Ž . Chemical Geology 166 2000 139–158 www.elsevier.comrlocaterchemgeo Dating crystalline groundmass separates of altered Cretaceous seamount basalts by the 40 Arr 39 Ar incremental heating technique Anthony A.P. Koppers a,b, ) , Hubert Staudigel a,b , Jan R. Wijbrans b a Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, UniÕersity of California, San Diego, La Jolla, CA, 92093-0225, USA b Laboratory of Isotope Geology, Vrije UniÕersiteit, De Boelelaan 1085, 1081 HV Amsterdam, Netherlands Received 22 December 1997; accepted 13 September 1999 Abstract Alteration of submarine basalts compromises geochronology using conventional KrAr and 40 Arr 39 Ar techniques. To help overcome these problems, we re-evaluated the potential of groundmass dating techniques. Incremental heating on acid-leached groundmass samples, following an overnight bakeout at 2008C and using high-resolution heating schedules, eliminated most of the low temperature alteration effects in the submarine basalts studied. More than 75% of the groundmass Ž . 39 analyses n s32 display accurate age plateaus consisting of 30–70% of the total amount of Ar released. More than 50% K of the analyses have plateau ages concordant with their total fusion ages implying minor or proportional loss of radiogenic 40 Ar U and 39 Ar . Overall, we could show a high degree of coherence between ages of groundmass separates and comagmatic K phenocrysts. This suggests that the dating of aphyric basalts, which previously has proven problematic, can be accomplished with increasing confidence as well. Adding these rock types to the list of datable submarine basalts significantly enhances our ability to understand the eruptive history of linear submarine volcanic chains. q 2000 Elsevier Science B.V. All rights reserved. Keywords: 40 Arr 39 Ar geochronology; Seamount basalts; Groundmass; Incremental heating; Seawater alteration; Recoil 1. Introduction Uncertainties in the crystallization ages of subma- rine basalts as measured by 40 Arr 39 Ar geochronology Ž . arise from natural causes seafloor alteration and analytical problems. Seafloor alteration processes Ž 40 U . may include the loss of radiogenic argon Ar from the original volcanic phases such as glass and ) Corresponding author. Fax: q 1-858-534-8090; e-mail: [email protected]mafic minerals, and the addition of seawater potas- Ž sium to the secondary alteration phases Kaneoka, 1972; Fleck et al., 1977; Seidemann, 1977, 1978, . 1988; Roddick, 1978 . Depending on the timing and the duration of these processes, the loss of 40 Ar U and the addition of potassium may have minor effects when these alteration processes are confined to a short period immediately following crystallization, or they may produce lower ages compared to the ex- pected crystallization ages when alteration has con- Ž tinued over extended periods of time Fleck et al., 0009-2541r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. Ž . PII: S0009-2541 99 00188-6
20
Embed
Dating crystalline groundmass separates of altered Cretaceous seamount basalts by the 40Ar/39Ar incremental heating technique
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Dating crystalline groundmass separates of altered Cretaceousseamount basalts by the 40Arr39Ar incremental heating technique
Anthony A.P. Koppers a,b,), Hubert Staudigel a,b, Jan R. Wijbrans b
a Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, UniÕersity of California, San Diego, La Jolla, CA,92093-0225, USA
b Laboratory of Isotope Geology, Vrije UniÕersiteit, De Boelelaan 1085, 1081 HV Amsterdam, Netherlands
Received 22 December 1997; accepted 13 September 1999
Abstract
Alteration of submarine basalts compromises geochronology using conventional KrAr and 40Arr39Ar techniques. To helpovercome these problems, we re-evaluated the potential of groundmass dating techniques. Incremental heating onacid-leached groundmass samples, following an overnight bakeout at 2008C and using high-resolution heating schedules,eliminated most of the low temperature alteration effects in the submarine basalts studied. More than 75% of the groundmass
Ž . 39analyses ns32 display accurate age plateaus consisting of 30–70% of the total amount of Ar released. More than 50%K
of the analyses have plateau ages concordant with their total fusion ages implying minor or proportional loss of radiogenic40ArU and 39Ar . Overall, we could show a high degree of coherence between ages of groundmass separates and comagmaticK
phenocrysts. This suggests that the dating of aphyric basalts, which previously has proven problematic, can be accomplishedwith increasing confidence as well. Adding these rock types to the list of datable submarine basalts significantly enhancesour ability to understand the eruptive history of linear submarine volcanic chains. q 2000 Elsevier Science B.V. All rightsreserved.
mafic minerals, and the addition of seawater potas-Žsium to the secondary alteration phases Kaneoka,
1972; Fleck et al., 1977; Seidemann, 1977, 1978,.1988; Roddick, 1978 . Depending on the timing and
the duration of these processes, the loss of 40ArU andthe addition of potassium may have minor effectswhen these alteration processes are confined to ashort period immediately following crystallization, orthey may produce lower ages compared to the ex-pected crystallization ages when alteration has con-
Žtinued over extended periods of time Fleck et al.,
0009-2541r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.Ž .PII: S0009-2541 99 00188-6
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158140
.1977; Pringle, 1993; Lo et al., 1994 . Problems alsoare caused by recoil of nucleogenic isotopes during
Žthe irradiation of these basalts Turner and Cadogan,1974; Dalrymple and Clague, 1976; Huneke, 1976;Huneke and Smith, 1976; Seidemann, 1978; Walker
.and McDougall, 1982 . Very fine-grained magmaticphases and secondary minerals in the altered basaltswill not retain all 39Ar and 37A during irradiation.K Ca
Recoil of 39Ar from high potassium, low tempera-KŽ .ture LT sites into low potassium, high temperature
Ž .HT sites leads to an increase in apparent age forŽ .the LT increments Huneke, 1976 , whereas recoil of
37A introduces a decrease in the apparent agesCa
when the calcium sites are degassing in the HTincrements during incremental heating experimentsŽ .Davis et al., 1989; Pringle, 1993 .
For these reasons, the occurrence of alterationphases in submarine basalts diminishes the suitabilityof whole rock dating by the conventional KrAr
Žmethod e.g., Kaneoka, 1972; Ozima and Saito, 1973;Baksi, 1974; Dalrymple and Clague, 1976; Ozima et
.al., 1976; Seidemann, 1977, 1978 , and to a greatextent, by the 40Arr39Ar incremental heating methodŽe.g., Ozima et al., 1977; Seidemann, 1988; Pringle,
.1993; Lo et al., 1994; Pringle and Duncan, 1995a,b .One approach in avoiding the problems due to alter-ation is the dating of unaltered phenocryst phases,whenever available in the basalts. Another approachis the dating of well-crystallized groundmass sepa-rates by the 40Arr39Ar incremental heating tech-nique. In previous studies, this technique has been
Žused successfully for dating fresh and young -15. ŽMa basaltic rocks e.g., McDougall and Harrison,
.1988; Nauert and Gans, 1994; Sharp et al., 1996 . Inthis paper, we describe the development and applica-tion of this technique to the dating of altered Creta-ceous seamount basalts. We present details of samplepreparation and mass spectrometry that were usedfor minimizing the possible alteration disturbance onthe 40Arr39Ar results. Groundmass ages are thencompared with age determinations of phenocrysts
Ž .from the same rock plagioclase, hornblende or withgroundmass separates from other rocks sampled atthe same drillrdredge site. Finally, the argon de-gassing patterns are discussed in the light of thealteration effects, recoil and K–Ca geochemistry.
Overall, we can show that our technique providesreliable crystallization ages in about 75% of all
groundmass experiments. When used with care, thistechnique extends our capability to date submarinebasalts that do not contain datable phenocryst phasesŽ .biotite, hornblende, plagioclase, nepheline, etc. .This is particularly useful when working with lavasfrom the shield building phase of submarine in-traplate volcanoes, where these datable phases arenot very common. Our experience from a compre-hensive study of Cretaceous seamounts in the West-ern Pacific suggests that addition of these techniquessubstantially improves our ability to resolve the ageprogression along submarine volcanic chains and,
For this study, we selected 12 typical groundmassor aphyric basalt separates from 32 basalts studied in
Žthe West Pacific Seamount Province WPSP, Fig. 1;.Koppers, 1998 , where we can compare groundmass
analyses with 40Arr39Ar age determinations of co-magmatic phases or related samples. Analytical re-sults on groundmass samples not discussed in this
Ž .paper can be found in Koppers 1998 and KoppersŽ .et al. 1998 . Samples were obtained from the
TUNES 6 expedition of the Scripps Institution ofŽ .Oceanography, the Ocean Drilling Program ODP
Ž .Leg 144 Premoli Silva et al., 1993 , and Y.Ž .Pushcharovsky USSR Academy of Sciences . The
selected samples range in age from 85 to 120 Maand come from five different seamount trails in the
Ž .WPSP Table 1 .Most of the WPSP basalts studied are highly
vesicular indicating near-surface extrusion and equi-Žlibration with atmospheric argon Dalrymple and
.Moore, 1968 . This is supported by the subaerial orrelatively shallow submarine eruption for most sam-
Žples Table 1; Winterer et al., 1993; Larson et al.,.1995 . The samples are alkali basalts or more evolved
Ž .varieties hawaiites, trachybasalts that have almostŽ .completely 98–100% crystallized groundmasses
with average crystal sizes between 10 and 30 mm.However, two samples from Maloney guyot containsubstantial quantities of microcrystalline material
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 141
Ž .MAL-2: 85% and MAL-5: 75% . Plagioclase andTi-magnetite are common constituents and rangefrom 10% to 88% and 5–40% in modal composition,respectively. Ti-augite abundance varies from absentin the hawaiites up to 40% in the ankaramites.Olivine, now entirely replaced by clay minerals, is
Ž .not a major constituent in the groundmasses 0–10% .K-feldspar is present only in the trachybasalts, whereit occurs either as laths or as rims surroundingplagioclase microlites. The groundmass separatesstudied typically are enriched in potassium compared
Ž .to their whole rock samples Appendix B .
3. Analytical methods
3.1. Mineral separation and acid-leaching
During sample preparation, we aimed to minimizethe quantity of alteration products in the materialanalyzed, including aphyric basalts, groundmassesand mineral separates. Alteration zones were re-moved by sawing, and the remaining material wascrushed in a hardened steel jaw crusher followed by
Ž .Fig. 1. Sample locations within the West Pacific Seamount Province WPSP . These Cretaceous seamounts originated from hotspotŽ .volcanism, which is associated with the South Pacific Isotopic and Thermal mantle Anomaly SOPITA in the French Polynesian area
Ž .Staudigel et al., 1991; Koppers, 1998; Koppers et al., 1995, 1998 . The DUPAL isotopic anomaly and present-day active hotspots of theŽ .Pacific Ocean filled circles are plotted for reference.
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158142
Tab
le1
Sum
mar
yof
40A
rr39
Ar
data
ofgr
ound
mas
san
dco
-mag
mat
icm
iner
alse
para
tes
Lab
code
Loc
atio
nD
epth
sA
gesp
ectr
umIn
vers
eis
ochr
onT
otal
fusi
onIC
P-A
ES
3940
36Ž
.Ž
.Ž
.L
atit
ude
Lon
gitu
deD
redg
eE
rupt
ion
Roc
kS
ampl
eA
geM
aK
rC
aA
rM
SW
Dn
Age
Ma
Arr
Ar
MS
WD
Age
Ma
Kr
Ca
KO
CaO
Kr
Ca
2Ž
.ty
pety
pe%
inte
rcep
tM
inM
axin
terc
ept
Lim
alok
guyo
tL
IM-3
5.55
172.
3512
55su
baer
ial
basa
nite
gmdm
68.8
"1.
01.
0147
25.9
468
.2"
0.8
0.04
0.56
15.6
70.
04L
IM-4
5.55
172.
3512
55su
baer
ial
basa
nite
gmdm
68.0
"0.
60.
6341
0.6
574
.3"
1.9
0.01
0.32
15.2
60.
02
Wod
ejeb
ato
guyo
tW
OD
-111
.90
164.
9213
35su
baer
ial
haw
aiit
eap
hba
s82
.1"
0.6
0.84
3721
.56
82.2
"0.
50.
261.
738.
820.
23W
R-c
ore
80.9
"1.
255
310
7.4
"1.
4W
OD
-10
11.9
016
4.92
1335
suba
eria
lha
wai
ite
WR
-cor
e81
.9"
1.4
935
85.2
"1.
0
Nee
n-K
oiaa
kgu
yot
NE
K-1
14.2
816
0.92
1500
suba
eria
lha
wai
ite
plag
103.
1"
0.7
0.02
100
0.6
1110
3.0
"0.
729
9.8
"13
.70.
610
3.1
"2.
10.
02gm
dm10
4.5
"0.
80.
1245
10.1
396
.4"
0.7
0.08
0.87
10.2
40.
10N
EK
-214
.28
160.
9215
00su
baer
ial
haw
aiit
epl
ag10
2.5
"0.
90.
0254
0.9
510
2.5
"0.
929
2.5
"13
.70.
996
.9"
3.3
0.02
gmdm
102.
0"
0.5
0.37
486.
610
98.6
"0.
50.
241.
688.
020.
24
Vli
nder
guyo
tV
LI-
417
.12
154.
3322
0065
040
0ha
wai
ite
plag
95.4
"1.
50.
0149
5.2
495
.2"
1.9
305.
0"
59.2
5.0
92.8
"1.
60.
01gm
dm96
.6"
0.7
0.34
416.
65
96.2
"0.
60.
191.
478.
800.
19
Ita
Mai
Tai
guyo
tIT
A-1
12.9
815
6.68
4000
2890
2580
haw
aiit
epl
ag12
0.0
"0.
80.
0364
1.7
612
0.0
"0.
829
5.8
"2.
51.
512
0.4
"1.
40.
03gm
dm11
7.9
"0.
90.
5442
5.9
711
2.3
"1.
10.
451.
934.
290.
52IT
A-2
12.9
715
6.75
2700
1590
1280
haw
aiit
epl
ag11
8.5
"0.
80.
3510
013
.08
118.
5"
0.8
293.
0"
25.5
13.5
118.
5"
0.7
0.34
gmdm
118.
0"
0.7
0.80
499.
16
116.
4"
0.8
0.60
3.09
5.24
0.69
Jenn
ings
guyo
tJE
N-1
20.8
715
6.17
2600
1030
850
alk
basa
ltgm
dm10
3.4
"0.
50.
5131
4.7
610
4.7
"0.
60.
211.
6610
.11
0.19
JEN
-420
.87
156.
1726
0010
3085
0an
kara
mit
egm
dm10
3.5
"0.
60.
4860
25.1
1410
1.6
"0.
50.
151.
3912
.54
0.13
Mal
oney
guyo
tM
AL
-221
.05
157.
1525
0093
077
0an
kara
mit
eap
hba
s97
.7"
0.6
0.21
3213
.66
96.5
"0.
60.
111.
2210
.32
0.14
MA
L-5
21.0
515
7.15
2500
930
770
hbl
basa
nite
hbl
100.
7"
0.7
0.08
773.
76
100.
8"
0.7
288.
1"
11.5
3.9
101.
9"
1.6
0.03
gmdm
100.
7"
1.5
0.08
4611
.76
105.
9"
1.7
0.03
0.54
9.74
0.06
Kr
Ca
rati
osar
eca
lcul
ated
asw
eigh
ted
mea
nsfo
rth
eag
epl
atea
use
gmen
ts,
and
usin
gre
com
bine
dto
tals
of39
Ar
and
37A
rfo
rth
eto
tal
fusi
ons.
The
ICP
-AE
Sda
taar
eex
pres
sed
asw
t.%
and
repr
esen
tth
eco
mpo
siti
onK
Ca
prio
rto
irra
diat
ion
ofth
egr
ound
mas
ssa
mpl
es.
MS
WD
valu
esfo
rth
eag
epl
atea
usan
din
vers
eis
ochr
ons
are
calc
ulat
edus
ing
Ny
1an
dN
y2
df,
resp
ecti
vely
.R
epor
ted
40A
rr39
Ar
date
sar
ew
eigh
ted
age
esti
mat
esan
dŽ
.Ž
.er
rors
onth
e95
%co
nfid
ence
leve
lin
clud
ing
0.2
–0.
3%st
anda
rdde
viat
ions
inth
eJ
valu
e.S
ampl
esw
ere
mon
itor
edag
ains
tT
aylo
rC
reek
Rhy
olit
eT
CR
sani
dine
27.9
2M
a;D
alry
mpl
ean
dD
uffl
ield
,19
88.
WR
-cor
eŽ
.Ž
.an
alys
esof
WO
D-1
and
WO
D-1
0ar
efr
omP
ring
lean
dD
unca
n19
95b
.T
heer
upti
onde
pths
for
the
WP
SP
seam
ount
basa
lts
are
mod
eled
base
don
ath
erm
alsu
bsid
ence
John
son
and
Car
lson
,19
92of
the
unde
rlyi
ngŽ
.Ž
.Ju
rass
icoc
eani
ccr
ust
that
was
part
iall
yre
juve
nate
d40
–60
%be
twee
n10
0an
d80
Ma
Sch
lang
eret
al.,
1984
;N
agih
ara
etal
.,19
96;
only
inca
seof
the
olde
rIt
aM
aiT
aigu
yot
asu
bsid
ence
hist
ory
prio
rto
Ž.
reju
vena
tion
has
been
incl
uded
.A
naly
tica
lda
taw
ill
beav
aila
ble
upon
requ
est
akop
pers
@uc
sd.e
du.
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 143
grinding in a steel ring swing mill. Grinding wasaccomplished in 1–2 s time steps, alternated withsieving at 250 mm until 90% of the rock powder wasreduced to -250 mm. The remaining 250–500 mmfraction included the hardest components of the al-tered basalt and are, by assumption, the freshest partsof the groundmass and phenocrysts. Mineral sepa-rates were prepared from the 120–250 or 250–500mm fractions, whereas groundmass and aphyric basaltseparates were prepared from the 250–500 mm frac-tion. The groundmass separates analyzed typicallycontained grains including more than 10–20 rela-tively large but individual groundmass crystals, withthe exception of the two samples from Maloneyguyot that contained significant quantities of finercrystalline material.
Plagioclase and hornblende separates were ob-tained by heavy liquid separation using an
Ž .overflow-centrifuge Ijlst, 1973 , followed by mag-netic separation using a Frantz magnetic separatorand hand-picking to remove the remaining alteredgrains. Groundmass and aphyric basalt grain sepa-rates were obtained by hand-picking of grains withneither alteration products nor pheno- and xenocrysts.
The separates were further cleaned by acid-leach-Ž . Žing with 3.5 N HCl ;60 min , 7% HF ;5 min;. Ž .only for plagioclase and 1 N HNO ;60 min in3
an ultrasonic bath that was heated to approximately508C. After acid-leaching, the groundmass grainswere checked using an optical microscope for thepresence of secondary phases, and where requiredacid-leaching was extended for at least 60 min in 7N HCl and for 60 min in 1 N HNO until 90–100%3
of the secondary phases were dissolved. Finally, theleached separates were washed in ultraclean water,
Ž .dried on a hot plate ;708C , and hand-picked for asecond time to remove the grains still containingremnants of alteration.
3.2. Irradiation and laser incremental heating tech-nique
Samples were wrapped in Al foil, loaded in 9 mmID quartz tubes alternated with packages of TCR-
Žsanidine flux monitor standard 85G003, 27.92 Ma;.Dalrymple and Dufflield, 1988 at one out of five
positions, and irradiated for 12–18 h in the Oregon
State University TRIGA reactor CLICIT facilityŽ .Dodd and Anderson, 1994 . Values of the irradia-tion parameter J for individual sample packageswere calculated by parabolic interpolation betweenthe measured standards. Estimated uncertainties forJ are between 0.2% to 0.3% standard deviation.Corrections for interfering reactions involving Ca
Ž .and K are outlined in Wijbrans et al. 1995 .Argon extractions for incremental heating were
Ž .performed at the Vrije Universiteit Amsterdam byusing a defocused, continuous argon ion laser beamon singlerdouble layered grain samples. Sampleamounts for groundmass, hornblende and plagioclaseanalysis were 10–30, 5 and 7–10 mg. Emphasis wasplaced on minimizing the problems that may arisefrom uneven heating of the sample grains with theargon laser. For this reason, we used only grain sizes
Ž . Ž-500 mm cf. Hall, 1990 , the laser beam 454.5–.514.5 nm wavelength was defocused to a focal
Ždistance of 580–590 mm main focusing lens is an.f:500 achromate and the low-energy halo around the
beam was cut out by use of an iris-diaphragmŽ .TEM00 mode . The heating time for each gas ex-traction step was set to 5–7 min to allow for heatingevery grain twice by moving the sample holder
Žslowly along parallel lines of an X–Y grid once in.the X-direction, and once in the Y-direction . Lastly,
the number of incremental heating steps was ex-tended to ensure a thorough heating of the sample atall temperature segments and to provide more detail
Žon the outgassing behavior of our samples minerals.8–12 and groundmasses 16–30 increments . During
incremental heating an emphasis was placed on theŽ .LT segment 0.05–0.50 W laser power to maximize
removal of LT alteration and atmospheric compo-nents. Groundmass gas fractions were purified for5–7 min using a Zr–Al SAES AP10GP getterŽ .4008C , followed by a combined 8–10 min purifica-tion including a double Fe–V–Zr SAES St172 getterŽ .2508C . The purified gas fractions were analyzed on
Ža MAP-215r50 mass spectrometer Wijbrans et al.,.1995 .
Newly developed data reduction software at theŽVrije Universiteit ArArCALC Õ16; Koppers, 1998;
.see also http:rrwww-pacer.ucsd.edurararcalc.htmŽ .follows Dalrymple et al. 1981 . Linear or asymp-
totic curve fitting was used to reduce peak intensitiesŽwith respect to inlet time McDougall and Harrison,
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158144
.1988 . Blank intensities were measured every 3–5sample runs. Incremental heating plateau ages andisochron ages were calculated as weighted meansusing 1rs 2 of the 40Arr39Ar ratio as weighting
Ž .factor Taylor, 1982 and as YORK2 least-squaredŽlinear fits with correlated errors York, 1969; Brooks
.et al., 1972 . In addition, mean squared weightedŽ w xdeviations MSWDsSUMSr Nyn ; York, 1969;
.Roddick, 1978 have been calculated for both thew xplateau and the isochron ages based on Ny1 and
w xNy2 df , respectively. Age plateaus are defined bythose incremental heating steps that have ages falling
2 2in the 1.96 s qs confidence envelope, where(Ž .1 2
s and s are the standard deviations of the lowest1 2Žand highest age in the plateau after Dalrymple and
.Lanphere, 1969 . Note that in this work we do notspecify a minimum amount of the total gas releasewhen defining an age plateau, as occurs in thecommonly accepted definition of an age plateauŽ .Section 4.2; Appendix A . If the scatter around theage plateaus or isochrons is beyond analytical errorŽ .MSWD)1 then the reported analytical error is
' Žmultiplied by the MSWD York, 1969; Kullerud,.1991 .
3.3. Chemical analyses
Ž .Groundmass samples 4–20 mg were analyzedfor K and Ca using an Inductive Coupled Plasma-
ŽAtomic Emission Spectrometer ICP-AES Varian.Liberty — Series II at the Vrije Universiteit. Elec-
tron microprobe analyses of the groundmass phaseswere performed on a JEOL. JXA8800M instrumentat the Vrije Universiteit. The precision of the elec-tron microprobe and ICP-AES analyses are betterthan 1% and 3%, respectively. Analysis of thegroundmass phases was possible if the grain size ofthese microlites exceeded 10 mm. However, theinterstitial spaces generally contain microlites smaller
Ž .than 1 mm or they partly consist of devitrifiedvolcanic glass. Abundances measured from such in-terstitial material using spot sizes of 10 mm, there-
Ž .fore, represent the modal distribution of devitrifiedglass and microlites, instead of primary glass compo-sitions. Due to sample heterogeneity within the
.in potassium the accuracy of the ICP-MS analysesmay be decreased, in particular, for samples withlow potassium content. Groundmass samples con-taining abundant needles of apatite or poikilitic pla-gioclase may show an increased variation in calciumcontent.
4. Results
The 40Arr39Ar release patterns of 12 well-crystal-lized, basaltic groundmass samples were studied foreffects of alteration and recoil, the minimization ofthese effects, and the relation to groundmass geo-chemistry. To show the reliability of the derived40Arr39Ar groundmass ages, these ages are compared
Žto ages of comagmatic mineral separates six sam-.ples or additional samples from ODP Leg 144 drillŽ .sites Limalok and Wodejebato guyot and TUNES 6
Ž . 40 39dredge sites Jennings guyot . The Arr Ar resultsand the KrCa ratios determined by ICP-AES for thebulk groundmass separates are summarized in Table1; the K–Ca geochemistry determined by electronmicroprobe analyses are given in Table 2. Reportederrors on the ages are two times the standard error.
4.1. Mineral 40Arr39Ar data
Coarse grained phenocrysts are less susceptible toalteration than the groundmass. They are more likelyto preserve pristine radiogenic 40Arr39Ar signaturesrepresenting the crystallization age of seamountbasalts. For this reason, the ages of six comagmaticplagioclase and hornblende phenocryst samples arereported that later will be used for a first ordercomparison with the results of groundmass analyses.
The 40Arr39Ar age determinations on these min-Ž .eral separates Table 1; Fig. 2 meet all quality
criteria for interpreting 40Arr39Ar experiments onvolcanic rocks, as outlined in Appendix A. They
Ž . Ž . Ž .have 1 broad age plateaus 54–100% , 2 concor-Ž .dant plateau, isochron and total fusion ages, and 3
isochron 40Arr36Ar intercept values are equal toŽ .modern atmosphere 295.5 . These ages, therefore,
can be accepted as geologically meaningful reference
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 145
Petrography and chemistry of the analyzed groundmass separates. Electron microprobe analyses are expressed as wt.% and include standarddeviations of the mean. If this standard deviation is )35% then the entire measured range is listed.
ages for the groundmass analyses. The high qualityof the mineral age determinations is furthermoredemonstrated by the reproducibility of two plagio-
clase samples from the same dredge site at Neen-Koiaak guyot which yield concordant ages of 103.1
Ž . Ž ."0.7 Ma NEK-1 and 102.5"0.9 Ma NEK-2 .
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158146
Ž .The low KrCa ratios for the plagioclases -0.028Ž .and hornblende 0.034 measured indicate that the
mineral separates are free of alteration and representŽ .true primary phases cf. Sebai et al., 1991 . The
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 147
Fig. 2. Groundmass, plagioclase and hornblende 40Arr39Ar incremental heating analyses. For most groundmass analyses, three additionalŽ . 39 Ž . 40 Uanalytical parameters are plotted upper panels against the amount of Ar released: i the radiogenic component Ar expressed as a
40 Ž . 39 37 Ž . 37percentage of the amount of Ar measured, ii the molar KrCa ratio based on the Ar over Ar ratio, and iii the release of ArK Ca Ca
expressed as a percentage of the total amount of 37Ar released during the experiment and normalized to increment size. ReportedCa40Arr39Ar dates are weighted age estimates and errors at the 95% confidence level including 0.2–0.3% standard deviation in the J value.
Ž . Ž .All samples were analysed using Taylor Creek Rhyolite TCR sanidine 27.92 Ma; Dalrymple and Dufflield, 1988 as a standard.
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158148
ITA-2 plagioclase forms the only exception with asignificantly higher KrCa of 0.34 that is explained
Žby the high potassium content K Os0.49"0.122.wt.% of these phenocrysts.
4.2. Groundmass 40Arr39Ar data
Groundmass experiments yield characteristic ageŽ . Ž .spectra Fig. 2 that typically display 1 high appar-
Ž .ent ages for the LT increments, 2 age plateaus thatcomprise less than 70% of the total amount of argon
Ž . Ž .gas released Fig. 3 , and 3 low apparent ages forthe HT increments accompanied by the first signs ofpartial melting. The LT discordant sections tend todefine ‘‘exponential’’ curves starting at high appar-ent ages that monotonically decrease towards the ageplateaus. Initial outgassing steps are characterized by
Ž .high atmospheric components ;80% merging intoage plateaus with highly radiogenic componentsŽ40 U .Ar )95%, Fig. 2 . Moreover, the KrCa ratiosfor the LT increments are higher than the ratios ofthe bulk groundmass and the age plateau; they arecompatible with the observed KrCa ratios of the
Žalteration minerals Table 2; Mankinen and Dalrym-.ple, 1972; Pringle, 1993; Lo et al., 1994 . This is
demonstrated in Fig. 4 where the apparent ages areplotted vs. KrCa ratio for six representative sam-ples. On one hand, these diagrams display an in-
Žcreased spread in KrCa values for the LT higher. Ž .KrCa and HT lower KrCa increments reflecting
Fig. 3. Histogram for the groundmass age plateau widths. Theseplateau widths are calculated as cumulative amounts of 39ArK
included in the plateau ages expressed as a percentage of the totalamount of 39Ar released.K
the heterogeneous style of degassing. On the otherhand, these diagrams display a tight clustering forthe age plateaus at KrCa values higher than that forthe bulk groundmasses. The more constant KrCaratios for the age plateaus suggest a rather homoge-neous argon source that has not significantly beendisturbed by alteration and recoil. More importantly,the higher KrCa ratios for the age plateaus suggestthat most of this radiogenic argon is retained in highpotassium minerals or, otherwise enriched,patchesrsegregations in the groundmass.
Plagioclase and clinopyroxene typically degas athigh temperatures and cause a lowering of the appar-ent ages, as is observed for pyroxene in experiments
sions to low apparent ages at the HT incrementscomprise between 22% and 55% of the total amountof argon released, and are characterized by decreas-ing KrCa ratios towards low plagioclase and
Ž .clinopyroxene values -0.04 , and a protracted re-37 Žlease of 50–90% of the total amount of Ar TableCa
.3; Figs. 2 and 4 . These observations indicate thatplagioclase and clinopyroxene start degassing at hightemperatures and are not significantly contributing tothe age plateaus. Moreover, mass balance calcula-tions for CaO show that plagioclase microlites onlyrepresent a maximum of 30 vol% of the bulk potas-sium inventory in the groundmass samples studiedŽ .Table 3 . This leaves the high-potassium, glassy or
Žmicrocrystalline spaces between microlites - 2.vol%; Table 2 as the major potassium source con-
tributing to the age plateaus. For more evolved rocksthe main potassium reservoir may include K-feldsparŽ .Table 2 .
Ž .The groundmass age of ITA-1 117.9"0.9 MaŽ .is younger than its plagioclase age 120.0"0.8 Ma
but is concordant with the groundmass and plagio-Žclase ages of ITA-2 118.0"0.7 Ma; 118.5"0.8
.Ma . This may suggest that the younger groundmassage represents the most accurate of both analyses,and that the older plagioclase age may be the resultof either contamination with xenocrysts or incom-
Žplete degassing during eruption in deep water ITA-1was dredged from the flank of Ita Mai Tai guyot,
.4000 m deep . All other mineral–groundmass pairsyield ages indistinguishable at the 95% confidence
Ž .level Fig. 2 . The groundmass ages from lava flows
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 149
Fig. 4. Age vs. KrCa diagrams for groundmass 40Arr39Ar analyses. Squaress increments included in age plateau, circless low and hightemperature increments displaying discordant gas fractions. APsage plateau increments; LTs low temperature increments; HTshigh
Ž . Žtemperature increments. For comparison the bulk groundmass KrCa ICP-AES analysis and the groundmass plagioclase KrCa electron.microprobe analysis have been plotted as vertical bars representing the measured range in KrCa ratio. However, the patterns of the
Ž . Ž .depleted samples LIM-3 K Os0.32 wt.% and LIM-4 K Os0.56 wt.% are not representative for the groundmass analyses. Both2 2
experiments show high KrCa values for the age plateaus up to nine times their bulk KrCa values, indicating the presence of K-rich argonŽ .reservoirs nepheline in these basanitic groundmasses.
Ž .drilled at sites 873 Wodejebato guyot, WOD andŽ . Ž871 Limalok guyot, LIM from ODP Leg 144 Pre-
.moli Silva et al., 1993 can also be readily com-pared, because of their close stratigraphic relation:
ŽWOD-1 82.1"0.6 Ma; 80.9"1.2 Ma after Pringle.and Duncan, 1995b is sampled only 9.4 m above
Ž . ŽWOD-10 81.9"1.4 Ma , whereas LIM-3 68.8". Ž1.0 Ma is sampled only 7.8 m above LIM-4 68.0".0.6 Ma . The age concordance between the ground-
mass LIM-3 and LIM-4 is extraordinary since theyŽhave relatively low K O abundances 0.32–0.562
.wt.% . Finally, groundmass samples from an alkalic
Ž .basalt JEN-1; 103.4"0.5 Ma and an ankaramiteŽ .JEN-4; 103.5"0.6 Ma from the same dredge siteat Jennings guyot are indistinguishable in age.
Our results suggest that we obtained reliable agesŽin 75% of all groundmass experiments ns32, in-
cluding data from Koppers, 1998; Koppers et al.,.1998 . In all cases, these age plateaus include at least
three incremental heating steps containing between30% and 70% of the total amount of 39Ar releasedKŽ .Fig. 3 . The lower limit for the total amount of39Ar released in these age plateaus is somewhat lessK
than the minimum criterion suggested for reliable
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158150
Table 3
ICP-AES Argon Argon mobility Discordance Inventories39 37 40 37 39Lab code K O CaO K O CaO Ar Ar KrCa TF Ar Ar Ar K O2 2 k ca atm ca K 2
Calculated argon mobility, inventories and discordances for the groundmass separates. The K O and CaO abundances determined from the240Arr39Ar analyses are calculated using an internal plagioclase standard measured together with the groundmass analyses; the difference of
Ž 40 39 . 39 37the K O and CaO abundances determined by ICP-AES before Arr Ar analysis then defines the total Ar and Ar mobility. LT2 K Ca
mobility of 39Ar has been calculated based on the inverse isochron diagrams assuming 39Ar recoil loss only. The KrCa discordancesK K
depict the difference in KrCa determined by the 40Arr39Ar and ICP-AES analyses expressed as a percentage of the ICP-AES analysis,Ž .whereas the total fusion TF discordances depict the difference in total fusion and plateau ages expressed as a percentage of the plateau age.
The K O and CaO inventories are calculated using the bulk composition of the groundmass samples and electron microprobe analyses of2Ž .clinopyroxene and plagioclase Table 2 . All data are listed as percentages, except for the K O and CaO abundances which are given in2
wt.%.
40 39 ŽArr Ar analyses on volcanic samples Appendix.A . Nonetheless, the observed age concordances of
the groundmass, phenocryst and duplicate samplesindicate that the groundmass ages are reliable — atleast, to a first order. When examining the ageplateaus closer, some minor alteration and recoileffects can be identified, as follows from the broadrange in observed MSWD values calculated for the
Ž .age plateaus 0.6–25.9; Table 1 . In some cases, thisis caused by an incomplete exponential decreasefrom the LT increments into a less than perfectly
Ž .horizontal age plateau e.g., LIM-3, NEK-1 . In othercases, the high MSWD is caused by random scatter
Žaround the average age of the plateau e.g., ITA-1,.ITA-2, JEN-4, MAL-5, NEK-2, VLI-4 . Although
we demonstrated that the plateau ages of groundmasssamples are geologically meaningful, the calculatedhigh MSWD values still reflect the disturbance in theage plateaus due to minor alteration and recoil.
5. Discussion
The 40Arr39Ar analyses of well-crystallizedgroundmass samples are characterized by high ap-
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 151
parent ages and increased KrCa ratios for the LTincrements, and low apparent ages and decreasedKrCa ratios for the HT increments. These character-istics cannot be explained by the effects of one
Ž .single process alteration or recoil acting on theargon system. The groundmass samples are polymin-eralic and may contain a variety of submicroscopic
Ž .phases, including devitrified basaltic glass. Differ-ences in the diffusive properties and K–Ca invento-ries for these phases may introduce another process:the preferential degassing of individual phases each
Žretaining distinct argon signatures cf. McDougall.and Harrison, 1988 . The disturbed age spectra of the
groundmass analyses, therefore, can only be under-stood as a complex function of alteration, recoil andthe preferential degassing of the groundmass mineralassemblage.
In order to formulate criteria for the successful40Arr39Ar analysis of altered groundmass samples,we need to know how these processes disturb theargon age spectra and how to minimize their effects.
Ž .To this extent this discussion considers 1 the mini-Ž . 40mization of alteration effects, 2 the amount of Ar,
39Ar and 37Ar loss and redistribution caused by alter-Ž .ation and recoil, and 3 the relation between ground-
mass K–Ca geochemistry and 40Arr39Ar data.
5.1. Minimization of alteration effects
Secondary minerals in the groundmass that have afirst order impact on the results of 40Arr39Ar analy-ses are produced by replacement of groundmassolivine with clay minerals, by devitrification of glassin the interstitial spaces, and by deposition of alter-
Žation products such as clay minerals, zeolites, phos-.phates and carbonate in veins, cracks and vesicles.
The alteration of plagioclase and clinopyroxene isvolumetrically less important. In this study ofseamount basalts, the first order problems of alter-ation were avoided through selection of groundmasssamples that are more than 98% crystalline, followedby acid-leaching. By selecting well-crystallized sam-ples, problems due to the presence of devitrified
Žbasaltic glass were minor cf. Fleck et al., 1977;McDougall and Harrison, 1988; Nauert and Gans,
.1994 . In basalts that originally contained large pro-portions of glass, the KrAr age may be lowered by
Ž .as much as 50% e.g., Pringle, 1993 . By acid-leach-
ing, secondary minerals filling in vesicles, cracksŽ .and veins, and replacing primary minerals olivine
were preferentially removed. This effect of acid-leaching is, for example, demonstrated in the lower-ing of the atmospheric component to 2–30% of the
Žtotal amount of argon released Table 3; cf. Davis et.al., 1989; Pringle, 1993 .
The remaining secondary phases after acid-leach-Ž .ing clays <10 mm appear to be located in inter-
stices and on the surface of groundmass minerals.However, sample irradiation and bakeout prior to40Arr39Ar analysis may induce an additional loss ofargon from these fine-grained clay minerals owing to
Žrecoil and diffusion loss Baksi, 1974; Lo et al.,. Ž .1994 . For example, Lo et al. 1994 modeled the
effect of a 2008C bakeout on the constituents ofaltered groundmass samples. Their results show thatduring a 24-h bakeout about 80% of the argon gas
Žwas lost from philipsite their representative alter-.ation mineral due to diffusion, but only 10% was
lost from plagioclase and negligible amounts fromclinopyroxene and fresh glass. In our study, thecombined effect of a 12- to 14-h irradiation and a
Ž .12- to 24-h bakeout 2008C is evident from thecalculated 39Ar and 37A losses up to 55% andK Ca
Ž30% in Table 3. Alteration minerals e.g., clays and.zeolites are expected to preferentially outgas com-
pared to primary mineral phases, given their highlydiffusive properties and small crystalrgrain sizesŽ .Walker and McDougall, 1982; Pringle, 1993 . Thisis confirmed by the observed increased 38Ar re-Cl
lease, high apparent ages, high proportion of atmo-Ž .spheric argon -80% and high KrCa ratios in the
LT increments, which are typical of alteration prod-Ž .ucts Table 2; Figs. 2 and 4 . An extended LT
Žheating schedule -6008C or 0.05–0.50 W laser.power is required to effectively remove the remain-
der of the discordant gas fraction before plagioclase,clinopyroxene and the interstitial spaces start de-
Ž .gassing cf. Lo et al., 1994 . When applying suchheating schedules we observed an exponential de-crease of apparent ages paralleled by a decrease inKrCa ratios and an increase in the proportion of
40 U Žradiogenic Ar to values greater than 95% Figs. 2.and 4 . Thus, with each temperature increment the
argon fractions included smaller components derivedfrom alteration minerals and larger components de-rived from primary groundmass phases. This discor-
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158152
dant part of the groundmass experiment is typicallycompleted after degassing 10–30% of the total 39Ar
Ž .inventory Fig. 2 . The effects of alteration on the40Arr39Ar age plateaus are, therefore, minimized bycombination of careful sample preparation, irradia-tion, an overnight bakeout and an effective LT heat-ing schedule. Equally important is the fact that bythis method the transition between alteration derivedargon and the less disturbed argon derived from themagmatic phases is well quantified.
5.2. 40Ar, 39Ar and 37Ar loss and redistribution
The loss and redistribution effects of 40Ar, 39Arand 37Ar on 40Arr39Ar age spectra of fine-grained,altered whole rock samples generally have been at-
Žtributed to alteration and recoil e.g., Turner andCadogan, 1974; Huneke and Smith, 1976; Dalrympleand Clague, 1976; Dalrymple et al., 1977; Fleck et
.al., 1977; Roddick, 1978; Pringle, 1993 . Inverse36Arr40Ar vs. 39Arr40Ar isochron diagrams are veryuseful in studying these effects on groundmass sam-
Ž .ples Fig. 5 since such diagrams do not express thepreferential degassing of individual groundmass
Ž37phases Ar is not included on the axes; cf. KelleyCa.et al., 1986 . In addition, groundmass samples do not
contain non-equilibrated phenocrysts, xenocrysts orxenoliths that may introduce excess or inherited ar-
Žgon signatures Dalrymple and Moore, 1968; Mc-
Fig. 5. Schematic isochron diagram explaining LT and HT devia-tions for the groundmass 40Arr39Ar analyses. Note the oppositeeffects of LT–HT 39Ar recoil; at LT, the high potassium sites in,K
e.g., alteration materials loose 39Ar due to recoil, which thenK
may be internally redistributed into the low potassium sites thattypically outgas at HT. This diagram is drawn not to scale.
Dougall and Harrison, 1988; Singer and Pringle,.1996; Singer et al., 1998 .
In Fig. 6, four representative groundmass analyseshave been plotted together with the isochrons of theircomagmatic plagioclase. The groundmass samples
Ž .display: 1 a parallel fit of the age plateau incre-Ž .ments with the plagioclase isochrons, 2 offsets of
the LT increments to values below the plagioclaseŽ .isochrons, and 3 offsets of the HT increments to
values above the plagioclase isochrons. The close fitof the groundmass age plateaus to the plagioclaseisochrons indicates that the age plateaus are a prod-uct of binary mixing between an atmospheric com-
Ž40 36 .ponent Arr Ars295.5 and a radiogenic compo-nent representing the crystallization age of theseamount basalts. It is generally not possible toperform meaningful isochron calculations forgroundmass samples, because the radiogenic compo-nent in the age plateau increments is very high andso these points show only a small range in 39Arr40Arand 36Arr40Ar. The LT offsets can be explained byrecoil loss of 39Ar lowering the 39Arr40Ar at con-K
36 40 Ž .stant Arr Ar Fig. 5 . The maximum calculated39 Ž .Ar losses for the LT increments 0.2–11.7% are,K
however, small compared to the bulk 39Ar lossKŽ . 39Table 3 ; this may suggest that most Ar loss isK
associated with the degassing of the alteration miner-als during irradiation and bakeout prior to the40Arr39Ar analyses. The HT offsets can be explained
Ž .by the effects of three processes: 1 the recoil lossof 37A indirectly causing higher 36Arr40Ar andCa39Arr40Ar ratios as a result of advancing interference
Ž . 39corrections, 2 the addition of Ar resulting inK
higher 39Arr40Ar ratios at constant 36Arr40Ar follow-ing an internal 39Ar redistribution after recoil fromK
Ž . 40 Uthe LT sites, and 3 the loss of Ar resulting inhigher 39Arr40Ar and 36Arr40Ar ratios due to the
Ž .alteration of the primary phases Fig. 5 . The twolatter processes are most evident because they ex-plain the higher 39Arr40Ar ratios falling far to the
Ž .right of the plagioclase isochrons Fig. 6 ; the firstprocess explains the bulk 37A loss as listed inCa
Table 3. Each process will, nevertheless, result inlow apparent ages compared to the crystallizationages defined by the plateau ages. To what extentplagioclase and clinopyroxene are affected by HTrecoil is difficult to establish. These calcium-bearingphases together account for more than 80% of the
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 153
Fig. 6. Isochron diagrams for groundmass 40Arr39Ar analyses. The isochrons calculated from comagmatic, phenocrystic plagioclase analysesŽ . 36 40are plotted for comparison their ages are listed in Table 1 . The plagioclase isochrons are forced to intercept the Arr Ar axis at the
Ž .atmospheric value of 295.5 since no excess argon is detected for these samples see text . Symbols and labels as in Fig. 4. Note the parallelfit of the groundmass age plateau increments with the plagioclase isochrons.
Ž .calcium inventory Table 3 and typically degas inŽ .the HT increments 50–90%; Table 3 . The effect of
37A recoil from plagioclase and clinopyroxene,Ca
thus, is expected to be more significant for the HTŽ .increments cf. Pringle, 1993 limiting the undis-
turbed plateau segment to the mid-range tempera-tures. It is interesting to note that the observed
decrease in the proportion of 40ArU for the HTŽ .increments Fig. 2: upper panels appears related to
the degassing of a trapped atmospheric argon com-ponent from plagioclase and clinopyroxene. This isalso shown in Fig. 6 where the HT increments areoffset towards the atmospheric air intercept in theinverse isochron diagrams. This means that the loss
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158154
of 40ArU due to alteration or the internal redistribu-tion of 39Ar must have been significantly contribut-K
ing to the lowering the apparent ages for the HTincrements.
5.3. Groundmass geochemistry as related to ground-mass dating
Discordances in the 40Arr39Ar spectra of ground-mass samples thus may be best explained by prefer-ential degassing and recoil of alteration phases forthe LT increments, and of plagioclase and clinopy-roxene for the HT increments. What if these pro-cesses are more detectable for potassium-poor sam-ples than for more alkalic samples? For example, itfollows that the bulk KrCa ratios measured by the40Arr39Ar analyses are almost invariably lower thanthe KrCa ratios measured before irradiation. Thisindicates that during sample irradiation and bakeout,but before 40Arr39Ar analysis, the groundmass sam-ples loose 39Ar preferentially over 37Ar . The dif-K Ca
Ž .ference i.e., discordance in KrCa ratios may,therefore, be used as a measure for 39Ar recoil inK
the groundmass alteration minerals. As a conse-quence, concordance in the bulk KrCa ratios indi-cates an insignificant contribution of alteration to thegroundmass analyses. Another measure is provided
Žby the discordance between total fusion age i.e.,.recombined age and the plateau age. Too low or too
high total fusion ages are respectively explained by40 U Žthe loss of Ar due to submarine alteration Fleck
.et al., 1977; Roddick, 1978 or by the recoil loss of39Ar . Concordance in the total fusion and plateauK
ages indicates that the main potassium reservoir isnot significantly influenced by alteration or recoil, orthat despite the effects of recoil, the groundmass
Žremained essentially a closed system for argon i.e.,.internal recoil redistribution . In Fig. 7A–B, the
observed discordances in total fusion age and KrCaare plotted vs. bulk KrCa ratio. Based on thisrepresentation of geochemical and 40Arr39Ar data it
Ž .is evident 1 that with increasing KrCa ratios theeffects of alteration and recoil become less important
Ž .in the groundmass analyses, and 2 that for KrCaratios lower than 0.15–0.20 such effects will signifi-cantly increase. Measuring bulk KrCa ratios, thus, isof great value during sample selection for 40Arr39Ar
Ž .groundmass analyses Appendix A . This does not
Fig. 7. Discordances vs. bulk KrCa ratios determined by ICP-AES.Ž . 40 39A Discordance for Arr Ar total fusion age expressed as apercentage of the plateau age. Error bars are calculated on the95% confidence level including a 0.2–0.3% standard deviation inthe J value. The plateau ages are charted as an average 95%
Ž .confidence envelope of 0"1% solid lines . Except for fourconcordant groundmass analyses, an increased discordance of
Ž ."20% is prominent at KrCa-0.15. B Discordance for bulkKrCa values determined by 40Arr39Ar analyses expressed as apercentage of the ICP-AES analyses. Primary KrCa ratios are
Žplotted as an arbitrary confidence envelope of 0"10% solid.lines . The deviations show an increased discordance for KrCa
Ž .ratios -0.2. C Discordance for KrCa values of the age plateausexpressed as a percentage of the ICP-AES analyses. The primaryKrCa ratios are charted as an arbitrary confidence envelope of
Ž .0"10% solid lines . Note that for less evolved groundmassesŽ .KrCa-0.3 , the age plateaus are characterized by KrCa valuesup to nine times the measured ratios for bulk groundmasses.
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 155
necessarily mean that successful 40Arr39Ar analyseson groundmass samples are impossible for un-
Ž .evolved seamount basalts cf. LIM-3, LIM-4 . How-ever, for such potassium-poor rocks the groundmassshould either be much less affected by alteration or itshould be characterized by interstitial spaces signifi-
Ž .cantly enriched in potassium up to 360%; Fig. 7Ccompared to bulk abundances. The LIM-3 and LIM-4groundmass samples may be enriched in the intersti-tial spaces for potassium due to the presence ofŽ . Ž .normative nepheline cf. Christie et al., 1995 .
6. Summary and conclusions
Groundmass incremental heating experiments re-sult in outgassing patterns that are different frombulk rock and phenocryst outgassing patterns, andtheir plateau ages do not consistently meet com-
Ž .monly accepted quality criteria Appendix A set forage determinations on volcanic samples. This may be
Ž .reflected in two sources of uncertainty: 1 thegroundmasses studied were partially subject to theloss or redistribution of 37A , 39Ar , and 40ArU dueCa K
Ž .to alteration and recoil, and 2 isochron calculationsfor the plateau segments were precluded due to highradiogenic argon components. Nonetheless, ground-mass samples have several advantages that makethem useful materials for dating. They show limitedgrain size variations and an enrichment in potassiumŽ .Appendix A . These advantages are apparent in anefficient removal of LT alteration effects and thehigh degree of coherence in the ages of ground-masses and comagmatic phenocrysts from individualsampling sites.
By performing )15 incremental heating steps ongroundmass samples the resolution of both discor-dant and primary radiogenic argon fractions wasimproved significantly compared to previous workŽe.g., Ozima et al., 1977; Seidemann, 1988; Lo et al.,
.1994; Pringle, 1993; Pringle and Duncan, 1995a,b .For more than 75% of the groundmass samplesŽ .ns32 these experiments resulted in age plateausincluding between 30% and 70% of the total amountof 39Ar released. We have demonstrated that inK
order to produce reliable age information for ground-
mass samples that their age spectra should includemore than three concordant incremental heating stepsrepresenting at least 30% the total amount of 39ArK
released. Total fusion and plateau ages that are con-cordant at the 95% confidence level and bulk KrCaratios )0.2 for the groundmass samples may beused as indicators for successful groundmass analy-ses as well.
In this study, we could show that acid-leachedgroundmass separates of submarine basalts provide asuccessful means of dating rocks that may otherwiseyield much more scattered results. While this studyadds very little to the geochronology of basalts withdatable phenocryst phases, it substantially expandsthe range of datable submarine rocks into aphyric
Ž .and olivine and clinopyroxene basalts. This expan-sion has been critical to our success in dating manyseamounts in the Western Pacific, in particular, when
Ždating their shield building lavas Koppers, 1998;.Koppers et al., 1998 . Dating shield building lavas is
important for constraining age progressions alongvolcanic lineaments, since they record the relativelyshort-lived arrival of the main pulse of volcanic
Ž .material at a particular hotspot location. The moreeasily datable alkaline caps are formed up to 10 Ma
Ž .later e.g., Clague et al., 1989 , and therefore, maynot give the most accurate information on the timingof shield building volcanism.
Acknowledgements
Malcolm Pringle is thanked for sharing his MSWDcalculation code for age plateaus and for many dis-cussions on the strategies for acid-leaching alteredseamount basalts. Gareth Davies, Ian McDougall,Warren Sharp and Bob Duncan are thanked for theirconstructive review comments improving both theEnglish and the structure of this paper. John Kistcarried out the ICP-AES analyses, and Wim Lusten-houwer is thanked for assistance with electron mi-croprobe analyses. Contribution 98.07.02 of theNetherlands Research School of Sedimentary Geol-
Ž .ogy NSG . Financial support by the NetherlandsŽFoundation of Earth Sciences Research GOA-NWO:
.750.60.005 and the National Science Foundation
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158156
Ž .NSF-OCE: 91-02183, and 98-11163 . This paper ispart of the PhD thesis of AAPK.
Appendix A. Criteria for interpreting 40Arrrrrr39Arexperiments
Internal testing of 40Arr39Ar ages on low potas-sium and altered volcanic rocks is necessary becausethe most fundamental assumption in geochronology— closed system behavior — may have been vio-lated. Alteration and recoil in these type of basaltsmight have caused an open system behavior for
Žargon Davis et al., 1989; Pringle, 1993; Pringle and. 40 39Duncan, 1995a,b . To test for reliable Arr Ar
Ž .crystallization ages, Fleck et al. 1977 , LanphereŽ . Ž .and Dalrymple 1978 and Pringle 1993 utilized the
following quality criteria for interpreting 40Arr39ArŽ .incremental heating experiments: 1 HT plateaus in
the age spectra should include more than three con-cordant incremental heating steps representing at least
39 Ž .50% of the total amount of Ar released, 2K
isochron and plateau ages should be concordant atŽ . 40 36the 95% confidence level, 3 the Arr Ar inter-
cepts on the isochron diagrams should be concordantwith the atmospheric value of 295.5 at the 95%
Ž .confidence level, and 4 the MSWD values shouldbe small compared to statistical F-distributions de-pending on the number of steps included in the agecalculations.
These criteria are generally well suited for inter-preting 40Arr39Ar incremental heating experimentson mineral separates or young whole rock samplesŽLo et al., 1994; Sharp et al., 1996; Singer and
.Pringle, 1996 . However, these criteria are to someextent arbitrary and based on convention; we demon-strated in this paper that reliable 40Arr39Ar plateauages might be extracted from less than ‘‘ideal’’groundmass experiments. When interpreting suchgroundmass experiments some of these criteria are
Ž .not usable, because: 1 useful isochron analyses areprecluded due to the high radiogenic component in
Ž .these samples, 2 geologically meaningful ageplateaus typically include between 30% and 70% of
39 Ž .the total amount of Ar released, and 3 MSWDK
values for age plateaus may be higher than expectedfrom analytical scatter alone and, thus, need to be
Žattributed to geological scatter Kullerud, 1991;
.Wendt and Carl, 1991; Kalsbeek, 1992; Wendt, 1992caused by alteration and recoil.
Appendix B. Groundmass geochemistry
Increasing modal plagioclase in the groundmasspositively correlates to an increase of K O in the2
Ž .bulk groundmass 0.18–3.09 wt.% and groundmassŽ .plagioclase 0.26–0.70 wt.% . It also correlates with
the appearance of K-feldspar as a groundmass phaseŽ .Table 2 . These observations confirm earlier studiesreporting that K O is enriched in basaltic ground-2
masses when compared to abundances of both wholeŽrock samples and phenocrystic minerals Mankinen.and Dalrymple, 1972; Lo et al., 1994 . For this
reason, groundmass samples typically have increasedradiogenic argon components with respect to the
Žalteration components cf. Pringle, 1993; Lo et al.,.1994; Nauert and Gans, 1994 . For example, the
Ž .K O abundance for groundmass VLI-4 1.47 wt.%2
is significantly higher than for its whole rock analy-Ž .ses acid-leached: 1.29 wt.%, unleached: 1.39 wt.%
Ž .or its phenocrystic plagioclase 0.19 wt.% .
References
Baksi, A.K., 1974. Isotopic fractionation of a loosely held atmo-spheric Argon component in the Picture Gorge basalts. EarthPlanet. Sci. Lett. 21, 431–438.
Brooks, C., Hart, S.R., Wendt, I., 1972. Realistic use of two-errorregression treatments as applied to rubidium–strontium data.Rev. Geophys. Space Phys. 10, 551–577.
Ž .Decker, R.W. Eds. , The eastern Pacific Ocean and Hawaii,the geology of North America. Geol. Soc. of America, Boul-der, CO, pp. 187–287.
Dalrymple, G.B., Clague, D.A., 1976. Age of the Hawaiian–Em-peror bend. Earth Planet. Sci. Lett. 31, 313–329.
Dalrymple, G.B., Dufflield, W.A., 1988. High precision 40Arr39Ardating of Oligocene from Mogollon–Datil volcanic field usinga continuous laser system. Geophys. Res. Lett. 15, 463–466.
Dalrymple, G.B., Lanphere, M.A., 1969. Potassium–Argon Dat-ing. Freeman, San Francisco, 258 pp.
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158 157
Dalrymple, G.B., Moore, J.G., 1968. Argon-40: excess in subma-rine pillow basalts from Kilauea volcano, Hawaii. Science161, 1132–1135.
Dalrymple, G.B., Clague, D.A., Lanphere, M.A., 1977. Revisedage for Midway volcano, Hawaiian volcanic chain. EarthPlanet. Sci. Lett. 37, 107–116.
Dalrymple, G.B., Alexander, E.C., Lanphere, M.A., Kraker, G.P.,1981. Irradiation of samples for 40Arr39Ar dating using theGeological Survey TRIGA reactor. U.S. Geological Survey,Professional Paper, 1176, 55 pp.
Davis, A.S., Pringle, M.S., Pickthorn, L.-B.G., Clague, D.A.,Schwab, W.C., 1989. Petrology and age of alkalic lava fromthe Ratak Chain of the Marshall Islands. J. Geophys. Res. 94,5757–5774.
Dodd, B., Anderson, T.V., 1994. Design, construction, commis-sioning and use of a new Cd-Lined InCore Irradiation TubeŽ .CLICIT for the Oregon State University RIGA reactor.Internal report OSU Reactor Center.
Hall, C.M., 1990. Calculation of expected thermal gradients inlaser 40Arr39Ar step heating experiments. EOS 71, 653.
Hanes, J.A., 1991. K–Ar and 40Arr39Ar geochronology: methodsŽ .and applications. In: Heaman, L., Ludden, J.N. Eds. , Short
Course Handbook on Applications of Radiogenic Isotope Sys-tems to problems in Geology. Mineral Association of Canada,Toronto, pp. 27–57.
Huneke, J.C., 1976. Diffusion artifacts in dating by stepwisethermal release of rare gases. Earth Planet. Sci. Lett. 28,407–417.
Huneke, J.C., Smith, S.P., 1976. The realities of recoil: 39Ar recoilout of small grains and anomalous age patterns in 40Arr39Ardating. Proc. Lunar Sci. Conf. 7, 1987–2008.
Ijlst, L., 1973. A laboratory overflow-centrifuge for heavy liquidmineral separation. Am. Miner. 58, 1088–1093.
Johnson, H.P., Carlson, R.L., 1992. Variation of sea floor depthwith age: a test of models based on drilling results. Geophys.Res. Lett. 19, 1971–1974.
Kalsbeek, F., 1992. The statistical distribution of the mean squaredweighted deviation — Comment: isochrons, errorchrons, andthe use of MSWD-values. Chem. Geol. 94, 241–242.
Kaneoka, I., 1972. The effect of hydration on the KrAr ages ofvolcanic rocks. Earth Planet. Sci. Lett. 14, 216–220.
Kelley, S., Turner, G., Butterfield, A.W., Shepherd, T.J., 1986.The source and significance of argon isotopes in fluid inclu-sions from areas of mineralization. Earth Planet. Sci. Lett. 79,303–318.
Koppers, A.A.P., 1998. 40Arr39Ar geochronology and isotopegeochemistry of the West Pacific seamount province: implica-tions for absolute Pacific plate motion and the motion ofhotspots. PhD Thesis, Vrije Universiteit Amsterdam, TheNetherlands, 263 pp.
Koppers, A.A.P., Staudigel, H., Christie, D.M., Dieu, J.J., Pringle,M.S., 1995. Sr–Nd–Pb isotope geochemistry of Leg 144 WestPacific Guyots: implications for the geochemical evolution ofthe ‘‘SOPITA’’ mantle anomaly. In: Haggerty, J.A., Premoli
Ž .Silva, I., Rack, F., McNutt, M.K. Eds. , Proceedings of theOcean Drilling Program, Scientific Results. Ocean DrillingProgram, College Station, TX, 144, pp. 535–545.
Koppers, A.A.P., Staudigel, H., Wijbrans, J.R., Pringle, M.S.,1998. The Magellan seamount trail: implications for Creta-ceous hotspot volcanism and absolute Pacific plate motion.Earth Planet. Sci. Lett. 163, 53–68.
Kullerud, L., 1991. On the calculation of isochrons. Chem. Geol.87, 115–124.
Lanphere, M.A., Dalrymple, G.B., 1978. The use of 40Arr39Ardata in evaluation of disturbed K–Ar systems. U.S. GeologicalSurvey Open-file Report, 78-701, pp. 241-243.
Larson, R.L., Erba, E., Nakanishi, M., Bergersen, D.D., Lincoln,J.M., 1995. Stratigraphic, vertical subsidence, and paleolati-tude histories of Leg 144 guyots. In: Haggerty, J.A., Premoli
Ž .Silva, I., Rack, F., McNutt, M.K. Eds. , Proceedings of theOcean Drilling Program, Scientific Results. Ocean DrillingProgram, College Station, TX, 144, pp. 915–933.
Lo, C.-H., Onstott, T.C., Chen, C.-H., Lee, T., 1994. An assess-ment of 40Arr39Ar dating for the whole rock volcanic samplesfrom the Luzon Arc near Taiwan. Chem. Geol. 114, 157–178.
Mankinen, E.A., Dalrymple, G.B., 1972. Electron microprobeevaluation of terrestrial basalts for whole rock K–Ar dating.Earth Planet. Sci. Lett. 17, 89–94.
McDougall, I., Harrison, T.M., 1988. Geochronology and Ther-mochronology by the 40Arr39Ar Method. Oxford Univ. Press,212 pp.
Nagihara, S., Lister, C.R.B., Sclater, J.G., 1996. Reheating of oldlithosphere: deductions from observations. Earth Planet. Sci.Lett. 139, 91–104.
Nauert, J.L., Gans, P.B., 1994. 40Arr39Ar geochronology of wholerock basalts. US Geol. Surv. Circ. 1107, 231.
Ozima, M., Saito, K., 1973. 40Arr39Ar stepwise degassing experi-ments on some submarine rocks. Earth Planet. Sci. Lett. 20,77–87.
Ozima, M., Saito, K., Honda, M., Aramaki, S., 1976. Sea waterweathering effect on K–Ar age of submarine basalts. Geochim.Cosmochim. Acta 41, 453–461.
Ozima, M., Honda, M., Saito, K., 1977. 40Arr39Ar ages of guyotsin the western Pacific and discussion of their evolution. EarthPlanet. Sci. Lett. 51, 475–485.
Premoli Silva, I., Haggerty, J., ack, F., et al., 1993. ProceedingsODP, Initial Reports. Ocean Drilling Program, College Sta-tion, TX, 144.
Pringle, M.S., 1993. Age progressive volcanism in the musiciansseamounts: a test of the hot spot hypothesis for the lateCretaceous Pacific. In: M.S. Pringle, W.W. Sager, W.V. Sliter,
Pringle, M.S., Duncan, R.A., 1995. Radiometric ages of basementlavas recovered at Lo-En, Wodejebato, MIT, and Taikuyo-
( )A.A.P. Koppers et al.rChemical Geology 166 2000 139–158158
Daisan Guyots, ODP Leg 144, Northwestern Pacific Ocean.In: Haggerty, J.A., Premoli Silva, I., Rack, F., McNutt, M.K.Ž .Eds. , Proceedings of the Ocean Drilling Program, ScientificResults. Ocean Drilling Program, College Station, TX, 143,pp. 547–557.
Roddick, J.C., 1978. The application of isochron diagrams in40Arr39Ar dating: a discussion. Earth Planet. Sci. Lett. 41,233–244.
Schlanger, S.O., Garcia, M.O., Keating, B.H., Naughton, J.J.,Sager, W.W., Haggerty, J.A., Philpotts, J.A., Duncan, R.A.,1984. Geology and geochronology of the line islands. J.Geophys. Res. 89, 11261–11272.
Sebai, A., Feraud, G., Bertrand, H., Hanes, J., 1991. 40Arr39Ardating and geochemistry of tholeiitic magmatism related to theearly opening of the Central Atlantic rift. Earth Planet. Sci.Lett. 104, 455–472.
Seidemann, D.E., 1977. Effects of submarine alteration on K–Ardating of deep-sea igneous rocks. Geol. Soc. Am. Bull. 88,1660–1666.
Seidemann, D.E., 1988. The hydrothermal addition of excess 40Arto the lava flows from the Early Jurassic in the Hartford basinŽ .northeastern U.S.A. : implications for the time scale. Chem.Geol. 72, 37–45.
Sharp, W.D., Turrin, B.D., Renne, P.R., Lanphere, M.A., 1996.The 40Arr39Ar and KrAr dating of lavas from the Hilo 1-kmcore hole, Hawaii Scientific Drilling Project. J. Geophys. Res.101, 11607–11616.
Singer, B.S., Pringle, M.S., 1996. Age and duration of theMatuyama–Brunhes geomagnetic polarity reversal from40Arr39Ar incremental heating analyses of lavas. Earth Planet.Sci. Lett. 139, 47–61.
Singer, B.S., Wijbrans, J.R., Nelson, S.T., Pringle, M.S., Feely,T.C., Dungan, M.A., 1998. Inherited argon in a Pleistoceneandesite lava: 40Arr39Ar incremental-heating and laser-fusionanalyses of plagioclase. Geology 26, 427–430.
Staudigel, H., Park, K.-H., Pringle, M.S., Rubenstone, J.L., Smith,W.H.F., Zindler, A., 1991. The longevity of the south pacificisotope and thermal anomaly. Earth Planet. Sci. Lett. 102,24–44.
Taylor, J.R., 1982. An Introduction to Error Analysis. UniversityScience Books, Mill Valley, CA, 270 pp.
Turner, G., Cadogan, P.H., 1974. Possible effects of 39Ar recoil in40Arr39Ar dating of lunar samples. Proc. Lunar Sci. Conf. 5,1601–1615.
Walker, D.A., McDougall, I., 1982. 40Arr39Ar and K–Ar datingof altered glassy volcanic rocks: the Dabi volcanics. Geochim.Cosmochim. Acta 46, 2181–2190.
Wendt, I., 1992. The statistical distribution of the mean squaredweighted deviation — Reply. Chem. Geol. 94, 242–243.
Wendt, I., Carl, C., 1991. The statistical distribution of the meansquared weighted deviation. Chem. Geol. 86, 275–285.
Wijbrans, J.R., Pringle, M.S., Koppers, A.A.P., Scheveers, R.,1995. Argon geochronology of small samples using theVulkaan argon laserprobe. Proc. Koninklijke NederlandseAkademie van Wetenschappen 98, 185–218.
Winterer, E.L., Natland, J.H., Van Waasbergen, R.J., Duncan,R.A., McNutt, M.K., Wolfe, C.J., Premoli Silva, I., Sager,W.W., Sliter, W.V., 1993. Cretaceous guyots in the NorthwestPacific: an overview of their geology and geophysics. In:
Ž .Pringle, M.S., Sager, W.W., Sliter, W.V., Stein, S. Ed. , TheMesozoic Pacific: Geology, Tectonics, and Volcanism. Am.Geophys. Un., Washington, DC, 77, pp. 307–334.
York, D., 1969. Least squares fitting of a straight line withcorrelated errors. Earth Planet. Sci. Lett. 5, 320–324.