In vacuo crushing experiments and K-feldspar thermochronometry

Earth and Planetary Science Letters, 117 (1993) 169-180 Elsevier Science Publishers B.V., Amsterdam

169

[PT]

crushing experiments and K-feldspar thermochronometry In v a c u o

T. Mark Harrison, Matthew T. Heizler and Oscar M. Lovera Department of Earth and Space Sciences and Institute of Geophysics and Planetary Physics, 3806 Geology Building, University of California, Los Angeles, CA 90024, USA Received August 17, 1992; revision accepted February 28, 1993

ABSTRACT Using in c,acuo crushing methods, both C1- and K-correlated components of trapped 4°At have recently been identified in hypersolvus alkali feldspar and K-feldspar-bearing 'chert' and stromatolite samples. If these components re-emerge during the late stages of thermal degassing, interpretation of 4°Ar/39Ar age spectra could be complicated. We have extended these observations by performing similar measurements on a well-characterized low-temperature K-feldspar, MH-10. A plot of 4 ° A r * / K against C I / K confirms the presence of a chlorine-related excess argon component from sites accessible by crushing MH-10, although it amounts to only 0.078% of the total potassium-derived 39Ar from this sample. No potassiumcorrelated component of excess argon could be clearly identified. Surprisingly, isothermal duplicate heating steps reveal a Cl-correlated component (4°ArE/Cl = 2.7 _+ 0.2 X 10 -4) that appears to be related to the modification of very small ( < 1 /xm) inclusions. This method appears to provide a basis with which to correct for the excess argon that is commonly observed during the initial stages of step heating of K-feldspars. The product of the 4°ArE/CI and the C1/K (determined for each step via the 3SAr/39Ar ratio) yields an "excess age" that is simply subtracted from the measured 4°Ar*/4°K to yield an age corrected for the presence of excess argon. Crushing appears to produce an artifact that seriously affects the step heating age spectrum but does shed light on possible complexities in the internal distributions of 4°Ar and 39At. On balance, we find that any adverse consequences arising from these effects on interpreting the 4°Ar/39Ar results in terms of thermochronomettic models are exceedingly small.

1. Introduction

Coupled with a reputation for yielding ages younger than rock forming events, the recognition of excess radiogenic argon (Ar E) in the early 1960s [1] contributed to a dampening of enthusiasm for the K - A r method in establishing basement chronologies [2]. Interest was rekindled following development of the 4°Ar/39Ar step heating variant which can, in principle, test underlying assumptions in K - A r dating [3]. Recently, two interesting and provocative studies [4,5] have utilized an additional dimension, in v a c u o crushing, for the examination of argon isotopes from Kfeldspar-bearing samples. Their results suggest an approach which has the potential of providing new insights into an old problem: the incorporation of excess argon in minerals. Turner and Wang [4] presented 4°Ar/39Ar re-

sults, obtained by a combination of crushing and step heating from fine-grained metasedimentary rocks, that led them to suggest caution in interpreting K-feldspar age spectra in terms of thermochronometric models [e.g., 6-8]. Their results point to potentially important and problematic aspects regarding the siting of excess argon in these composite materials and the implications for more commonly used samples (e.g., sub-solvus K-feldspars of igneous origin) need to be directly addressed. I n v a c u o crushing of their two samples prior to thermal degassing yielded 1% and 4.8% of the total 39ARK and well-correlated plots of 4°Ar*/39ArK versus 38ArcI/39ArK, suggesting the presence of both Cl-correlated (4°ArE/C1) and K-correlated (4°Are/K) components of radiogenic 4°Ar (4°Ar * ). Although it was possibly hosted by fluid inclusions, Turner and Wang [4] explored several al-

0012-821X/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved

170

T.M.

TABLE 1 Argon isotopic results, experimental conditions and diffusion coefficients ( rush n f l c m l ) ( ° ( )

41) \ M m "w~

38 W( 1/30'WKh

t6,kr]t9ArZxll) a3~:'xrxlO~Smol3qAr~l~d(51

M H lOu K h, t d w a r I • 4 5 6 g o io 1t 12 13 ~4 15 t6 [18 19 211 21 22 2t 25[~ 250 3OO t25 371) 40O 450 450 5(30 51~q 350 550 550 (z)0 600 650 650 65O ~(Xl 700 750 801)

81X) 85O ~) 950 10fK) 1009 1050 1050 1100 1100 11 (X) 1150 1200 1250 13110 1400 1500

(r~h#[lemp(°C}

1214 I o65 1252 I 124 tttI t051 / [49 1044 11182 u52 (, 010 4 9626 8921 815 I go~ i) 767 6 18118 ~(m I) 730] 465 3 "65 8 7164 720 4 441 "94 8 5.~9 I 15"~ 2 6 7 21 485 ~ (~184 3662 ~696 2298 4 4 42 25.07 1')63 21 (11 2017 2 I 19 21 l 0 21 61 2 I 97 2222 22 53 23 61 2-* 71 25 05 25~S 2 " 50 2925 2~ 61 2962 29.70 30 29 3064 ~0 86 ~ I 62 31 57 311 (1~ 291~1 55 0 8 416.5

40Ar~qAr~

0 6d~f) o 7186 ,) 754~, 0 5269 0 7285 (150~9

308 6 114 7 451 7 3 6 t (1 3793 318 2 41~ 5 ~47 1

(I 5204 O 496] (14964 04O38 04620 0 ~96 04]'5 0) 3971 I) 4 2 7 4 D4144 01570 0 M59 0 t697 0 1 ] 97 0 3860 0~794 0 3687 I) 0 0 I)232 0 0078 0 01('~5 0t)133 () @~88 0 0301 011022 00180 O(DI5 0 (K)OI fl 0{)59 0 I)2';12 0 (X)I8 0 (XI02 OOOO8 00 I) (IOO2 0 ~XX)~ 0 (~C(~ 011111311 0~101 D 0~)1 0OOO2 o~ao2 110002 I) 110¢)5 o (~)5 O.O013 I).~21 011030 0 (K,W.2 0 (KI49 00O63 0OO67 o f105~ 0 (I~M7 00f)6di

~242 232 6 270 v 224 2 2310 274 6 18t 4 2443 1604 t~132 31)1 2 468 4 41)5.o 47118 423 0 1483811 257-; 0 1"225) 426-_ 138 7 85 2 t 51 IJ 4 t 57 3 17o 1 586 4250 1258 0 28411 0 230~1 00836 0 3053 0 ~411 0.4394 114210 06468 0 2050 0.1014 0 3854 0 5667 I) 6~,48 115OO0 0~122 i 152 1 275 1 931 1 018 2944 3 764 2 931 2 163 08153 t f)(M 95 68 133] [I

0.0

38Ar(I/39ArKb

36Ar#t ArSx]0~

0.036 11~1O5 OO67 01152 0 (~)9 0 O28 t) 0 2 0 0 017 01)32 0 082 0 IR)5 0 070 0055 0047 0 039 O 118 (} 1~)4 0 0(H 0209 0 020 0 l 17 0022 0 020 1) I I I 0035 0 183 0 }14 I).881 I 24 142 3 22 146 2 71 O085 773 802 12 ~ 961 34 6 55 2 717 304 32.2 70 (I 2O~ 21 8 41 2 885 88 I 66(, 448 t6 5 ~2 I 20 6 334 346 27 6 52 v 185 115 3 17 0 3l)5

6907 12~9 1914 7 7 19 IL~} 5 32 97 3O DI 33 I)2 21 9~ 2992 20 85 2298 20 o v 2246 2153 2241 23 37 23 86 24 65 2622 26 89 26 80 28 33 2q 02 29 12 20 DI 20 5 ] 2846 29 I t 29 45 2964 31 57 31129 29 82 28 N) ~7 27 l q 6Q

00O82 00 0 1934 0 0187 0060~ 00 0 0 (I 0 2 1 8 0 tl~ ~ t OO2O3 o (X~)2 00026 O0 00020 O.O(D6 0 01Kt7 0(} 00 00 0 0 0 (~)2 11111)13 0 OfXR) 0 ~5 0 f~)10 0 (/4)18 O Of)lq 0 (~)32 0 ~X142 O(~M ~ DO(M4 0 01)38 0 f~)8N 0 004S 0 (1~47 0 0041 0 (I

221)6 265 7 2(10 2 21117 ~.221 4262 81 6 6 I I 77 i 1.78 2 823 6 82~ 7,)07 ~ 2gR 4072 2 023 2 ffa6 4 386 4 3O4 4~85 5 534 5 2f~ 2 825 2 ~62 6869 ~ ~1~ I ()~0 I (M,O f) 053I 1 069 4 54(I 6t21 9 473 5 214 4 23O 0 IXI(×) 44 71 44 ~9

194

24 2 38 3 89 8 42 6 ~8 7 5.:. 7 593 623 048 670 69 / 71t 73 ~ 75 t) 792 91 9 99.8 0 9 09 100

3qAr x l 0 5mol 39At ~ l e ~ d ( % )

M H lOu' ( > 1 5 0 tl~) K 4 e l d s ~ a r ~50 41X) 45O 45O 5()q 51~) 51~) 55O 55O (~1~1 61111 65o 650 7t)O 7OO ~5 0 75O 8(FO 8[g) 850 850 o(xI 950 10011 I00~) 1051) 11(~) [ [(~) t 100 1 I(R) I I(1~? 1[(#) 12(#) : 2 5o I ~G~I / 4(1~1 15511

0 (~)2 OOO ~ 0.t~7 (I 0 i / 00[ 2 0 f)14 O I) 15 DO] 6 I) 1118 O024 0028 0 033 OO~7 0.040 0.043 0 I)51 I) 1151 0 051 0fh56 0 067 01175 0 [176 0 O78 0 fig5 0088 0 I11O 0 122 0 182 I) 266 0363 0583 (}68~ 0868 08-4 140 [95 2 811 3 45 5 82 0 5,) 14 5 ]7 2

0019 0 I)38 0 023 01)I8 0 I01 0 1 ~6 O 027 11233 O IV6 0362 0 422 0730 0 962 0908 202 3 72 2 55 254 340 3 4~ 2 55 2 83 5 {X) 238 5 al) 31)0 a 55 ~ 87 272 4 811 5 87 59g 1 4~ - (~ 7 51 1118 0 4 IN

40At* ( q ) c

41l ~,r-/3~ ~,rK

~,ppa~nl age ± I o

Iim¢ ( m in i

IIII10/1 (K)

log l)/r2 ts

(J = 0 0 0 7 9 4 2 W e & h t = 0 0 0 9 6 8 e) a ~1 ~ ,~ v 59 ~ ~14 898 i~)0 80 2 '~) 1 91 I 92 " '11. ~ 92 3

923 tX)(I 93~ ~) 6 71 1 ~ ~ g7 8 7(1 I 8118 81) 5 82 6 0 7~ 4 16 "26 17 2 ~8 2 47 1 74 I 64 I 95 2 96O 168 ')7 g 98 v 991 9O 4 q~ I ,~o2 09 2 t)o ~ 9911 095 094 99 4 ocl 2 99.~ 993 ,m/ 987 085 978 07 4 965 053 959 972 98 5 05,1 16 2 2 82

40Ar'(%)c

11)07 (~71 5 II 19 IOIh 1019 n57 t 1026 ~414 0850 8~ 0 g ~9 4 -96 4 823.9 733 q 7548 6955 190 8 3793 fMI 1 ! 26 9 61 q 1 5~ t 505 5 ? 2 32 ~4 24 41135 27 IO 26 2O 21 36 45 ~2 2~ 72 ~61K) 2240 31 84 24 6 ~ 10 52 21) 02 21) 12 2104 21) 08 2146 21 ,R2 22 O3 22~2 2347 24 57 24 8 6 25 - 6 2~ t2 2002 20 25 20 22 2 o 11 2970 2974 29 7 ~ 10 7 ~ I110/

4 1 0 1 ~ ~3 ~ ' ~ , ± ')6 4 1 U ± ~4 ~';78 ~ ~g ~083 ± 40 ]881 1: I I t 9 9 ~ ± /5 3 8 5 6 ± 36 10301:46 ~756 ± 20 t675 ± t" ~502 ± 20 3646±41 ~46"6 ± 21 ~51~1± 6 3383 ± 13 1285 ± IO0 251),6 ± 295 t26[) ± 21 2gGq ± 21 3208 ± 7 3103 ± 26 ~ 149 ± ~q 412 ± 024 4~4 ± 2 3 6 5O2 1:61 352 ± ~ 1 1141 I ± 7 ; tl)" () 1 : 1 4 5 5 8 8 ± ~tl 41 I 3 ± 2 3 45~ 6 ± 1 2 2965 ± I 3 4 ( ~ 7 ± IO 2 3228 ± 04 2600 1:0 2 2773 ±05 2675 ±03 2 7 8 8 1 :0 2 278.11 ± 0 4 2 8 3 0 ± 116 2~8 3 ± 0 4 290 5 ± 04 205 6 ~ 0 3 3084 ±04 3216 ± I 1 ~25 I ± O4 335 8 ± 0 8 3543 ± 05 3741±1) 5 ~ 7 6 8 ± 0.5 3765 ± 13 175 2 ± D 5 382 1 ± Dt 3826±1).8 3 8 2 4 ± D.5 1 9 4 0 ± 0.7 3060±O5

16 18

1')12 i ')12

. . . . . /. 6=2. . .

12 17 24 14 87 n 14

1 555 1 486 1 ~83 1 383 l 294 1294

Iv 30 /2 iN 14 29 70 22 24 21 ~o

1215 1215 I 145 1145 1.083 i t)~i 1 08 g IO28 1028 0 978 f) 0 3 2

1~

08~1 ii 81~1

11t2 H 52

2

,I 0 16 8 8v 8 37 8 58 7 05 76t , 7 11 63 ~ 65t 5.60 55~ 5 66 5 RI 536 4 ~O 4 53 4 ~) 442

i1:.

.I). 818 ....... 11786 0786 0 756 I) 7 5 6 O'28 I) 728 f) " 2 8 O7[)I O 679 1 t6 5 7

4 17 4 17 444 4 33 4 56 4 W 459 484 441 401 ~40

29 .~

t82 9 ± 0 7

10 16 It) 16 10 20 40 13 1I 16

2851 26 78 2248

t682z03 t4" 9 ± 2 2O6 ± 55

12

0 6~6

2 71)

14 10

05o8

2 91

411Ar*,#ArK

App~nta$e±lo

l]~(mln)

I(X~J/Y(K~

l o g l ) / r 2 ( s"~)

(I = 0 007942: Weight = 0.0006 ~,~ 0021 O(ff~4 0090 0 I If) 0 222 0 ~52 0 382 06-13 II 8 ~ I2a 1 72 253 3 6t 4 63 689 I1 I 139 16 7 2O 6 244 27 2 304 36. ~9 1 45.4 48 7 53 R 58 i 612 66 5 73 I 7o 8 814 81)~ 98~ r~) 5 100

t 80 tt 5 35 7 ~ 37 48 0 34 6 106 77 3 b96 9115 827 85 5 86 7 G,0 I ')2 9 92 7 85 v 874 89 ~ 882 87 ,.~ 8 80 5 86 a 97 3 92 9 93 1 Rt) 9 86 q 7O 7 . i i1 57 4 ~42 89 2 '~15 172 2 ~o

47 67 45 34 [ 2O 6 1757 8801 2035 14 86 29 5! 18 4~ 2906 18 81 205o 1o 97 2 i 23 20 91 2t 78 22115 2256 2 t 33 24 56 25 tl 25 o4 2 " 61 27 I% 28 12 28 67 2~ [ " 2R / 6 2851) 28 OR 27"4 28 " 5 28"2 28.55 28 87 24 (~t 6 55t

580±920 555±88 1277±33 236~z'~4 056113 270~1 I 201±25 3799±60 2404±88 ~74 7 ± 3 0 251 2 ± 1 8 2"33±16 265 6 ± 1 2 281 1 ± 2 4 2 7 7 1~-0.8 287 O±l I 291 1±17 2973±2 5 Ic4~ 7 ~ 9 ~215±2 7 3t04z1 4 3 ~8.0±1 ~ ~57 7±1 2 ~(~) 7 ± t 6 3 6 1 . 6 ~0 9 ~70 0 ± 2 0 t760-220 ~6~ l ± l 6 ~ 6 8 1±2 7 363 2 ~ ) H ~59 2 ± 2 4 3 7 l I):t:63 t70 7.1 v ~68 6 ~ ) : 3724~) 7 t / 5 21: g 5 01 5 ~ 3 0

22 34 II 16 13 28 14 14 20 10 17 10 19 7 2O i9 26 14 ~2 10 27 18 2Q g 94 14 14 21 2~ 76 224 625 10 16 81 tl

i 6115 1 486 1 183 1 38~ 1204 I 294 t 294 1215 / 215 I 145 / 145 1 083 I I)8t 1028 I 0251 O°78 t) 07~ O 932 0 932 (I 81~) 0 88~ 11853 II81 g 11786 D 78h O756 I) v28 I ) ~2 8 I) 2 8 O 728 D-28 0 728 0 679 I) 657 0 636 /I 598

1057 9 86 0 32 9 48 g 42 g 46 868 7 N~

7"2 6 96 oq" 6 t4 6 34 5 80 577 5 29 544 5 O9 5 24 402 5 15 4 88 a 72 4 48 4 71 4 53 4 3I 4 52 4 62 4 88 5 18 5 51 4 25 ~ 59 2 ')2 ~ 18

!

HARRISON

ET

AL

IN

CRUSHING

VACUO

EXPERIMENTS

AND K-FELDSPAR

ternative explanations for the K-correlated component, including the effects of recoil and voids in the K-feldspar, but were unable to reach a definite conclusion. The late release of additional 3SAra during thermal degassing, together with rising apparent ages, led them to infer that argon, released from these composite materials by an incompletely understood mechanism, had affected the age spectrum. We have extended their approach to a K-feldspar sample more typical of those used in our thermochronologic studies and can confirm the release of Cl-correlated argon during crushing. However, we conclude that these effects have negligible consequences on the interpretation of the age spectrum in terms of thermochronometric models. Observation of a consistent relationship between the release of Cl-derived argon and excess 4°Ar* among isothermal duplicate heating steps provides a key to seeing through the effects of the excess argon commonly observed in the initial heating steps of K-feldspars.

0.001

i

171

THERMOCHRONOMETRY

[

2. Results MH-10 K-feldspar, described in detail elsewhere [9], has been used extensively in the development of an interpretive model for 4°Ar/39Ar age spectra that assumes the presence of a distribution of diffusion domain sizes [6-10]. Its use in a test of the Turner and Wang proposal is ideal, because it has been the subject of thorough kinetic, XRD, T E M and visible light investigations, with the result that its microstructure and argon diffusion behavior are very well understood [9,10]. This sample contains regions of pristine K-feldspar 5 0 - 1 0 0 / x m in size, separated by zones characterized by a structurally modified feldspar. This latter material makes up ~ 5 vol.% of the sample and contains minor, apparently not connected, fluid-filled (?) inclusions that range from 0.05 to 1-2 /xm in diameter. About 20% of the K-feldspar domains defined by these features contain micron to sub-micron perthite development, but otherwise the K-feldspar is remarkably pristine

'

1200 (=J

i

i

I

i

i

i

i

i

i

1100

M H - I O'~

1000

0.0008

900 800 0.0006

~.~ 700 t~

0.0004

~

500

+Oo (y°~ o

0.0002

600

l _

_

400 0

300 200 100

o 0

0.0'005

0.001

0.0'015

0.002

~.K-a~"

0

0

6

C~/'K

0.2,1 I ,

0.1 0.2 0.3 0.4 015 0.6 017 018 0.9 1.0 SaAT=/S~IT X

Fig. 1. (a) Isochron plot of argon isotopes evolved from twenty crushing and three blank steps of irradiated MH-10.u K-feldspar. No linear array is apparent in these results, suggesting the presence of additional component(s). (b) Correlation plot of 4°Ar*/39Ar~ against 38Arcl/39ARK for the same results as in (a) in which 'atmospheric' 4°Ar has been removed by 36Ar'295.5. Note that 4 ° a r * / K = 4°Ar */39ARK. 9.72 × 10 s and C1/K = 38Arcl/39Ari,:. 0.277. The fit shown incorporates sixteen of the twenty crushing data but does not include the three 'blank' steps. Despite a reasonably good correlation (4°ARE/CI = 1.02 × 10-3), the meaning of this array remains obscure.

172

and undeformed. For example, MH-10 contains negligible dislocations and no 'voids' observable by T E M [9]. We have performed an in v a c u o crushing experiment on MH-10 using a simple arrangement in which the driver of an all-metal mini valve displaced a stainless steel shaft into a blank flange, shaped as a mortar. This mortar contained 24.3 mg of K-feldspar ~ 0.5 m m in size. Following baking over night, argon was released in 23 steps by progressively torquing the valve driver, eventually reaching 10 N-m. This method appears to be at least as effective in reducing grain size as 1000 crushes using a magnetically operated pestle [5]. Microscopic examination of the fine powder created by crushing reveals that about two-thirds of the grains contain at least one edge as large as 50-100 Ixm, with the rest mostly in the form of angular fragments up to 100 ixm in size. Only 4% of the sample was in the form of 150-200 Ixm sized particles and no grain was larger than 200 /xm. As relatively little gas was lost over the last two crushing steps, we infer that most of the argon accessible by this approach was released. Details of the irradiation environment, extraction system, mass spectrometry and correction factors are given in [8], with the exception of (38Ar/39Ar)K = 1.20 _+ 0.03 × 10 -2. The C I / K ratios were calculated using the relationship 38ArcI/39ArK • 0.277, determined by analysis of irradiated KC1. The results from both the crushing of the whole sample and step heating of an unfractionated aliquant of the resulting powder were designated MH-10.u. Analyses performed on the > 150 /xm split of the remaining powder were labelled MH-10.u'. All isotopic results are presented in Table 1.37Ar/39Ar ratios are not given because the sample was irradiated more than 1 yr prior to analysis. No correction was necessary for 36Ar produced b y / 3 - decay of 36C1. An isochron plot of the argon ratios obtained from the crushing experiment (Fig. la) yields a complex, J-shaped pattern. However, when atmospheric corrected (see caption of Fig. 1), sixteen of the 23 crushing steps yield a moderately well correlated array on a plot of 4OAr*/39ArK versus 38Arc|/39Ar K that extrapolates to a negative 4°Ar*/39ArK value for zero C 1 / K (Fig. lb). Not included in this fit are the three lowermost data on the left ('blank' steps in which the sample

T.M. HARRISON

E T AL.

remained under stress but the piston was not further advanced, including an overnight blank) and four of the five initial crushing steps (in the right uppermost corner), that fall off this line. In strong contrast to earlier results [4,5], the total 39At liberated during this process (4.42 × 10-15 mol) amounts to only 0.078% of the sample total. This small fraction, however, does account for 7 Ma of the total integrated K - A r age and appears to correspond broadly to the anomalously old ages found in the first few tenths of a percent of 39Ar released from conventionally heated splits [10]. We have since crushed two contrasting K-feldspar samples by this method (Cooma Granodiorite K-feldspar and Fish Canyon sanidine), and obtained only 0.084 and 0.11% of their total 39Ar, respectively. The similarity between these three results suggests that a value of ~ 0.1% may be typical of many alkali feldspars. A 9.68 mg split of the crushed powder was wrapped in Sn foil and heated at temperatures from 250 ° to 1500°C in a double-vacuum resistance furnace [8]. The resulting age spectrum has a pronounced peak at about 80% 39Ar released and yields an integrated total fusion age of 341.3 Ma, or 348.3 Ma when the 4 ° A r * / K released during crushing is added back. This latter age is indistinguishable from the average K - A r age of 347 + 2 Ma previously obtained for MH-10. The clearest differences between these results relative to the uncrushed material are: both the significantly lowered 38ArcJ39ArK values and the lack of plus ages over 109 yrs in the initial gas release, anomalously old ages at around 80% gas release that then drop sharply, and a decrease in argon retentivity (see diffusion coefficients in Table 1 compared to [10]). Only this latter effect was previously observed in fractions of MH-10 reduced in size to tens of microns prior to irradiation [10]. By sieving the remaining ~ 14 mg through a 100 mesh screen, 0.6 mg of > 150 txm grains were recovered. The separated fraction was wrapped in Sn foil and step heated at temperatures from 350 ° to 1550°C using procedures identical to those described above (Table 1). This analysis, designated MH-10.u', contrasts with MH-10.u in yielding both a similar age spectrum (Fig. 2) and Arrhenius data (Table 1) to the many results previously obtained for MH-10 [e.g., 6 -

IN VACUO

CRUSHING

EXPERIMENTS

AND

K-FELDSPAR

T H E R M O C H R O N O M E T R Y

lo-'J I 10-~ 40O

[ , i

:

K

'

t"d"~

"6" 35o

o

300

t~

&

<

MH-IO.t MH-IO.u'

250

200

-

o

,

~o

,

2o

3o

+o

,

so

+o

,

7o

~o

9o

~oo

Cumulative %39Ar released

Fig. 2. Age spectra for the thermal degassing results of MH-10.u (unfractionated sample) and MH-10.u' ( > 150 /xm grains) following in vacuo crushing. Also shown for reference is an age spectrum of MH-10 (MH-10.t grain size ~ 140/xm [10]) that was crushed prior to irradiation. The molar C1/K ratio for each heating step of MH-10.u is shown in the accompanying box. Although the age spectrum of the split crushed after irradiation has been profoundly modified compared to the control spectrum and that of the > 150/zm split, the K - A t age is unchanged.

8,10]. The incremental total fusion age, excluding the crushing results, is 346 Ma. This is within the expected bounds for MH-10, indicating that these > 150 ~ m crystals have not lost 4°ArE amounts comparable to that removed from the split containing particles well below the largest diffusion domain size. 3. Discussion 3.1 Cl-correlated 4°Ar from crushing

Although restricted to a much smaller fraction of total argon release, the 4°ArE/C1 = 1.02 _+ 0.17 × 10 -3 (chlorine-correlated excess 4°Ar*) suggested by the crushing results from MH-10.u (Fig. lb) is similar to the highest value reported by Burgess et al. [5] and a factor of four higher than Turner and Wang's sample, JX04 [4]. While Burgess et al. [5] attributed direct chronological significance to the age intercept of the 4°Ar*/

173

39ARK versus 38Arcl/39ArK plot, Turner and Wang [4] were more cautious. Several factors, including discordance between apparent ages derived in this way and by step or laser heating [5], large uncertainties [4,5], or the lack of an independent benchmark with which to assess concordancy [4], prevent definitive interpretation of the chronological meaning of these results. Our result (Fig. lb) suggests caution in the a priori interpretation of correlations arising from crushing as having significance for the age of the K-feldspar and associated features. Although even the linear array of data from the crushing experiment (Fig. lb) contains scatter in excess of analytical precision (MSWD = 7.9), the projected line is nearly 4 s away from passing through the K - A r age. A plot of all [4°Ar*] against [3SArcl] results (not shown) yields a broadly linear relationship that passes through the origin, suggesting no K-correlated excess 4°Ar*, but this line is heavily weighted by the 'blank' steps. All data on an 4°Ar/39ArK against 36Ar/39ArK against 38Arcl/39ArK plot (not shown) yield a positive 4 ° A r / 3 9 A r K intercept (corresponding to an age of ~ 600 Ma), but contain a relatively large dispersion. A similar 3-D analysis of only those results used in the 2-D regression (i.e., Fig. lb) results in a significantly better fit and necessarily also yields a negative intercept on the 4°Ar/39ArK axis. Possibly because of the dominance of the Cl-correlated argon, we have not been able to establish the existence in MH-10.u of a K-correlated component of 4°Ar* that is accessible via crushing. The 39Ar released during crushing may be derived all or in part from the K-feldspar structure. This is because the highly linear and reproducible Arrhenius plots obtained from MH-10 over the initial gas release (fig. 4a in [10]) do not suggest derivation of significant 39ARK from sources other than the K-feldspar structure. Consider the case in which the K-feldspar has been reduced in size to 20 x 20 × 20 /zm cubes. What fraction of the K-derived 39Ar will be exposed by the creation of these new surfaces? Given that the K-feldspar unit cell is ~ 10 .~ on edge, approximately 0.03% of the 39mrwould be directly adjacent to a freshly created boundary. This small fraction is in the order of that lost from MH-10 during crushing and we speculate that the 39Ar at these bound-

174

T.M. HARRISON

aries might be readily lost during the fracturing process. Although the effective particle size of the crushed aggregate is undoubtedly greater than 20 p~m, we expect the fractured surfaces would have a fractal character, thereby exposing a higher proportion of unit cells than that estimated by the calculation above. Even in the unlikely case in which all the 39ARK evolved during the crushing experiment was hosted by, for example, inclusions in the modified feldspar zones [9], the effect of a non-volume distribution of < 0.1% of the potassium on the Arrhenius plot would still be very small as this tiny fraction of 39Ar is usually released in the first one or two steps [10]. It is often these initial steps that yield diffusion coefficients slightly higher than those predicted by extrapolation of the higher t e m p e r a t u r e data [7,11], suggesting that many other K-feldspars may also be minimally affected by this potential complication. Provided K-feldspar samples yield robust and linear Arrhenius relationships in the initial stages of gas release, it appears unnecessary to perform in vacuo crushing routinely to assess the fraction of potassium held in non-volume sites. 3.2 Cl-correlated

4°mr from thermal degassing

To establish whether trends on Arrhenius plots are meaningful or only apparent, we have adopted the practice of performing isothermal duplicate heating steps [7]. During the step heating of MH10.u (and MH-10.u') in this fashion, we noticed a relationship b e t w e e n 38Arcl/39Ar K and apparent age. From 450 ° to 600°C, the first heating step at a given temperature yields both a significantly higher age and 38Ara/39ArK than the subsequent, duplicate step (note that the 250°C pair could not be used due to very high atmospheric contamination). Above 600°C (i.e., at > 3.5% of cumulative 39mr release), this correlation is no longer seen because C1/K values have dropped to background levels. To understand this relationship requires some additional background. We can write the total atmospheric-corrected 4 ° A r * released in the laboratory as the sum of two components: 4°Ar * = 4°ARE + 4°ArR

(1)

ET AL.

TABLE 2 Isothermal duplicate results used in Fig. 3 Temperature (°C)

CI/K (10-3)

4OAr*/K

450 450 800 500 550 550 600 600

8.338 0.609 4.986 0.415 1.634 0.054 0.499 0.054

4.442 2.304 3.497 2.185 2.397 1.896 2.032 1.955

4OAI-R/K (10.6)

(10-6)

A4OAI'*/K (10-6)

2131 2.135 2 116 2.069 1,944 1.881 1.893 1.939

ACI/K (10-3)

2.137

7.728

1313

4.570

0500

1.580

0.078

0.444

where 40ArR is in situ radiogenic 4°Ar. The ratio 4 ° A r * / K can be expressed as: 4°Ar*/K = (4°ArE/CI • C1/K ) + 4°ArR/K

(2)

Assuming that 4°ArE/C1 is constant across some interval of degassing, the difference between isotopic results from any two heating steps of equal age (i.e., A4°)krR/K = 0) is given by:

m4°mr* / K = 4°ArE/C1 • AC1/K

(3)

and thus a plot of A4°Ar*/K versus AC1/K should yield a linear relationship with a slope equal to the 4°Arz/Cl ratio. We have recast our isothermal duplicate data as follows: the 4°Ar * / K ratio of the initial step in the pair is subtracted from the second 4°Ar*/K ratio. This value is then plotted against AC1/K, calculated from the difference in 38Arcl/39ArK ratios of the contiguous isothermal steps. The 2.5

MH-10.U 4O

450OC -4

Are/Cl = (2.7-+0.2)x10

2.0

LS 5000

C

1.0

550~C

MH-10.U'

~00oC 00°C

0.5

0.0 / 0.0

~

(~(2.3-+0.3)x10"4 '

2.0

'

,

4.0

,

r

6.0

8.0

ACI/K (xl0 "3) Fig. 3. Plot of A4°Ar*/39ArK against A C I / K (molar) for the four isothermal step-heating pairs between 400 ° and 600°C for MH-10.u. The slope of the line corresponds to an 4°ARE/Cl = 2.7_+ 0.2 X 10-4, indicating the presencc of a Cl-correlated component of excess 4°At*. The initial isothermal duplicates from MH-10.u' (450-600°C)yield an average value of 2.3 _+0.3 x 10 -4, which is indistinguishable from that of MH-10.u.

IN VACUO C R U S H I N G E X P E R I M E N T S AND K-FELDSPAR T H E R M O C H R O N O M E T R Y

duplicate results corrected in this way are highly correlated and indicate an 4°mrE/C1 ratio of 2.7 ± 0.2 x 10 -4 (Table 2; Fig. 3). Although less precise because of their smaller total signals, the isothermal duplicates in the temperature range 450-600°C from the > 150 /xm split (MH-10.u', Table 1) yield an essentially identical average 4°ArE/C1 ratio of 2.3 + 0.3 X 10 -4. Isothermal duplicate results from two previously measured 'virgin' splits, MH-10.q and MH-10.r [10], were re-examined and found to contain 4°Ar~/C1 ratios of 3.1 x 10 - 4 and 3.5 x 10 -4, respectively. These somewhat higher values are not unexpected as both samples presumably still contained the elevated 4°ArE/CI component (i.e., 1.02 x 10 -3) found during in vacuo crushing of MH-10.u. Why is the first step of the isothermal duplicate pair disproportionately affected by the C1correlated excess argon component and why are the crushing and step-heating 4°mrE/C1 ratios different by a factor of four? The answer to the first question appears to be straightforward and we address it here, returning later to the second question. Attainment of a new peak temperature during laboratory degassing appears to be the key element in the release of Cl-correlated 4°Ar. The subsequent step at the same temperature appears to have little or no effect on releasing these isotopes. The potential for reaction due to exposure at a certain temperature is largely realized upon reaching that threshold and leads us to conclude that these correlated argon isotopes are derived from a system containing a spectrum of thermal stabilities. Heating at 450°C causes the more susceptible portions of this system to release their contained argon, while a second step at 450°C releases a higher fraction of volume-sited argon. When the temperature is raised to 500°C, features that had barely resisted deterioration now fail, releasing Cl-correlated argon. This process continues until all accessible features are exhausted. Both the degassing behavior during isothermal replicate steps and microstructural observations are consistent with this effect, arising from the decrepitation of fluid inclusions. Many factors control the strength of an inclusion, including size, shape, fluid density and composition, and heating (strain) rate [17-19]. As temperature is

175

raised, fluid pressure within sealed pores increases while the strength of the confining medium drops, eventually causing the host to fail. For a given geometry, larger pores will tend to rupture first. For a distribution of geometries containing a similar fluid pressure, irregularly shaped inclusions will tend to have higher stress in their tips compared to more equant ones and preferentially fail [20]. The most likely candidates to host the Cl-correlated 4°Ar are the apparently unconnected inclusions, typically with diameters of 0.05-0.5 t~m, that inhabit zones of 'modified' feldspar [9]. Although some of these features appear to persist when heat treated to 950°C, a new generation of pore trains, ~ 20 nm in size, is created by laboratory heating [9], and are probably the result of the decrepitation of larger inclusions. There is some indication that an extremely small-scale feature, making up no more than 0.1 vol.% of the sample, is produced by laboratory heat treatment [10]. This might reflect nano-scale fracturing of the modified zones. Further insight into the siting of the Cl-correlated component comes from results of samples of MH-10 heated in the laboratory prior to irradiation (e.g., MH-10.n in [10]). Despite experiencing temperatures as high as 1100°C, initial step heating results yield similar 38Aro/agArK values to those observed from the in vacuo crushing. That is, pre-irradiation heating has caused 4°Ar* to be released but the C1/K ratios in the early stages of thermal degassing are not significantly reduced. In vacuo crushing of a sister split preheated to 1100°C for 71 min prior to irradiation (MH-10.v) releases a similar fraction (0.11%) of agAr, a much greater (5 times) quantity of 4°At*, and about half the O-derived argon compared to MH-10.u (Table 1). It appears that physical changes to the sample during laboratory heat treatment have both permitted some of the 4°Ar* mobilized during heating to be captured in features accessible by crushing and resulted in the loss of C1. Although this particular agency would have no relevance to the capture of 4°Ar* during slow cooling, we do not rule out a related natural mechanism having such a property. As is often the case, apparent ages over the first few percent of gas release are anomalously old and highly variable. Using the relationship 4°ArE/C1 = 2.7 x 10 -4, we have applied a 4°ARE

176

T.M.

H A R R I S O N

ET

AL.

1 0 "2

Although the method described above was pivotal in our recognition of this effect, it is not absolutely necessary to perform isothermal duplicates to make the correction for Cl-correlated excess argon, provided that the 4°ArE/CI and 4°ArR/K ratios are constant throughout the contaminated portion of the age spectrum. However, because we have no a priori knowledge of whether an in situ radiogenic age gradient is present, it is advisable to restrict calculation of A to adjacent isothermal steps.

10a k~ 10 "~

600 /-

550

i L t

5OO

450 400 350

i_

if i i i

i i i

i

i

i i

i

- - - MH-10.U

(uncorrected)

- -

(CI

MH-10.u

corrected

ages)

i I_l

r . . . . . .

-

~. 300 7-

250

3.3 Cl-correlated spectrum

4°Ar and its effect on the age

_F-

200 150

I

0

1

I 2

C u m u l a t i v e % 39A r r e l e a s e d

Fig. 4. Age spectrum of the first 3% of gas released from MH-10.u showing the results corrected (solid line) and uncorrected (dashed line) for the presence of Cl-correlated excess 4°Ar. The product of the 4°ArE/CI (Fig. 3) and the C1/K (determined for each step via the 38Ar/39Arratio) yields an "excess age" that is simply subtradted from the measured 4°Ar*/4°K to yield an age corrected for the presence of excess argon. The corrected results suggest an age distribution consistent with continued cooling.

correction to the ages between 0.12 and 2% of 39Ar release (see caption in Fig. 4). Uncorrected, the age spectrum over the first few percent of gas release is highly variable but, following correction, reveals a geologically reasonable and broadly increasing pattern of ages (Fig. 4). We interpret this spectrum as recording the continuation of cooling of the host pluton to temperatures as low as ~ 125°C [6]. This apparent success supports the view of Turner and Wang [4] that correction for a Cl-correlated component should be straightforward, although, in this case, identification of the function arose from thermal degassing rather than in vacuo crushing. We have recently replicated this observation using a K-feldspar (XR-2A) from our studies of southern Tibet. This sample contains substantial Cl-correlated excess argon (4°ArE/CI = 1.45 __+0.04 × 10 -4) over the first 9% of gas release, with ages ranging between 16 and 71 Ma. However, using this approach, all eleven ages correct to a geologically meaningful value of 9_+2 Ma.

Midway through the thermal degassing of JX04 that followed in vacuo crushing, Turner and Wang [4] observed an increase in the 4 ° A r * / K ratio towards values associated with the K-correlated component, and later increased C I / K ratios. They speculated that the same reservoirs that had been tapped during the crushing were releasing extraneous argon late in thermal degassing. It was this possible relationship that inspired their cautionary statement with regard to thermochronometric interpretations. In contrast, our results for MH-10 do not indicate the presence of a K-correlated 4°ARE component released by crushing (section 3.1) but, rather, suggest that the origin of the increased C I / K is volume-sited chlorine and that the anomalous form of age spectra from crushed samples is due an artifact produced by the crushing experiment. C 1 / K ratios increase in the last 40% of gas release from values of 0.00005 to 0.0019 but even this latter C 1 / K is still more than two orders of magnitude lower than the peak values obtained during crushing (Table 1). Over this same range of 39Ar release, ages rise well above that of pluton emplacement. Worth noting are both the jumps in age associated with temperature increases between 1050-1150°C and the opposite trend with respect to the AC1/K of isothermal duplicates in this temperature range compared with the 450600°C steps. Could the anomalously old ages be due to either of the two distinctive 4°ArE/C1 components already identified? The answer appears to be no for two related reasons. Assuming a 'background' 38Aro/39ArK and age of 0.0002 (Table 1) and 373 Ma [21], respectively, a plot of

IN VACUO C R U S H I N G E X P E R I M E N T S A N D K - F E L D S P A R T H E R M O C H R O N O M E T R Y

Aa°Ar*/K versus AC1/K for the last 40% of gas release yields what could be interpreted as yet a third Cl-correlated component of 4°ARE (4°ArE/ CI = 8.0 + 0.2 X 10-5). We seriously doubt that this latter relationship is meaningful as we observe a similar increase in C1/K over the same portion of gas release in a ~ 140/xm split crushed prior to irradiation (MH-10.t in [10]) and in the > 150 /xm fraction of the crushed sample (MH10.u'). Both these samples yield plateau ages in the expected range of 365-373 Ma. In fact, log (r/r o) plots for both 38Aro and 39ARK released from MH-10.u show similar behavior over the last 80% of 39Ar release, suggesting that the associated chlorine is volume sited. The inference we make from these results is that the late released chlorine from MH-10.u, amounting to 50 ppm or about 75% of the sample total, is distributed within oxygen sites in the K-feldspar lattice. This conclusion is consistent with measurements of insoluble CI in K-feldspar, which suggest volume chlorine contents at least as high as we have measured [15]. Under this 0

'

177

interpretation, 38Arft is released from these sites only at high temperatures, due to their low activation energy a n d / o r the very sluggish nature of anion vacancy transfer under anhydrous conditions [12-14]. Direct ion imaging of the chlorine distribution in MH-10 confirms that high levels of C1 occur both within and adjacent to the zones of modified feldspar (Fig. 5) but that the vast majority of the feldspar contains relatively low (0.1-1 ppm) and uniformly distributed chlorine concentrations. As can be seen from Fig. 5, the regions of heterogeneous C1 are typically 5 x 15 /xm in size. However, TEM imaging [9] has shown that even the largest fluid-bearing inclusions in MH-10 are only 1-2 /xm in size and are more typically much smaller, indicating that the vast majority of the chlorine in this feldspar must indeed be volume sited. The close association between the 5 x 15 /xm regions of high chlorine concentration and the sub-micron inclusions, containing the Cl-correlated component of 4°mrE released during the initial stages of step-heating, raises a question.

9RP

25 24 23 21 2e 18 15 13 i0

Fig. 5. Scanning ion image of the 35C1- distribution in a modified zone of MH-10 K-feldspar. The image was acquired by a CAMECA 4f ion microprobe using a primary probe of 10 kV Cs ÷ at a mass resolving power of 2000 and using the normal incidence electron gun to neutralize sample charging. The imaged field is 75 x 75 tzm and the lateral resolution is 0.7 txm. Chlorine concentrations were quantified using an NBS glass containing ~ 50 ppm CI. In this image, relative chlorine abundances are shown by the grey scale where the background concentration is approximately 1 ppm. CI is heterogeneously distributed, with high concentrations persisting over 5 × 10 tzm regions. TEM imaging reveals that the largest fluid-bearing inclusions are 1-2/~m in size, and typically much smaller. Other ion images (e.g., 28Si) and previous TEM observations indicate that these regions are alkali feldspar and thus the vast majority of the chlorine is volume sited.

178

Assuming that the chlorine that is now sited in the K-feldspar structure was associated with a given 4°ArE/C1 in the precursor fluid phase, what happened to the related 4°ARE? During the episode at elevated temperature in which fluids were introduced, causing reprecipitation of small portions of the K-feldspar, this 4°ARE may have either been partitioned into the fluid phase (increasing the 4°ArE/C1 ratio) or become structurally bound and then subsequently lost, due to its very high diffusivity. Once the inclusions formed, continued evolution of t h e 4°ArE/CI ratio in the fluid may have occurred due to preferential partitioning of CI into the solid relative to 4 ° A r E.

Not surprisingly, the act of crushing has induced an artifact into the age spectrum. We carried out twenty age spectrum analyses on coarse splits ( > 130/.~m) of MH-10, including the age spectrum from the > 150 ~ m split of the crushed sample (MH-10.u') [7,8,10,21], and had never before seen the anomalous rise in 4°Ar/39Ar at 80% gas release (to apparent ages 25 Ma older than pluton emplacement [18]) followed by an abrupt decrease (Fig. 2). We have, however, seen this identical behavior in a split of MH-10 reduced in size to ~ 54 /xm prior to irradiation [10]. Thus, the cause of this anomalous effect must be due to crushing the particles smaller than the largest diffusion domain size (50-100 p~m in MH-10 [9,10]). The MH-10.u age spectrum (Fig. 2) appears to mimic features of JX04, suggesting that features in the JX04 age spectrum may also be a consequence of crushing. We speculate that the anomalous age spectrum behavior is a consequence of exposing regions of the crystal that were originally remote from a natural diffusion boundary. Small (a few percent) internal heterogeneities in the distribution of 4°Ar a n d / o r 39Ar could be envisaged (e.g., 4°gr* residing in defect clusters, 39Ar recoiled into K-poor zones) that would normally (i.e., during step-heating of the intact domain structure) be homogenized by the randomizing effects of diffusion or even by the eventual melting of the sample. By artificially creating internal surfaces via crushing, these anomalous zones could permit either relatively enhanced or retarded release of 4°Ar or 39Ar, which could lead to seemingly parentless 4°Ar* in a manner somewhat analogous

T.M. HARRISON ET AL.

to the 'unsupported' Pb* observed in complexly zoned zircons [22]. Since we see the anomalous age spectrum behavior in samples crushed both before and after irradiation, and because of the substantially enhanced diffusivity of 39Ar that results from crushing, we suspect 4°Ar* residing in higher-order defects [23] may be the culprit. Although an additional complication, these isotopic heterogeneities apparently do not materially affect thermochronometric reconstructions, provided that natural diffusion boundaries are not disturbed during laboratory analysis. Turner and Wang [4] express the view that the disturbed step heating pattern of their sample JX04 might conventionally be read as a sample having an age in excess of 1000 Ma that experienced episodic argon loss at about 350 Ma. This interpretation would presumably lead to an erroneous inference regarding the thermal history of such a sample if the increased apparent ages late in gas release were due to the 4°ArlUK component. Although we do not share their basic premise, having previously argued [11,16] that virtually identical patterns appear to be diagnostic of extraneous argon, the potential presence of excess argon is not an impediment to performing calculations such as the multi-domain analysis [6-8]. This is because evaluation of the relationship between inflections in the age spectrum and log ( r / r o) plots permits a self-consistency test of the assumptions used in calculating the model age spectrum. Indeed, in at least one case (FA-8 in [11]), we have used the multi-diffusion domain approach as a tool to confirm the presence of excess argon.

4. Summary Discovery through in vacuo crushing of CIand K-correlated components of 4°ARE in Kfeldspar-bearing samples [4], components that appear to re-emerge late in thermal degassing, stimulated us to perform similar measurements on a well-characterized sub-solvus igneous K-feldspar (MH-10). We confirm the presence of a chlorinerelated component ( 4 ° A r E / C I = 1.02 × 10 -3) from sites accessible by crushing MH-10, although it is associated with only 0.078% of the total sample potassium. However, we could not identify a K-correlated component. Isothermal

IN VACUO C R U S H I N G E X P E R I M E N T S AND K-FELDSPAR T H E R M O C H R O N O M E T R Y

duplicate heating steps reveal a Cl-correlated component (4°ArE/C1 = 2.7 × 10 -4) that appears to be related to decrepitation of fluid inclusions. There appears to be no adverse consequence arising from these effects that would significantly impact on a thermochronometric interpretation of the 4°gr/39Ar results for MH-10. Rather, identification of the 4°ArE/C1 component released during step heating permits us to see through the effects of excess argon, allowing a previously inaccessible portion of the age spectrum to reveal thermal history information. We strongly urge others to measure and report 3SAr results from K-feldspar samples routinely. The implications of Turner and Wang's [4] observations are potentially important and only beginning to be exploited. We expect that there is a continuum of behavior between the samples they describe and the one that we have documented here. However, it is possible that a reader may misinterpret their caution, arising from an effect seen in fine-grained sediments, as a criticism of the articles they cite as representative of thermoehronometry. One suggestion from our findings is that care should be exercised when extrapolating behavior observed in materials outside the realm of those commonly used for thermochronometry (e.g., chert and stromatolite, hypersolvus feldspars and authigenic overgrowths [4,5,24]), lest it divert attention from the fundamental characteristics shared by many K-feldspars originating at mid to deep crustal levels. Our results lead us to the following summary model for MH-10. Crushing in v a c u o releases 39ARK from K-feldspar sites adjacent freshly broken surfaces, 4°ArE sited in or adjacent to domain boundaries and inclusions, and 3SArcl from inclusions ( + domain boundaries). Initial step heating activates loss of 4°Ar* and 39ARK from the Kfeldspar structure by diffusion, but also causes decrepitation of fluid inclusions, formed during a sub-solidus alteration event, that release a distinctive 4°ArE/C1 composition. These features form progressively smaller inclusion trails but are eventually exhausted of fluid. Further step heating causes continued diffusive loss of 4°Ar* and 39ARK, but eventually reaches high enough temperatures that 3SAra originating in anion sites begins to be mobilized by diffusion and then released from the mineral. A small fraction of the

179

4°Ar* that may have been trapped in defects during transport in nature is released from these energy wells, diffusively mixed with the associated 39Ar, and ultimately released at domain boundaries.

Acknowledgements We thank Marty Grove for his interest in this study and assistance with the isotopic measurements, Ben McClellan for fabricating the crushing device on short notice, Didier Renard for the ion probe measurements, Kevin McKeegan and Roberto Gomez for providing the 3-D plotting and regression routines, Jeff Fillipone for helpful discussions, and Chris Roddick and Peter Zeitler for constructive reviews. Support for this research was derived from a grant from the Office of Basic Energy Research, Department of Energy.

References 1 I. McDougall and D.H. Green, Excess radiogenic argon in pyroxene and isotopic ages on minerals from Norwegian eclogites, Nor. Geol. Tidsskr. 44, 183-196, 1964. 2 S. Moorbath, Recent advances in the application and interpretation of radiometric age data, Earth Sci. Rev. 3, 111-133, 1967. 3 I. McDougall and T.M. Harrison, Geochronology and Thermochronologyby the 4°Ar/39Ar Method, OxfordUniv. Press, New York, 212 p. 4 G. Turner and Wang Songshan, Excess argon, crustal fluids and apparent isochrons from crushing K-feldspar, Earth Planet. Sci. Lett. 110, 193-211, 1992. 5 R. Burgess, S.P. Kelley, I. Parsons, F.D.L. Walker and R.H. Worden, 4°Ar-39Al analysis of perthite microstructures and fluid inclusions in alkali feldspars from the Klokken syenite, South Greenland, Earth Planet. Sci. Lett. 109, 147-167, 1992. 60.M. Lovera, F.M. Richter and T.M. Harrison, 4°Ar/39Ar thermochronometry for slowly cooled samples having a distribution of diffusion domain sizes, J. Geophys. Res. 94, 17917-17935, 1989. 70.M. Lovera, F.M. Richter and T.M. Harrison, Diffusion domains determined by 39Arrelease during step heating. J. Geophys. Res. 96, 2057-2069, 1991. 8 T.M. Harrison, O.M. Lovera and M.T. Heizler, 4°Ar/39Ar results for multi-domain samples with differing activation energy, Geochim. Cosmochim. Acta 55, 1435-1448, 1991. 9 J.D. Fitz Gerald and T.M. Harrison, Argon diffusion domains in K-feldspar I: Microstructures in MH-10, Contrib. Mineral. Petrol., in press. 10 O.M. Lovera, M.T. Heizler, and T.M. Harrison, Argon diffusion domains in K-feldspar If: Kinetic properties of MH-10, Contrib. Mineral. Petrol., in press.

180 11 T.M. Harrison, C. Wenji, P.H. Leloup, F.J. Ryerson, and P. Tapponnier, An Early Miocene transition in deformation regime within the Red River fault zone, Yunnan, and its significance for Indo-Asian tectonics, J. Geophys. Res. 97, 7159-7182, 1992. 12 T.M. Harrison and I. McDougall, Excess 4°Ar in metamorphic rocks from Broken Hill, New South Wales: Implications of 4°mr/39Ar age spectra and the thermal history of the region, Earth Planet. Sci. Lett. 55, 123-149, 1981. 13 S. Claesson and J.C. Roddick, 4°mr//39Ar data on the age and metamorphism of the Ottfjallet diorites, Sarv Nappe, Swedish Caledonides, Lithos 16, 61-73, 1983. 14 P.K. Zeitler and J.F. Fitz Gerald, Saddle-shaped 4°mr//39Ar age spectra from young, microstructurally complex potassium-feldspars, Geochim. Cosmochim. Acta 50, 1185-1199, 1986. 15 E.V. Kostetskaya, M.E. Markova and Z.I. Petrova, Distribution of chlorine in rocks and minerals of the Paleozoic Dzhilda granitoid complex (western Transbaykaliya), Geochem. Int. 6, 172-179, 1969. 16 D.A. Foster, T.M. Harrison, P. Copeland and M.T. Heizler, Effects of excess argon within large diffusion domains on K-feldspar age spectra, Geochim. Cosmochim. Acta 54, 1699-1708, 1990. 17 E. Roedder, Fluid Inclusions (Reviews in Mineralogy, Vol. 12), Mineral. Soc. Am., 644pp., 1984. 18 R.J. Bodnar, P.R. Binns and D.L. Hall, Synthetic fluid

T.M. HARRISON ET AL.

19

20

21

22

23

24

inclusions VI. Quantitative evaluation of the decrepitation behavior of fluid inclusions in quartz at one atmosphere confining pressure, J. Metamorph. Geol. 7, 229-242, 1989. D.L. Hall and J.R. Wheeler, Fluid composition and the decrepitation behavior of synthetic fluid inclusions in quartz, in: 4th Biennial Pan-American Conf. on Fluid Inclusions, Program and Abstracts (Dep. Earth Sciences, Univ. Calif., Riverside), p. 39, 1992. J.C. Jaeger and N.G.W. Cook, Fundamentals of Rock Mechanics, Chapman and Hall, London, 3rd Ed., 593pp., 1979. M.T. Heizler, D.R. Lux and E.R. Decker, The age and cooling history of the Chain of Ponds and Big Island Pond plutons and the Spider Lake granite, west-central Maine and Quebec, Am. J. Sci. 288, 925-952, 1988. L.P. Black, I.S. Williams, and W. Compston, Four zircon ages from one rock: the history of a 3930 Ma-old granulite from Mount Sones, Enderby Land, Antarctica, Contrib. Mineral. Petrol. 94, 427-437, 1986. P. Horn, E.K. Jessberger, T. Kirsten, and H. Richter, 39A1--4°mr dating of lunar rocks: effects of grain size and neutron irradiation, Proc. 7th Lunar Planet. Sci. Conf., pp. 1987-2008, 1975. J.P. Girard and T.C. Onstott, Application of 40Al'//39/Skr laser-probe and step-heating techniques to the dating of diagenetic K-feldspar overgrowths, Geochem. Cosmochim. Acta 55, 3777-3793, 1991.