts, in: The Apollo 15 lunar samples, Lunar Sci. Proc. (1972) 40. 6 J.J. Papike, A.E. Bence and M.A. Word, Subsolidus relations of pyroxene from Apollo 15 basalts, Lunar Sci. Proc. (1972) 144. 7 J.L. Birck and C.J. AllSgre, 87Rb-87Srsystematics of Muntsche Tundra mafic pluton, Earth Planet. Sci. Lett. 20 (1973) 266 274. 8 C. Brooks, S.R. Hart and I. Wendt, Realistic use of two error regression treatments as applied to rubidiumstrontium data, Rev. Geophys. Space Phys. 10, 2 (1972) 551-577. 9 L. Husain, 4°A-39Achronology and cosmic-ray exposure age of the Apolto 15 sample, J. Geophys. Res. 79 (1974) 2588. 10 D.J. All~gre and R. Dars, Chronologie Rb-Sr et gravitologie, Geol. Rundsch. 55 (1965) 226-237. 11 J.L. Birck, M. Loubet, G. Mahnes, A. Provost, M. Tatsumoto and C.J. All~gre, Age and origin of Lunar soils (abstract) Third Lunar Sci. Conf., Houston (1973). 12 L.J. Moore, J.R. Moody, I.L. Barnes, J.W. Gramlich, J.J Murphy, P.J. Paulsen and W.R. Shields, Trace determination of rubidium and strontium in silicate glass standard reference materials, Anal. Chem. 45 (1973) 2384.
F~
LLJ
© Elevier Scientific Publishing Comapny, Amsterdam - Printed in The Netherlands
87Rb-86Sr AGE OF ROCKS FROM THE APOLLO 15 LANDING SITE AND SIGNIFICANCE OF INTERNAL ISOCHRONS JEAN LOUIS BIRCK, SERGE FOURCADE and CLAUDE J. ALLEGRE Laboratoire de GOochimie et Cosmochimie, Institut de Physique du Globe, Universit~ de Paris V1
and D~partement des Sciences de la Terre, Universitd de Paris VII, Paris (France)
Revised version received February 12, 1975
Internal isochrons for two Apollo 15 rocks give an age of (3.34 -+ 0.09) and (3.46 -+ 0.04) × 109 years with an identical 87Sr/86Sr initial ratio of 0.69928. Considering the possibility for the line obtained in a 87Sr/86Sr, 87Rb/86Sr diagram to be a mixing line, the significance of these results are discussed. 1. Introduction a7Rb-87Sr ages of Apollo 15 samples have already been published by several groups of investigators [ 1 4]. However, discrepancies can be noted among these results. As is usual in science, these ages had to be measured independently. The two rocks we measured have been sampled near the site of the LEM in the basaltic flow area. One of these (15555) has already been measured by other groups, the other one (15058) has not been dated yet. The 15555 rock is an olivine basalt. The study of the zoning in the pyroxenes seems to show that it is coming from a partially differentiated liquid [5]. The 15058 rock is a pyroxene-rich dolerite. The olivine can be seen only as fayalite in the mesostatis [5]. With the study of the pyroxene, Papike et al. [6] came to the conclusion that these two rocks belong to the center of two different lava flows.
2. Experimental techniques After rock crushing, the minerals are separated by use of heavy liquids and further purified by handpicking under the microscope. This procedure is done
in a clean room equipped with fdtered air. To avoid any contamination during this process, we systematically measured procedure blanks (Table 1). Using less than 10 cm 3 of heavy liquids for one separation and washing the separate thoroughly with acetone we estimate the contamination during mineral separation at less than 50 pg for Sr and Rb. Then, each mineral concentrate is spiked with 87Rb and 84Sr TABLE 1 Potz:ssium, rubidium, and st*ontium concentration (ng/cm 3) of the reagents Reagent
K
Rb
Sr
Bromoform Methylene iodine Acetone Methanol
0.95 1.17 0.45 0.46
0.0019 0.0028 0.0004 0.0021
0.024 0.031 0.0036 0.0042
HNO3 65% HC1 (6N) H20
0.32 0.0015 0.22 0.0007 0.2 0.00008 0.32 0.00017
0.0021 0.0078 0.0003 0.0025
HF 48%
0.88
0.0014
0.0042
0.020
0.035
Column blank
3O
and dissolved in HF + HNO3. After column separation, Sr and Rb are run on two separate mass spectrometers. Each of them is equipped with a programmable magnetic field and digital output. All the details of our technique were described previously [7]. The only change in analytical technique is the use of a new 99.7% 84 Sr enriched spike. During the course of this work, we carefully monitored the blanks and the mass spectrometry measurements of 87Sr/86Sr. (1) Blanks are repetitively measured. Our total blanks are K = 10 ng, Rb = 0.04 ng, Sr = 0.15 ng. These blanks can be grossly calculated with the K, Rb, and Sr values of the components quoted in Table 2. The sample/blank ratio is always superior to 800 for Sr and superior to 100 for Rb. Therefore, we have not applied any blank correction. (2) Mass spectrometer runs for strontium are always done on a Faraday cup using currents of about 0.5 to 1.10 -11 amps for the 8SSr isotope. During the course of this work, we have respectively run the CIT seawater, and the NBS 987 standards. However, this kind of measurement without 84Sr spiking, can-
TABLE 2 87Sr/86Sr measurements of the standards 1
(1) NBS 98 7 standard (comparison between 84Sr spiked and unspiked run) Spiked:
Unspiked:
0.71014 (12) 0.71025 (15)
(84Sr/86Sr)~ 3
0.71016 (9) 0.71015 (6)
(84Sr/86Sr)~.0.3
0.71014 0.71013 0.71022 0.71017 0.71020
(7) (10) (8) (9) (10)
(2) CIT water standard 2 (Ur/spiked) 0.70905 (15) 0.70912 (9) 1 Each value corresponds to a separate measurement. All measurements are normalized to an 86Sr/aSSr ratio of 0.1194. 2 The number in parentheses after each value of 87Sr/86Sr corresponds to the ±2a/x/N on the last digit. For example, 0.71014 (12) = 0.71014 ± 0.00012.
TABLE 3 Potassium, rubidium and strontium concentrations and isotopic comp o sitio n of lunar rocks 15058,85 and 15555,147.
Samples
Weight (mg)
K (ppm)
87Rb (ppm)
86Sr (ppm)
87Rb/86Sr 1
87Sr/86Sr2,3
26.7 38.3 31.2 12.9 2.7 3.2 1.5
332 545 605 143 2275 367 3360
0.2254 0.1197 0.297 0.1364 1.361 0.366 1.416
9.85 35.9 40.44 1.800 7.715 2.005 1.416
0.02262 0.00320 0.0073 0.0749 0.1743 0.1804 8.223
0.70040 (6) 0.699507 (60) 0.69659 (7) 0.70288 (9) 0.70789 (5) 0.70818 (7) 0.70767 (8)
31.4 9.2 2.9 23.2 2.3 18.9
325 580 483 176 904 312
0.2395 0.132 0.4072 0.1271 0.5543 0.245
9.044 29.85 3.048 2.654 8.413 3.147
0.0262 0.00437 0.1321 0.0473 0.06513 0.0769
0.70046 0.69951 0.70567 0.70148 0.70237 0.70286
15058,85
Total rock Plagioclase Plagioclase Pyroxene Cristobalite Ilmenite Cristobalite 15555,147
Total rock Plagioclase Ilmenite Pyroxene Cristobalite Olivine
1 Errors on the 87Rb/86Sr approximately ratio is 1%. 2 The numbers in parentheses correspond to the standard error +- 2ox/N on the last digit. 3 Values of 87Sr/86Sr are normalized to 86 St/88 Sr = 0.1194.
(15) (8) (20) (10) (10) (15)
31 not lead to perfect comparison with the actual measurements. We, therefore ran the NBS 987 standard with different proportions of 84Sr spike. The results quoted in Table 2 show clearly that the spiking procedure does not introduce any bias in our measurements and is also in very good agreement with the measurements done by Moore et al. [12] and Papanastassiou et al. [4] on the same standards. The analytical results quoted in Table 3 are given with the corresponding errors.
STsJS%
=
#k,~
0.708
1 ~ ,
0.7070.706
llAgdkkT- I I ~ D L ~
0705 0.704
f /
T: 3.46 ..0.04 BY. /= 0.69928
Q703 /
2J-
t-"
3. Results and discussion
2t°~to~eff/
"":--
0.700
~rP/ag/oc/ase/8-//
For both rocks, the experimental results (Table 3) plotted in a 87Sr/S6Sr, S7Rb/S6Sr diagram define a straight line (Figs. 1 and 2). A linear least-square fit to the data points yields the slope and intercept on the 87Sr/86Sr axis [8]. (1) The classical and most conservative way to interpret such a pattern is to admit that the slope is related to the age of the last isotopic rehomogeneization. The line is then an isochron. If we adopt this point of view, the ages for the two rocks are (3.34 -+ 0.09) × 109 years for 15555 and (3.46 + 0.07) X 109 years for 15058, using ~ = 1.39 X 10 -1~ yr -~ as the
5~,
!
0.706
OLIVINE
141 BASALT/
/
0.705
/#men/te
/
/ \T:3.34"_-OO9BY I:Q69927
0.704
/
0.703
/ 0.699
~:;I
I
lJ_~_J 0 " -
01
,~. . . . . . . . . . .
I
87Rb/86Sr 0.2 Fig. 1. 87Sr/86Sr, 87Rb/86Sr diagram for Apollo 15555,147 rock. The inset corresponds to ~, 87Rb/87Sr diagram. ~(87Sr/86Sr) measured - (87Sr/86Sr) best fit line) (
(87Sr]86Sr)best fit line
I is the (87Sr/86Sr) initial ratio.
)
1000
0.699
_
&l
;2
87Rb/~Sr Fig. 2. 87Sr/86Sr, S7Rb/S6Sr diagram for Apollo 15058,85. Symbols are similar as for Fig. 1. 87Rb decay constant. We can remark that these ages are in the range defined for the Apollo 15 samples by 87Rb-87Sr [1-4]. A quite good general agreement is also obtained by comparison with 39A-4°Aages [9] despite that fact that 87Rb-87Srages appear a little older in some cases (for instance Husain gives a 39A-4°Aage of 3.36 × 109 yr to 15058). (2) As pointed out and discussed extensively by All~gre and Dars [10] a straight line in the 87Sr/S6Sr, 87Rb/86Sr diagram can be obtained by mixing of the two end-members. The line is then not an isochron but a mixing line. This possibility should be seriously checked before saying anything about lunar chronology. A priori, two reasons seem to invite such an inspection. From the common knowledge of mineral chemistry, we can presume that ilmenite or olivine or even orthopyroxene are almost free of Rb and Sr. Calcium-rich plagioclase is, of course, rich in Sr but should be poor in Rb. A K-rich interstitial phase (close to cristobalite in density) is clearly rich in both Rb and Sr and is the dominant source of radiogenic 875r. Inspection of these rocks with a Cameca ion analyzer [11] (as reported at the third Lunar Science Conference) indicates clearly that several minerals are intimately mixed with the finegrained (5/1 or smaller) K-rich phase and, thus, that pure separates must be difficult to obtain (Fig. 3). Straight lines ha the Rb-Sr diagram may simply be mixing lines between plagioclase and the K-rich phase.
~ii~
ii
i il !~i
iili! II~
i~i ~
:i ~i!~!~iii~ ii~i~ii~!ili~i!~ii!iii~i~ i!i~i~iiiii~iii!!i~ii:i i!ili ~
Fig. 3. A. Secondary ion emission pictures for Apollo 15555,147 basalt. The sample area consists of a large ilmenite grain enclosing fine-grained K-rich phases, composite pyroxene grains, and small amounts of cristobalite and plagioclase. The area is ap-
48-t., /|
i i¸¸
Fig. 3. B. Secondary ion emission pictures for Apollo 15058,85. The sample area consists of large cristobalite grains spotted with fine-grained K-rich phases plagioclase, clinopyroxene and titano-magnetite. The area is approximately 200 ttm in diameter.
34
H~ghKp~
///
i/O01~m" y
Plog~cla~
i
~
10
,J
~o $r ~ m
Fig. 4. Rb-Sr correlation diagram for rocks 15555 and 15058 (see the text for comments). To check this possibility we plotted the Rb versus Sr content in Fig. 4. Plagioclase plots almost on the Sr axis. But the other minerals plot on a straight line. This straight line can be easily interpreted as a
,,',, /
/ /
Rb
/
\
• \
+
\ \
\
\
\
\ \
/
/
Pho~ x
minerals.
\\ \
i
oJ~,s~'ved
e, \
\
II~ioclases
Sr
High K phase
£ ~oT, e x
87Rb/ 86Sr
Fig. 5. Correspondence between Rb-Sr correlation diagram and 87Sr/86Sr, 87Rb/86Sr diagram.
mixing line between a high-K phase for one endmember and a point on the Sr axis (Fig. 5) for the other. With these results, nothing disproves the possibility that the straight line in the 87Sr/86Sr, 87Rb/86Sr diagram is a mixing line between one end-member with high Rb/Sr ratio and another almost free of Rb. Note that even if the Sr content is variable, several phases free of Rb have the same ordinate in the 87Sr/86Sr, SVRb/86Sr diagram. Before saying anything about the significance of such phenomenon, we have tried to plot the results of other workers in the same way [ I - 4 ] . All of the data we have studied this way are compatible with a three-phase model. One rich in Rb/Sr, one clearly rich in Sr (plagioclase), and one poor in both Rb and Sr but much poorer in Rb than in Sr. Then, the last two phases almost degenerate in a commbn point on the 87Sr/S6Sr, 87Rb/86Sr evolution diagram. Nevertheless, in some cases the K-rich phase has a variable R b - S r ratio and can then, by itself, produce the necessary spread to establish the line as an isochron. It is not yet clear, however, whether this effect was created by plagioclase contamination in the high-K mineral separate or whether it is a real effect. We are not, therefore, claiming that all of the 87Rb-87Sr mineral data should be interpreted as mixing lines but that it is a serious possibility at least for the Apollo 15 samples. It is now necessary to discuss the consequences of these observations. If the line is a mixing line, we should examine the significance of ages. In the strict sense, we are calculating the apparent age of the high-K phase using an externally imposed 87Sr/S6Sr initial ratio. Since the ages obtained agree remarkably with 39A-4°Aages, it seems reasonable to assume that the initial 87Sr/a6Sr ratio of this high-K phase was not too different from the measured one. The only thing that can occur is that slight differences can create the small spread in the calculated ages observed for the Apollo 15 rocks. In conclusion, it may be said that straight lines in the R b - S r evolution diagram may result from mixing of a phase rich in both rubidium and potassium with phases poor in rubidium. Slight contamination (L5%) by the former may even obscure 39A-4°A release patterns. When the lines are interpreted, as is usual, as isochrons, these may only represent the time of K-rich
35 phase "mixing". In the case of the Apollo 15 rocks several hypotheses arise, which remain to be tested: -Segregation of a K-rich phase as the ultimate product of volcanic differentiation and/or the result of impact events subsequent to the crystallization of the rocks. - C o n t a m i n a t i o n of the m a g n a by subcrustal potassic material during ascent to the lunar surface, and/or by furmadic activity subsequent to emplacement. -Because the rocks studied show generally concordant ages and the K-rich phases seem to have had initial 87Sr/86Sr ratios in equilibrium with the other phases, we believe that basalts are emplaced around 3.35 × 10 -9 yr but the small variations around these values like 3 . 2 5 - 3 . 4 8 X 10~9 yr are not. Acknowledgements Michel Semet and Nobu Shimizu have reviewed the manuscript. References 1 G.J. Wasserburgand D.A. Papanastassiou, Age of Apollo 15 mare basalts. Lunar crust and mantle evolution~Earth Planet. Sci. Lett. 13 (1971) 97-104. 2 B.W. Chapell, W. Compston, D.H. Green and N.G. Ware, Chemistry, geochronology and petrogenesis of lunar sample 15555, Science 175 (1972) 415.
3 V.R. Murthy, N.M. Evensen, B.M. John, M.R. Gioscio, J.C. Dragon and R.O. Pepin, Rubidium-strontium and potassium argon age of lunar sample 15555, Science 175 (1972) 419. 4 D.A. Papanastassiou and G.J. Wasserburg, Rb-Sr ages and initial strontium in basalts from Apollo 15, Earth Planet. Sci. Lett. (1973) 17 324-337. 5 G.M. Brown, C.H. Emelens, J.G. Holland, A. Peckett and R. Phillips, Petrology, mineralogy and classification of Apollo 15 mare basalts, in: The Apollo 15 lunar samples, Lunar Sci. Proc. (1972) 40. 6 J.J. Papike, A.E. Bence and M.A. Word, Subsolidus relations of pyroxene from Apollo 15 basalts, Lunar Sci. Proc. (1972) 144. 7 J.L. Birck and C.J. AllSgre, 87Rb-87Srsystematics of Muntsche Tundra mafic pluton, Earth Planet. Sci. Lett. 20 (1973) 266 274. 8 C. Brooks, S.R. Hart and I. Wendt, Realistic use of two error regression treatments as applied to rubidiumstrontium data, Rev. Geophys. Space Phys. 10, 2 (1972) 551-577. 9 L. Husain, 4°A-39Achronology and cosmic-ray exposure age of the Apolto 15 sample, J. Geophys. Res. 79 (1974) 2588. 10 D.J. All~gre and R. Dars, Chronologie Rb-Sr et gravitologie, Geol. Rundsch. 55 (1965) 226-237. 11 J.L. Birck, M. Loubet, G. Mahnes, A. Provost, M. Tatsumoto and C.J. All~gre, Age and origin of Lunar soils (abstract) Third Lunar Sci. Conf., Houston (1973). 12 L.J. Moore, J.R. Moody, I.L. Barnes, J.W. Gramlich, J.J Murphy, P.J. Paulsen and W.R. Shields, Trace determination of rubidium and strontium in silicate glass standard reference materials, Anal. Chem. 45 (1973) 2384.