Trace element and Sr-Nd isotopic constraints on the...
Transcript of Trace element and Sr-Nd isotopic constraints on the...
Geochemical Journal, Vol. 20, pp.173 to 189, 1986
Trace element and Sr-Nd isotopic constraints on the compositions of lithospheric primary sources of
Serra Geral continental flood basalts, southern Brazil
S. S. HUGHES', R. A. SCHMITT' and Y. L. WANG" and G. J. WASSERBURG2
Departments of Chemistry and Geology and the Radiation Center, Oregon State University, Corvallis,Oregon 973311, and The Lunatic Asylum, Division of Geological and Planetary Sciences,
California Institute of Technology, Pasadena, California 911252 U.S.A.
(Received December 12, 1985; Accepted April 23, 1986)
Multi-element abundances in twenty-four samples from the Serra Geral continental flood basalt system
(Parana Basin) were determined by INAA to assess the trace element signature of primary magmatric sources. Isotopic Sr and Nd data were obtained on a 24-sample composite and three individual samples of the suite to constrain the influences of older enriched material. Twenty basaltic samples have enriched LILE commensurate with most continental tholeiites and , display fractionated REE patterns similar to those in the Columbia River Group and other flood basalt provinces. The mafic units exhibit low-moderate Ti, variable V, Cr and Ni and relatively uniform Co and Sc . abundances. Four intermediate to silicic units exhibit higher overall incompatible element abundances, strongly fractionated patterns and depletions of compatible elements consistent with derivation by fractionation of basaltic parents. Isotopic data for two basalts and the composite analysis indicate 87Sr/86Sr = 0.7090 to 0.7105 and eNd = -6.6 to -6.9 requiring magma genesis in an evolved LILE-enriched upper mantle or lower crust. A silicic sample has 87Sr/86Sr = 0.7219 and eNd = -9.2 reflecting a stronger crustal influence. Chemical comparisons within the basaltic members allow the delineation of ten least-evolved compositions having a regionally characteristic trace element pattern and two additional samples (of one flow) representing a second, less fractionated pattern. Trace element models for both types predict magma segregation from lower lithospheric sources having relatively uniform enrichments of incompatible elements via regional metasomatism. The most viable scenarios for basalt magma production require either enriched mantle partial-melt liquids, some of which comingle with crustal components, or partial melting of ultramafic, LILE-enriched lower crust. Either process requires a zone of primary magma extraction in a region where initially uniform source enrichments act independently of subsequent contamination, probably in the crust/mantle transition zone.
INTRODUCTION
Source regions of continental flood basalts
(CFBs) and the nature of subcontinental lithosphere are enigmatic in light of equivocal trace
element and isotopic constraints. Trace element
studies of late Precambrian (Keweenawan and
Coppermine River) and Mesozoic-Tertiary
(Karroo, Parana, Siberian Platform, Deccan and Columbia River) CFB provinces (e.g., BVSP,
1981; Balashov and Nesterenko, 1966; Dupuy and Dostal, 1984; Mahoney et al., 1982; Nelson,
1980; Ruegg, 1976; Smith and Schmitt, 1981;
Hughes et al., 1983; Mantovani et al., 1985) have shown that (1) Cr and Ni contents are
often too low for the equilibration of large
volumes of CFB magma with unaltered mantle and (2) abundances of incompatible elements,
e.g. K, Rb, Ba, REE (rare earth elements) and
Th, are variable and high compared to oceanic tholeiite chemistries. As an example, La in
basaltic members of CFB systems ranges from
less than 5 to over 50ppm whereas normal mid
* Permanent address: Chengdu Geological College, Chengdu, Sichuan, P.R.C. ~.
173
174 S. S. Hughes et alt
ocean ridge basalts (Sun et al., 1979) contain
typically 1 to 7 ppm La. The higher incompati
ble element signatures in CFB magmas cannot be
attributed entirely. to fractional crystallization of picritic parents (e.g., Cox. 1980), but reflect
LILE (large ion lithophile element) enrichments
owing to source metasomatism, subsequent
crustal contamination or a combination of
these processes. Generally, CFB lavas display wide ranges in 144Nd/143Nd and 87Sr/86Sr ratios (e .g., DePaolo and Wasserburg, 1979; BVSP, 1981; Carlson et al., 1981 b; Mahoney et al., 1981; Mantovani et al., 1985) which suggest crustal interaction
(Carlson et al., 1981a). Combined Nd, Sr and O isotopic data for Columbia River Basalts
(Nelson, 1983; Carlson et al., 1981 b) and Nd-Sr isotopic evidence from Deccan flows (Mahoney et al., 1982) argue for significant crustal interaction in most CFB provinces. But the cluster of
data around eNd = 0 and (87Sr/8fiSr), = 0.705 for the volume-weighted major proportion of CFB lavas (DePaolo, 1983) supports a primordial mantle source (DePaolo and Wasserburg, 1976,
1979). While CFB lavas portray gross similarities in trace element and isotopic characteristics, the influences of mantle versus crustal
processes are probably unique for an individual province or subsystem. The tectonic association of Serra Geral (SG) lavas of southern Brazil
(Parana Basin) with the Jurassic-Cretaceous breakup of Gondwanaland (Fodor et al., 1985) affords an opportunity to evaluate possible source compositions and crust development
during the transition from continental to oceanic magmatism. Trace element data by Ruegg (1976),
although not including REE analyses, indicate somewhat higher incompatible element abun
dances in Parana rocks relative to most CFB
lavas. Cl chondrite normalized (N) trace element patterns of SG-CFB compositions includ
ing REE were first reported by Hughes et al.
(1983). The patterns indicated wide ranges in LILE, but relatively uniform basaltic composi
tions correlated well with better-known CFB systems. On the basis of Ba, Sr, REE and Sc
signatures, Hughes et al. (1983) proposed that
the Parana mafic magmas represent a series of
partial melts from an enriched plagioclase
peridotite source. ;; Silicic compositions, and their fractionated patterns were further at
tributed to an extensive crustal history involving
fractional crystallization rather than contamina
tion although the possibility of crustal influence
was not totally dismissed. Recently reported trace element abundances in the LPT (low P and Ti) Parana CFB samples by Mantovani et al. (1985) are consistent with worldwide CFB systems and agree with the
patterns reported earlier by Hughes et al. (1983). Mantovani et al., (1985) also report wide ranges in 87Sr/86Sr ratios (0.7074 0.7163) for the LPT samples which are commensurate with ranges in other CFB provinces. Models calculated by Mantovani et al. (198 5) show that the magmas parental to the LPT series originated in a variably LILE enriched and heterogeneous
garnet-free mantle source. They attributed the LPT series to plagioclase dominated fractionation and progressive contamination, thereby confirming the suggestion by Hughes et al.
(1983) that plagioclase was an important residual phase. Additional isotopic data available for SG
CFBs include the average initial 87Sr/86Sr ratio of 0.7057 for six samples by Compston et al.
(1968), and Sr-Nd data on two samples (DePaolo and Wasserburg, 1979) that yield similar 87Sr/86Sr ratios and eNd values of -1.5 and -4.2. These few data are consistent with SG-CFB sources in a LILE-enriched upper mantle and preclude derivation from a depleted mantle unless enrichment occurred subsequent to magma genesis. The 24 Serra Geral CFB samples documented by Fodor et al. (1985) were provided
for trace element and isotopic assessment of
the SG-CFB source region. These samples,
whose trace element patterns were originally
shown in the preliminary report by Hughes et al. (1983), represent four arbitrary geographic
sectors (our designations) in a 300km X 300km
region near Port Alegre (see Fig. 1 in Fodor et
Trace element and Sr-Nd isotopic constraints 175
al., 1985) in the southern portion of this CFB
province. Multi-element data obtained by instrumental neutron activation analysis (INAA) are reported in detail along with Nd-Sr isotopic analyses on selected samples in order to demonstrate the trace element characteristics of the lower lithospheric source at the time of magma segregation and to evaluate further the possibility of crustal involvement in primary magma
genesis. We address these problems in light of (1) the comparison of (low Ti) SG-CFB chemistries with better-known CFB systems and the likely prevalence of regional uniformities which can be related to primary sources of most CFB magmas, (2) initial Nd/Sm ratio in the source as determined by trace element modeling and Nd isotopic systematics, and (3) the feasibility of mathematical models which satisfy trace elements and Nd-Sr isotopic constraints. While
previous arguments for crustal contamination are accepted for silicic CFB compositions, we
question the need for contamination and source heterogeneity in the generation of mafic magmas having uniform trace element signatures.
ANALYTICAL PROCEDURES
Approximately 10 grams of each sample
was processed through a 99.9% pure alumina
plate jaw crusher and rotary mill to insure homogeneous samples and similar counting
geometries. Multi-element abundances were determined using a sequential INAA procedure
similar to that described by Laul and Schmitt
(1973). Aliquants (^'0.5g) of samples and rock standards BCR-1, BHVO-1, GSP-1 and
PCC-l, plus primary standards, were weighed
and irradiated for short-lived radionuclide analyses for 3 minutes at 25 kW in the Oregon
State University TRIGA reactor via a pneumatic
transfer system for a total neutron fluence of 4 X 1013. Each sample and standard was counted
for 400 seconds with a horizontal Ge(Li) gamma-ray detector coupled to a multichannel
spectral analyzer to obtain Ti, Al, Mg, Ca and V
abundances. After 4-8 hours decay, subsequent countings of 1-2 k seconds on similar equipment
yielded analyses for Na, Mn and Dy. Data for long-lived radionuclides were obtained by activa
tion of ^'200mg samples (accurately weighed)
and rock standards (plus a primary REE stan
dard) for 5 hours at 1 MW for a neutron fluence
of 5 X 1016. After 3-6 days decay time, samples
were initially counted 1-2k seconds for K and U,
then 8-10k seconds to determine abundances
of Na, La, Sm, Yb and Lu. Continued sequen
tial countings using longer decay and count
times (several weeks and 20-40k seconds, respec
tively) enabled data acquisition for Fe, Sc, V,
Cr, Co, Ni, Se, Rb, Sr, Cs, Ba, Ce, Nd, Eu, Tb, Zr, Hf, Ta and Th abundances. Isotopic data for
Nd and Sr were determined at the California
Institute of Technology following the proce
dures outlined by DePaolo and Wasserburg
(1976) and Papanastassiou et al. (1977).
RESULTS
Chemical compositions determined by se
quential INAA are listed in Table I according to the four geographic sectors. Isotopic and
elemental Nd and Sr data are presented in Table
2. Group 1 has been further subdivided on the
basis of relative trace element abundances
(primarily for clarity in presentation of data); although G-648 does not fit readily into any
subgroup, it is assigned to group I c. The low
moderate TiO2 abundances, together with the
LILE abundances, suggest that all of these
samples belong to the LPT series of Mantovani
et al. (1985). The twenty basaltic samples show
moderate ranges in LILE abundances exem
plified by K2O = 0.6-1.7 wt.%, Rb = 14-69 ppm, La = 9-27 ppm, Th = 2-6 ppm and U = 0.3-2.0
ppm while much greater ranges and higher LILE concentrations are evident in the four inter
mediate to rhyolitic samples with K2O = 2.25.4 wt.%, Rb = 73-210 ppm, La = 25-46 ppm,
Th = 8-19ppm and U = 0.4-4.7 ppm. The elements V, Cr and Ni are widely variable in
basaltic units with V = 250-540ppm, Cr =
6-250ppm and Ni = 10-160ppm compatible with wide ranges in other Parana rocks (Ruegg,
1976); whereas less variation is shown by Sc =
176 S. S. Hughes et al.
Q Q
z
0) N
..O
E 0)
0)
CV
0) 00
4. 0
0 a
0 U .V.
G) .i. U
4)
n>r 0. 0
0
O. 0 O
CS
7
O
0L 0
0
u
.0
M 0) 00 00 ~ 0 'IT ~o N Oh M 00 ti
'tt N so N O 0
N
C. M
N M M 5O O N O O
•O .0 00 .
2 N oD V; ~t 00 ~O N .-. tI) .•i m 00 M N to 0 Ow
kn C7% tn ~O h h A N In V O
00 O N z r. N Ln O a
NM ON l~ O O O~ ct 0 00 ON
O
d• N M 00 N O O
M
N O0 00 0~ N O h M O .
N .--1 .-•i M tt 5O O 0~ 0~ 0 V) as %0 a, N.,O
.O V)M ~O M et OO 01 O O 47 N N (li r+N-~t N %0 NN0
r. 00 O N V) .--s et '
rr .. V)CD V)as N -s0
,.4 N 00 50 .-i O OO V) 50 0 O% .-q
> ti 00 00 50 4 N 6 6 w -4
rn00 O n m 05
M M N m (75 O .•. NV) 0%N-.O
~o r 00 ,tr) .~ 00 N 50 N 00 00 ¢t N
.4 M M M 00 N N O
0 -4rn
.-. ~.-. V) a, N OO
N , O M 05 M O o0 N t- OO O\ .-i
ai -+V)N h (;5 N OO
.. 0-7 [~ 00 m O 5D O 00 05 .-. n O
~ ti 50 m N M M O
0000 N V) 00 v w) 50 t .r m 05 O) N O O
O M M .-y l
N 05 .-i OO 0~ V) O .•i
a .-. M N 00 N 0 0
N O t CN ON O'1: CD 50t-M5o..
N O O
0 a, N N t M N M 5O qt 00 N
~/ .. M 7 4) 00 N O O
0 O N N M O oo O 50 .•. M
V) N 50 05 N rr 0
a, 0O 5000 M M V et .-. r.
O .. h O O M V) 0% et M .•.
N 0 O O O O\Nt-O
v m
N 00 (7%06 50 " N v
M
N M O W) N 0 0 0 0 0 v) 0 N 0 .r N N M
N .-. .. ti .-~ V)
N
v) . et ~O t~ 0 0 0 0 0) 0 .-. N M .-S V) N
-. N N \D
N 0 50N 00 M0 d•-O
en
w) O O v) O O M V) V) 0 lD T
N N .r
0 0 0 0 0 00 .. 5D N
N M
b 0 0 0 0 0 v t -
r4
00
M000505n 00000 M V) M M .•i 00 00 M N
N N N N
505000
.V ry
et 00000
N .. .-a
N O O M 0 0 M 50 N 7 O 00
00 CD M0.-+0 0 N N .,
)n 00000 00 0 O O
N M
N O M O O O \O M %D V)
N N
N 050 000 00 0000-.0 M N 50 sr .-..-r N 0%
rr .--I .-1 N (n
05 0 N .-. 0 0 M 0 N 0 V) N
O0 O v) 40 O O en 50wI~r n0
M c.
.-50.t N O V) 05N0-0.(7`
M
00 0 0 0 0 0 M d• V) y
V) 4n 4n o O 0 0 N N
N N
50 O N O O O 5o N M 00
N N
r -. 00 N 00 O 00 p O O N .-~ .•. V)
oq oo0.0400 ON 000 M 5o w) I t 05 00 N 05 r
m N N
OO 00 O O O O 0 oo O M O M M M 0 in .~ t- ti O) 05
M . .-. N
0 0 N 0 O 0 0 5o w) 10. 0)M
o .t kn C) CD 0 N 0 N et
0 0 N 05 O v) 00 05 0t 50 50
M
N
M t` O O O
N
O 50 O .-~ O 00 N et etr4 N
50 O h 0 0 0 M M M a5
H .--1
O 00 w) C~ N [~ M 50 t~ M 50 N O N 0 f0 .-y M .r 05 0 .•. O M v) O O O 05 0 0 0
50 )n V 50 (75 0t N 50
N 50 V) 00 .0 V) n 00 50 N
--4 M 0 00 N O 0
O 05 l~ l~ t~ t~ n 50 T t~ a% O .-+ En .
a O
O 0 0 en st O 0 O cn 0 •-. 0 V N 0. M N N to 0 O
0 H .-1 M M
0 M O O M O, d M a
-01 .-.
05 l~ N N O " 5D 0 00 50 05 0-) N 00N.-.O O0 0I O0
m
O 0 1c pN00
Q.
0
E G
a >UUzN
o. ON 00 tl 50 .-. M
.-1 M
00 005-+0 l~ 8 ON 50
N M
E a
to C4 to U aG
O Net IqInNM OOH N0--00 N0
ON N N V) b %O 00 d: 0 00 et 0, 0,5000N O a 0 N
h It O N V) O~ O oq V)
40N -s00mO M N v
m N M t~ QD V) V) a~ N t1 )n 0 00 0~ kn O
00 .-. l~ V) N h 0\ .-. M N N0 w) %.O M0
N h ti \O ON N v1 00 N M
N N e ti O M N 0 N M N
ON 00 en O OR O SO O~ N M
V) N fn m .-r O to N 0 -~ 0 N
0~ li 't 't 0-00 40 5040m 0 N 50 N
O~ 00N 0050 r 0d 05 et ..O ON O N
40 r.
0 0050-sa,N et c 50m.-5Om~ O N M N
w) wl~ M M M
MO.-.NN-5000-0 N N
C4 %n OO O 50 a5 50 ' 7 t cn 0 w) 0 5o cn 0
N
ti O 00 00 50 M 00 0\ M It
V) et r. .-5 O v) M 0 . M N
V ~--~ N N 00 "T 0 N v v ~--~ O Ln N O r M
05 0 N N N O V)
M 50 ct
00 05 M N. V) 0-)400-)-.. 040N0
M N
N M M 50 V; o0 O 05 7 05 N a5 "t •. (D V) N 0
C-4 m 0, mO%00v)d 00 (D 00 M .-. O 4 N O
N
O N .S v) M 5o O tN 50
50M N0-).-•~.-. 50 O .. v ..
N D\ N R \O M OO O5 DD et .-i r. V) M O M N O .-1 M N
CD r4 cn N en V, .t M
05 M M r. 0 et N ON
50 OO 4) O 50 M It
N 7 N' to m 0
00 l~ 50 Nr %0 1.0
.. 50 N 50 N .-•s 00 O N 0 N
N .+ V) N t-N V) 5O
t~ N N N N O 0 O N It M
v 4) '0 E
0 00 00 M
O M O .-4 O M
SD 01 C M V7
M O N O 00
N O~ sO O 0 M N M N ti
N N 00 t~ 0 t` N 00 00M
N [~ O 00 sO
O 0 ti N ti .. ..
00 et ~O l~ .-.
O M O M -i 00
O O N N O
Om O M.-.
N O O "t
0 0 -. 00 O
N
00 M 0550(
O 0 V) r. V) N
oo 00 t M l1
O N O M O 05
0 V) N, ON
0 0 0 r n, 50 N
h N 00 50
0 0 0 ti
05 v) e} 4) N O M O M V)
M 5Q N V) O M O M O V)
050 MM N
tn
O M O cn 0
O 5D 05 N N
O M O V r. .-t
et N ON O N O N O
r-1
0 4 0 0. N 7 .r
oo V) N 50
O fn o M 0
n V) M 7
O N 0 N 0 00
N OD t` t O 0 m 0 v
N cn O o0 N
O et ti v) .-. 05 N
V) N 00 05 N
O O 0 ..
a 0.
N x F H ~
W
) N+1
dt
U
O
L~ _ Nm
M 'ti
+1 i"
d
U Oy O '0 N
d t3 C
Q
y -0 d
ti
00
tsO ' 4) 0.
~s O
Q +I O
O b
y 41 O,tea, U c~
N oO 3
y n °OON
h
d tl y 0 '
LXI
a ~ Q 0s m u
Trace element and Sr-Nd isotopic constraints 177
Table 2. Isotopic data for Serra Geral samples
Sample Sr (ppm) 87Sr/86Sr Nd (ppm) eNd Sm (ppm) 147Sm/144Nd
RS-65
RS-71
EV-01
Composite-24
161
214
242
207
0.72187±3
0.70899
0.71010 ±3
0.71052 ±5
35.31
22.97
17.76
24.10
-9.22 ±0.37
-6.63 ±0.33
-6.61 ±0.47
-6 .9
7.40
5.32
3.66
5.8
0.127
0.140
0.125
32-44ppm and Co = 39-49ppm. The uniformity
of Sc and Co abundances in basaltic units con
trasts with strong depletions of these elements in more silicious rocks as shown for Co vs. La in
Fig. 1.
Relative abundances of REE plotted in Fig. 2 as La/Sm vs. La/Yb, vary systematically and show enrichment of light REE (LREE) over C l chondritic (minus volatiles) abundances
(Anders and Ebihara, 1982). Also plotted in Fig. 2 are typical LREE-depleted mid-ocean ridge basalts (MORB, Sun et al., 1979) which complement the higher LREE in the continental units yielding a curve which passes through the C 1 value and may possibly pass through the origin. The curve reflects the tendency for La/ Yb ratios to increase faster relative to La/Sm ratios indicating less fractionation of LREE with overall increases in incompatible elements. Figure 2 also portrays minor REE differences between some geographic locales, especially
group I b, although the less-evolved units of most sectors overlap in the ranges La/Sm = 3.34.5 and La/Yb = 4.7-8.0.
Rare earth element abundances in SG sam
ples and other basaltic types, plotted as La vs. Yb and Sm vs. Yb in Fig. 3 show the demarca
tion which is evident between CFB and MORB
(Sun et al., 1979); however, alkalic basaltic compositions from Hawaii (Chen and Frey,
1983; Frey and Clague, 1983), the Rio Grande
Rise (Thompson et al., 1983) suggest chemical overlaps of tholeiitic CFB with alkalic oceanic
basalts of aseismic ridges, especially at higher
total REE contents. At lower REE abundances
SG and other CFB lavas maintain a more de
finitive segregation from the basalts of aseismic
5
E Q
0 0
4
50
40
30
20
10
0
• RS-76
RS ° * eRS-69 •1k0 e G-402 ° a r
• U G-648
GEOGRAPHICGROUP
o fa
• fb a is
o II • III
0 IV
•RS-65
MRS-30
RS-50
3
2
I
0jI
La/Sm
•
Serra Geral
ev ~,
i
i
P CI Chondrites
MORBasalts
e
0 ®'
0 0 0
La/Yb
0 10 20 30 40 50
La (ppm)
Fig. 1. Co vs. La in all Serra Geral CFB units. RS-30, RS-50 and RS-65, silicic members of the suite, have strongly depleted Se as well as Co abundances.
0 2 4 6 8 !0
Fig. 2. Ratios of La/Sm vs. La/Yb in Serra Geral com
positions relative to C1 chondrites (Anders and Ebihara, 1982). The curve reflects a complementary relation between LREE-depleted MORB compositions (Sun et al., 1979) and LREE-enriched Serra Geral primary magmas. Symbols are the same as in Fig. 1.
178 S. S. Hughes et al.
ridges in the southern Atlantic.
C 1 chondrite-normalized Sr, Ba, REE, Sc,
Hf, Ta and Th patterns of all SG-CFB samples
are shown in Fig. 4 with the REE segments
spaced according to ionic radii. Depletions in
Sr, Eu and Sc, produce negative anomalies in
evolved members of groups lc and 3 which are
coupled to higher total REE, Hf, Ta and Th.
All units display low Sr and many samples,
except in group 4, have REE patterns with small
negative Eu anomalies. Although minor aberra
tions are evident in some SG samples, the
basaltic trace element patterns within each
group (except l b) are essentially parallel and overall pattern variations between groups are
minimal.
50
40
E 30
d -j 20
I0
0
I0
8
CL 6
E
4
2
0
4
• A
++
+_
W
/
w +
A ~~ AAWO 0
O ~ o Cf
w
C.
2J
0
0
Ilk' •,•
0
W
00
00
A Alkalic Hawaiian Basalt
R
(Chen and Frey, 1983; Frey and Clague, 1983)
Rio Grande Rise Basalt
(average of 7, Thompson et al., 1983)
W Walvis Ridge Basalt(Humphris and Thompson, 1983; Liu and Schmitt, 1984)
/_
/
/
+
A
Aw0/ O
A / 00,~ /~ a' Q°O
4+ 4 A + •k`
w /
+ / 108
/
W /W y/y +o O +p
R
O
Z 0
A
A
•
00
W•
O
CO
0
Serra Geral (this study)Col. River Group (Carlson,
1984; BVSP, 1981)0 Siberian Platform (Smith and
Schmitt, 1981; 8alashov and Nesterenko, 1966)
+ Kewe enawan Ref. Suite
(B A MORE
VSP, 1981) (Sun et al., 1979)
0 1 2 3
Yb
4 5
(ppm)
6
Fig. 3. La vs. Yb and Sm vs. Yb for a variety of continental and oceanic basalts. The lower solid lines represent the well-defined boundary between MORB and continental tholeiites, whereas the upper dashed lines represent a less tenable boundary of alkalic compositions.
Figure
minimum
SG basalts
5 illustrates
trace element
and the same
Trace element and Sr-Nd isotopic constraints 179
the maximum and from two divisions in the Columbia River (CR)
patterns representing group: (1) Saddle Mountains formation and for reference samples (2) CR members stratigraphically below the
U) W
0 Z 0
U
U NN W -J CL
Q U)
100
50
20
100
50
20
100
50
20
100
50
20
10
5
GROUP
* (a ' Jb
• IC • G648
•II * IV
•III
10
5
2
10
5
2
10
5
2
10
5
2
Sr Ba La Ce Nd SmEuGd Tb Dy Yb Lu Sc Hf To Th
Fig. 4. Sr, Ba, REE, Se, Hf and Th plotted (REE ionic radii) relative to volatile free (minus C and H20) C1 chondritic abundances (Anders and Ebihara, 1982). Major group divisions are geographic sectors reported by Fodor et al. (1985). Scale on right is for Se only.
180 S. S. Hughes et al.
Saddle Mountains section (BVSP, 1981). The
REE segments in the CR patterns are nearly identical with the CR data in Carlson (1984)
which exhibit increases in total REE and in the ratio of light to heavy REE roughly with
stratigraphic position. Isotopic data for the
CR system (McDougall, 1976; Nelson, 1980;
Carlson, 1981 a) indicate an increasing radiogenic component with higher stratigraphic position
and argue in favor of crustal contributions to
the Saddle Mountain lavas. However, the low
sequences in the CR group have less radiogenic
compositions and are more likely derived with
minimal amounts of crustal interaction. The
patterns in Fig. 5 portray strong geochemical similarities of SG basalts in this study (although
not stratigraphically controlled) to the basalts in the lower CR sequence, an argument in favor
of similar petrogenesis.
Isotopic data (Table 2) for two basaltic samples (EV-01 and RS-71), a rhyodacite (RS65) and a composite analysis of all 24 units indicate high 87Sr/S6Sr ratios (0.709 to 0.722) relative to mantle xenolith minerals (e.g., Menzies and Murthy, 1980a,b; Jagoutz et al., 1979; Stosch et al., 1980). Negative eNd values
(-6.6 to -9.2) are within the mantle mineral realm and argue for a LREE enriched (NdN > SmN) source. The fact that Nd-Sr data for the two basaltic samples are similar to the composite analysis suggests that RS-65 and other silicic derivatives are petrologically distinct from the main sequence of SG-CFB liquids. The displacement of silicic units is also evident in Figs. 1 and 4.
0 W N J
Q: O z
W t a: 0 Z 0
U
U
103
102
10
+
0
!I.
Maximum and Minimum
-mo o.
o--o
Serra
CR8
CR8
Geral
Saddle
below
Patterns
Basalts
Mfns. Fm.
SM Fm.
\41 I
/
./
4• 4
L ~I/
DISCUSSION
Trace Element Constraints
The bulk and trace element geochemistries of mafic SG units, as well as REE (Figs. 3 and
5), are virtually indistinguishable from those of most CFB systems despite minor variations
within individual groups. This likeness argues in favor of a unifying geochemical process which
can be generalized to sources of CFB systems on
a worldwide basis although the continental
ft. 5. f patterns (from Fig. 4) in Serra Geral basalts, Columbia River
, rains River data from B VSP, 1981.
Sr Ba La Ce Nd SmEuGdTbDy YbLu Sc Hf Th
Upper and lower boundaries o trace element
basalts stratigraphically below the Saddle Moun
Fm. and CRB Saddle Mountains Fm. Columbia
tectonic settings are not always uniform. The
timely breakup of Gondwanaland certainly
allowed for changes in depth of mantle sources
beneath eruptive zones, but the rifting itself
probably had little impact on the regional geochemistry of pre-established magma sources in
the subcontinental lithosphere.
Descriptions of CFB primary sources is complicated by the requirement of either
picritic parents (Krishnamurthy and Cox, 1977; Cox, 1980) which fractionate to magmas having lower Mg/Fe ratios or melting of an evolved
(Fe-enriched) source. Also, any viable scenario for a CFB origin requires large volumes of chemically uniform magma, high rates of extrusion and, therefore, a regional "magmafer". If the least-evolved units of a CFB system display a characteristic trace-element pattern over a wide area, that pattern should reflect a relatively homogeneous stage of primary magma evolution prior to divergences produced by assimilative mixing or fractional crystallization. Assuming that fractionation of a picritic parent will not appreciably alter incompatible element ratios, the average least-evolved trace-element
pattern is primary and was determined before or during initial magma extraction.
Trace element and Sr-Nd isotopic constraints 181
80
60
40
20
10
8
6
4
2
0.3
0.2
0.1
o RS-70 (Rb/K) x /0
RS-7I
p RS-69RS-65 RS-30 RS-50i
p01 O MR-03 ------- *-- p--------------------~a 01
_ .~-0 GI 9
-~ * G-648 M-03Bulk Earth
EV-01 ~'
RS-73 O 0'-02.--~
p0
t7 pRS-2Q,' • •
G-402 La/ Yb
O CI Chondrites
Yb/Sc RS-50
RS-30
00 RS-65
G-641 ~.- J'~ M-03
p,. -O--- - --bMR-03
I 1 I I I I I2 4 6 8 10
Th (ppm)12 14 16 18
Fig. 6. Rb/K, La/Yb and Yb/Sc vs. Th in Serra Geral lavas. Bulk earth composition from DePaolo, 1983; CI chondritic values from Anders and Ebihara (1982). Symbols as in Fig. 1.
182 S. S. Hughes et al.
The nearly parallel configurations in Fig. 4 suggest a regional trace element uniformity of basaltic units, although G-648 has higher REE abundances and subgroup l b exhibits lower overall abundances (except Se) and lower La/ Yb ratios. The members of group 1 b, especially RS-20 and G-402, exhibit the lowest LILE abundances; however, siderophile and major elements fall within the ranges shown for all basaltic units and do not support a lesser degree of magma evolution. The observation that REE
patterns of lb members, which have lower total REE abundances, are not parallel to other
groups argues for a separate line of descent for subgroup l b and a source region with lower La/Yb ratios and total LILE abundances. Figure 6 illustrates the behaviors of Rb/K,
La/Yb and Yb/Sc ratios relative to the highly incompaticle element Th. The slight increase of Rb/K in evolved members portrays a systematic trend from the assumed bulk earth composition (DePaolo, 1983), but it is offset by G-648, M-03 and members of subgroup 1 a which follow an alternate line of evolution (discussed in greater depth by Fodor et al., 1985). Variations in La/Yb ratios with Th (Fig. 6) are less systematic, although mafic members cluster around La/Yb = 4-6. A positive trend from the
C 1 chondritic value passes through basaltic compositions and subtle differences in REE chemistry between groups are evident. The Yb/ Sc vs. Th data for most basaltic units plot along a line extending from the C 1 chondritic value through the most silicic members of the suite. Offsets of Yb/Sc ratios in M-03, MR-03 and RS65 suggest that diverging evolutionary patterns are initiated below Th = 5 ppm, the cut-off used herein for least-evolved magmas. The clustering in Fig. 6 of most basaltic compositions compared to the scattered points
for silicic and divergent units allows the delineation of samples which best represent the least
evolved trace element signature. A regionally
characteristic . trace-element pattern is established by averaging these basaltic compositions,
but excluding certain samples for the following
reasons: (1) G-648 (Fig. 4) has absolute REE
abundances roughly 50 percent higher than REE in other SG basaltic rocks; (2) subgroup la
(RS-69 and RS-71) has excessive Rb (Fig. 6) and high FeO, V and Th, suggesting a divergence in petrogenesis; and (3) subgroup l b has lower
total LILE abundances and lower La/Yb ratios
(Fig. 6), although bulk chemistry and siderophile trace elements do not substantiate a more primitive stage. The average of 10 SG basaltic compositions
considered to be least-evolved and regionally
represented is depicted in Fig. 7 as SGAV 10.
A second compositional average of G-402 and
RS-20 (SUB 1 b, Fig. 7) indicates either LILE
depletion from previous melt extraction or less
LILE enrichment in an alternate source. Our
petrogenetic models of the SG basaltic magmafer assume that (1) the SGAV 10 pattern is
inherent in magmas at the zone of basaltic
magma extraction, regardless of the magmafer
location (crust or mantle), and (2) the processes
that determined the geochemical character of the least-evolved magmas are uniformly opera
tive on a regional scale. Differentiation to more
silicic compositions, accompanied by an increase
in total incompatible elements and more frac
tionated patterns (Figs. 2-4), is dominated by
fractional crystallization of mafic parents
(Hughes et al., 1983; Fodor et al., 1985;
~,100 a)
50
0 t _U 20
U \ 10 Q)
5 0
2
Serra feral CFB Least-evolved Magmas
SGAV 10
SUB Ib -
(RS-20 a G-4021
Fig. p deviations,
, RS-73 1-02
patte less f
primary magmas.
Sr Ba Lo Ce Nd Sm Eu Gd Tb Dy Yb Lu Sc Hf To Th
7. SGA VIO: trace element attern with standard
r, representing least-evolved Serra Geral magis average of RS-16, RS-26, RS-37, RS-60, RS-68, M-01, M-02, and G-215. SUBl b: identical
rns of two samples from one flow which, although representative , reflects a second source or SG
Trace element and Sr-Nd isotopic constraints 183
Mantovani et al., 1985) with an allowance for
subordinate effects of crustal assimilation.
Similarities between evolved SG compositions
and post-Archean continental crust (McLennan et al., 1980) would probably cause the effects
of crustal contamination to be cryptic in trace
elements. However, the uniformity of frac
tionated compositions in basalts precludes the
random effects of secondary assimilation and
supports an enriched lower lithospheric source.
Isotopic Constraints
The isotopic data in Table 2 suggest crustal influence on the SG-CFB system in addition to magma equilibration in LILE-enriched regions of the lithosphere. Figure 8 illustrates eNd vs. 87Sr/ 86Sr data of this study plus those for two other
SG basalt samples of DePaolo and Wasserburg
(1979) obtained from a drill core near Santa Catarina (PAR, see Fig. 1, Fodor et al., 1985). The higher 87Sr/86Sr ratio and lower eNd of RS-65 places it near the "upper crust" realm
(DePaolo, 1981a), indicating that RS-65, and
probably other evolved samples contain uppercrust material. The less-evolved units, RS-71 and EV-01, and the composite SG analysis are consistent with a basaltic magmafer in either the lower crust or an enriched upper mantle although equilibration with mantle material is not supported by major elements (MgO C 10%) or transition metal abundances. By contrast, PAR 1 and 2 plot closer to the present day bulkearth composition supporting a less-evolved basaltic magma source. The isotopic differences between these sources are consistent with either variable amounts of older LILE-enriched material assimilated into the liquid or variations in the age of source-enrichment prior to magma extraction.
Assuming little or no difference in source ages, the SG-CFB source. is petrologically positioned between primary mantle and upper crust as shown by combined fractional crystallization and mixing curves in Fig. 7. The curves follow equations given by DePaolo (1981b) using a
primary magma composition of 87Sr/86Sr =
8
4
z o W
-4
-8
-12
Bulk Earth
® PM t PAR-1
o PAR 2 ~
30
Sr (m)
MAGMA 450
CRUST 50
87Sr/86Sr Nd (ppm)
20
45
eNd
II
PRIMARY "
20 RSl71EV-01
/ s
Composite
C /A
0.705
0.760
DSr = 0.96 ONd = 0.14
50A 50
100
-1
-13
-4,1
RS2:1 7o 1. wI'1 oo so
0* 1
1 I 200 I ! I
0.704 Q 708 0.712
87Sr/86Sr0.716 0.720
Fig. 8. eNd vs. 87Sr/86Sr for PAR-1 and PAR-2 (DePaolo and Wasserburg, 1979) and RS-71, EV-01, RS-65 and a 24-sample composite (this study). Curves represent combined effects of crustal assimilation and fractional crystallication using an assumed primary magma and the "E" crustal component of Carlson et al. (1981b). CA Values indicate cumulate to assimilate ratios and labelled points show the percentage of original mass added by assimilation.
184 S. S. Hughes et al.
0.705 and eNd = -1. The assimilated composition is assumed to be similar to the "E" com
ponent used by Carlson et al., (1981a) with 87Sr/86Sr = 0.760 and eNd = -13, but with an absolute Nd abundance (45 ppm) more com
patible with silicic SG rocks. The evolutionary lines represent specific ratios of cumulate lost to assimilate added (C:A) and the numbered
points mark the percentage of original mass added by assimilation. The position of RS-65 on the C:A = 2:1 curve implies an evolutionary trend marked by more than 80 percent crustal assimilation with twice that amount (of original mass) being removed as cumulate. Basaltic SGCFB units lie above the 2:1 line and simulate wsystem controlled by 30-50% assimilation of the the crustal component along with larger amounts of crystallization. However, 30-50% upper-crust addition to regionally uniform
basaltic magmas seems fortuitous unless a shallow homogenization process can be demonstrated. Alternatively, the positions of basaltic SG
units in Fig. 8 represent magma generation in an older enriched lithospheric source, possibly in the mantle/crust transition zone. The differences between regional Parana magmafers become those of location and age of LILE enrichment, assuming similar petrogenetic behavior and somewhat variable isotopic signatures. This discretion is supported by similar or lower Sm/Nd ratios in the two PAR samples
(Sm/Nd = 0.17 and 0.22, DePaolo and Wasserburg, 1979) compared to that of the SGAV 10 composition (Sm/Nd = 0.21). It is likely that source enrichments for PAR 1 (Fig. 8) occurred within a relatively short time before magma genesis.
TRACE ELEMENT MODELING
Models of SG-CFB source regions follow
three inherent premises. First, absolute trace
element abundances can only be approximated
due to ranges in partition coefficients for most
minerals (e.g., Fujimaki et al., 1984; Irving,
1978), whereas relative behaviors are more
accurately predicted. Second, uncertainties in the relative amounts of partial melting and fractional crystallization for large magmatic volumes force a combination of mineral/melt calculations into a single, presumably equilibrium, process. Residual minerals in source models may be either unmelted solids at the original point of extraction or subsequent
precipitates. Third, the most viable model should predict a flat chondrite-relative pattern for the most compatible elements, HREE, Sc and Hf which will remain relatively unfractionated during source enrichments or depletions. Assuming that the SG lavas are not derived from an initially depleted reservoir, enrichments determined by modeling represent sources prior to magma segregation. Predicted sources for SGAV 10 and SUB 1 b,
shown in Fig. 9 were calculated using the equi
librium non-modal melting equation of Shaw
(1970) to simulate the combined effects of anatexis and crystallization. Partition coef
ficients (D-values, Appendix) are those ap
plicable to tholeiitic basalt systems and D-values for Th are assumed zero. The strong depletions in Sr compared with minor negative Eu anom
alies support higher and lower relative D-values
for Sr and Eu, respectively, of plagioclase equi
librated at elevated fo 2 and lower temperatures
(Drake and Weill, 1975). Melting proportions correspond approximately with observed mass
modal proportions of minerals (excluding late
stage Fe-Ti oxides) in tholeiitic basalts although
the modeled sources depend more on residual
proportions. The preferred model for SGAV10 (Fig. 9)
assumes 10 percent melting of an initial plagioclase peridotite source and a residual mineralogy of of 60 opx 10 cpx 1 p129. The greater relative amounts of residual olivine and plagioclase may be related to subsequent crystallization of these
phases and do not necessarily indicate the initially solid proportions at greater depths. Enrichments in Sr, Ba, LREE and Th are consistent with melting of metasomatized ultramafic lithosphere having an initial (Sm/Nd)N ratio of 0.65. Lower degrees of melting, even for a less
Trace element and Sr-Nd isotopic constraints 185
reasonable source mineralogy required to minimize inflections in the REE-Sc-Hf pattern, require similar or lower (Sm/Nd)N ratios as shown for F = 5 percent melting.
The SUB 1 b composition yields a viable
source at 15 percent melting (Fig. 9) which is
similar to the SGAV 10 source in HREE, Sc and
HE Lower incompatible element abundances
100
10
QJ
N v
E
__
O
I00
v
10
I
F=10 %
0I
SOURCE 55 MELT 10
RESIDUUM 60
P1?x 10
to 10
CPX 5
40I
PI. 30 40 29
F=5%
SOURCE 78
MELT 10 RESIDUUM 81
3
25 2
1
25 0
\
I8 40 17
10
5
SGAV10
/
i
-J
I
I If
F=15%
01
SOURCE 77
MELT 20 RESIDUUM 87
5 20
3 20
0
P 1
15 40
11
F=10%
15
IO
SOURCE 83 4 2 11
MELT 20 20 20 40 RESIDUUM 90 2 0 8
SUB t b
, II - -------- ----------------------r i
-J
Sr 8a La Ce Nd Sm Eu Gd Tb Dy Yb Lu
Fig. 9. Trace element models using non-modal batch melting equations (Shaw, SGA VIO and SUB1 b Serra Geral magmas. D-values given in Appendix.
Sc
1970)
Hf
for
Th
primary sources of
186 S. S. Hughes et alt
indicate a less enriched or evolved source having a (Sm/Nd)N ratio of 0.85. Similar relative enrichments are portrayed by the modeled source at 10 percent melting. Models for both magma compositions from Fig. 7 indicate LREE-fractionated sources with (Sm/Nd)N ratios similar to the observed ratios. Essentially no fractionations of Ba and LREE are obtained due to their low bulk D-values in the residual mineralogy. It is likely that PAR-1 and PAR-2
(DePaolo and Wasserburg, 1979; Liu and Schmitt, 1984) compositions also yield source
(Sm/Nd)N ratios that are similar to their observed ratios. Model-dependent sources for PAR-1 and PAR-2 (not shown) should have minimum (Sm/Nd)N ratios of 0.52 and 0.67, respectively, which suggest slightly greater source enrichments despite the more primitive isotopic signatures.
PRIMARY MAGMA SOURCES
The apparent disparity between source composition and the observed eNd values argues in favor of source enrichments to produce the observed trace element patterns as well as source variability between subprovinces in the SG-CFB system. Added crustal components are operative in altering some isotopic signatures, but do not significantly alter the overall trace element
pattern. Slightly different trace element patterns in sources required by SGAV 10 and SUB 1 b derive from regional differences in the subcontinental mantle not unlike those which have been proposed for the Columbia River Basalt group (Carlson, 1984). Isotopic differences, apart from the pronounced effects in silicic compositions, may be imparted with variable ages of enriched lithospheric sources or variable proportions of crustal components added to isotopically uniform primary magma. In light of similar Sm/Nd source ratios required for most SG-CFB magmas, variations in source age appear to be the more viable hypothesis. Assimilation of a crustal component to alter isotopic ratios requires that the contaminant have a trace element signature close to that in
the primary melt.
At least two geologically reasonable scenarios are possible: (1) primary melts from enriched lithospheric mantle comingled and homogenized with underplated crustal com
ponents of variable ages, probably in stratigraphic succession; and (2) primary magmas were derived by partial melting of enriched lithosphere at variable time-stratigraphic horizons. Both hypotheses allow the existence of regional homogeneous primary magma subsystems and require the placement of CFB magmafers within the crust/mantle transition at levels permissible for regionally uniform partial melting. In either scenario trace element enrichments are obtained in the zone of primary magma production.
CONCLUSIONS
Basaltic low-moderate Ti members of the Serra Geral CFB system display characteristic trace element signatures that reflect a regional constancy of magmatic processes. Trace element models require at least two separate sources enriched in incompatible elements via metasomatic processes. Isotopic data indicate variable contributions of crust or evolved upper mantle in some basaltic magmas, a more substantial crustal component in silicic units, and support models of magma genesis in LILEenriched sources. The basaltic magmafer for a
given subprovince in the SG-CFB system may be obtained by either of two mechanisms: (1)
primary melts from enriched mantle comingle with assimilated lower crustal components of a
particular age or (2) partial melting of ultramafic LILE-enriched lower lithosphere occurs within a specific time-stratigraphic horizon. Similarity of these two scenarios requires the
placement of SG-CFB magmafers in the mantle/ crust transition zone allowing proportioned contributions from variable-evolved LILE-enriched material. The similarity of modeled SG-CFB source
chemistries, especially incompatible trace ele
ment ratios, with better known CFB systems
Trace element and Sr-Nd isotopic constraints 187
reflects the processes of primary source enrich
ment on a regional scale and not the random
effects of subsequent contamination by crustal
components. Evolution of the SG-CFB source region, despite the breakup of Gondwanaland,
represents a possibly uniform and continuous
global process of LILE enrichment in the continental lower lithosphere.
Acknowledgements-We thank A. Roisenberg for providing the suite of Serra Geral samples and T. V. Anderson, W. T. Carpenter and D. Pratt for assistance with neutron activations in the Oregon State University TRIGA reactor. Helpful comments and criticisms by R. V. Fodor greatly improved the manuscript. Financial support was provided by NASA grant NAG9-63 to R. A. Schmitt and grant NAG9-49 to G. J. Wasserburg.
REFERENCES
Anders, E. and Ebihara, M. (1982) Solar systemabundances of the elements. Geochim. Cosmochim.
Acta 46, 2363-2380.
Arth, J. G. (1976) Behavior of trace elements during magmatic processes a summary of theoretical models and their applications. J. Res. U. S. Geol. Surv. 4,41-47.Balashov, Y. A. and Nesterenko, G. V. (1966) Distri butions of the rare earths in the traps of the Siberian Platform. Geochem. Int. 3, 672-679.BVSP (Basaltic Volcanism Study Project, 1981) Basal
tic Volcanism on the Terrestrial Planets. Pergamon Press: New York. 1286p.Carlson, R. W. (1984) Isotopic constraints on Colum bia River flood basalt genesis and the nature of the subcontinental mantle. Geochim. Cosmochim. Acta
48, 2357-2372.Carlson, R. W., Lugmair, G. W. and McDougall, J. D.
(1981a) Crustal influence in the generation of continental flood basalts. Nature 289, 160-162.
Carlson, R. W., Lugmair, G. W. and McDougall, J. D.
(1981b) Columbia River volcanism: the question of mantle heterogeneity or crustal contamination. Geochim. Cosmochim. Acta 45, 2483-2499.Chen, C.-Y. and Frey, F. A. (1983) Origin of Hawaiian tholeiite and alkalic basalt. Nature 302, 785-789.
Compston, W., McDougall, I. and Heier, K. S. (1968) Geochemical comparison of the Mesozoic basaltic rocks of Antarctica, South Africa, South America and Tasmania. Geochim. Cosmochim. Acta 32, 129-149.
Cox, K. G. (1980) A model for flood basalt vol
canism. J. Petrol. 21, 629-650. DePaolo, D. J. (1981a) A neodymium and strontium
isotopic study of the Mesozoic calc-alkaline granitic
batholiths of the Sierra Nevada and Peninsular ranges,
California. J. Geophys. Res. 86, 10470-10488.
DePaolo, D. J. (1981b) Trace element and isotopic effects of combined wallrock assimilation and frac tional crystallization. Earth Planet. Sci. Lett. 53, 189-202.
DePaolo, D. J. (1983) Comment on "Columbia River volcanism: the question of mantle hetero geneity or crustal contamination" by R. W. Carlson,
G. W. Lugmair and J. D. McDougall. Geochim. Cosmochim. Acta 47, 841-844.DePaolo, D. J. and Wasserburg, G. J. (1976) In
ferences about magma sources and mantle structure from variations of 143Nd/144Nd. Geophys. Res.
Lett. 3, 743-746.DePaolo, D. J. and Wasserburg, G. J. (1979) Neo dymium isotopes in flood basalts from the Siberian
Platform and inferences about their mantle sources. Proc. Natl. Acad. Sci. U.S.A. 76,3056-3060.Drake, M. J. and Weill, D. F. (1975) The partition of Sr, Ba, Ca, Y, Eu2+, Eu3+ and other REE between
plagioclase feldspar and magmatic silicate liquid: an experimental study. Geochim. Cosmochim. Acta 39, 689-712.
Dupuy, C. and Dostal, J. (1984) Trace element geo chemistry of some continental tholeiites. Earth
Planet. Sci. Lett. 67, 61-69.Fodor, R. V., Corwin, C. and Roisenberg, A. (1985) Petrology of Serra Geral (Parana) continental flood basalts, southern Brazil: crustal contamination,
source material, and South Atlantic magmatism. Contrib. Mineral., Petrol. 91, 54-65.Frey, F. A. and Clague, D. A. (1983) Geochemistry of diverse basalt types from Loihi Seamount, Hawaii:
petrogenetic implications. Earth Planet. Sci. Lett. 66, 337-355.Fujimaki, H., Tatsumoto, M. and Aoki, K. (1984) Partition coefficients of Hf, Zr, and REE between
phenocrysts and groundmasses. Proc. Lunar Planet. Sci. Conf 14th, J. Geophys. Res. 89, B662-B672.Hughes, S. S., Wang, Y. L., Schmitt, R. A., Fodor, R. V., Corwin, C. and Roisenberg, A. (1983) Parana Basin lavas: chemical and petrographic charac teristics. Lunar Planet. Sci. 14, 337-338.
Irving, A. J. (1978) A review of experimental studiesof crystal/liquid trace element partitioning. Geochim. Cosmochim. Acta 42, 743-770.
Jagoutz, E., Palme, H., Baddenhausen, H., Blum, K., Cendales, M., Dreibus, G., Spettel, B., Lorenz, V. and
Wanke, H. (1979) The abundances of major, minor and trace elements in the earth's mantle as
derived from primitive ultramafic nodules. Proc.
188 S. S. Hughes et al.
Lunar Planet. Sci. Conf 10th 2031-2050. Krishnamurthy, P. and Cox, K. G. (1977) Picrite
basalts and related lavas from the Deccan Traps of
western India. Contrib. Mineral. Petrol. 62, 53-75.
Laul, J. C. and Schmitt, R. A. (1973) Chemical composition of Luna 20 rocks and soil and Apollo 16 soils. Geochim. Cosmochim. Acta 37, 927-942.
Lindstrom, D. J. (1976) Experimental study of the
partitioning of the transition metals between clino pyroxene and coexisting silicate liquids. Ph. D. Thesis, Univ. Oregon, 188p.Liu, Y.-G. and Schmitt, R. A. (1984) Chemical
profiles in sediment and basaltic samples from Deep Sea Drilling Project Leg 74, Hole 525A, Walvis Ridge.
Initial Rep. Deep Sea Drill. Proj. 74, 713-730.Mahoney, J., MacDougall, J. D., Lugmair, G. W., Murali,
A. V., Sankar, Das. M. and Copalon, K. (1982) Origin of the Deccan Trap flows at Mahabaleshwar
inferred from Nd and Sr isotopic and chemical evidence. Earth Planet. Sci. Lett. 60, 47-60.
Montovani, M. S. M., Marques, L. S., DeSousa, M. A., Civetta, L., Atalla, L. and Innocenti, F. (1985) Trace element and strontium isotope constraints on the origin and evolution of Parana continental flood basalts of Santa Catarina State (Southern Brazil)
J. Petrol. 26, 187-209. McDougall, I. (1976) Geochemistry and origin of
basalt of the Columbia River Group, Oregon and
Washington. Geol. Soc. Am. Bull. 87, 777-792.
McKay, G. A. and Weill, D. F. (1977) KREEP petro
genesis revisited. Proc. Lunar Sci. Conf 8th 2339 2355.
McLennan, S. M., Nance, W. B. and Taylor, S. R. (1980) Rare earth element-thorium correlations in sedi mentary rocks, and the composition of the continen
tal crust. Geochim. Cosmochim. Acta 44, 1833 1839.
Menzies, M. and Murthy, V. R. (1980a) Nd and Sr isotope geochemistry of hydrous mantle nodules and their host alkali basalts: Implications for localheterogeneities in metasomatically veined mantle.
Earth Planet. Sci. Lett. 46, 323-334.
Menzies, M. and Murthy, V. R. (1980b) Mantlemetasomatism as a precursor to the genesis of alkaline
magmas isotopic evidence. Am. J. Sci. Jackson Vol.
280A, 622-638.
Nelson, D. O. (1980) Strontium isotopic and trace element geochemistry of the Saddle Mountains and
Grande Ronde Basalts of the Columbia River Group. Ph. D. Thesis, Oregon State Univ., 224p.Nelson, D. O. (1983) Implications of oxygen-isotope data and trace-element modeling for a large-scale
mixing model for the Columbia River Basalt. Geol ogy 11, 248-251.
Papanastassiou, D. A., DePaolo, D. J. and Wasserburg, G. J. (1977) Rb-Sr and Sm-Nd chronology and
genealogy of mare basalts from the Sea of Tranquility. Proc. Lunar Sci Conf 8th 1639-1672.
Philpotts, J. A. and Schnetzler, C. C. (1970) Pheno cryst-matrix partition coefficients for K, Rb, Sr and Ba, with applications to anorthosite and basalt
genesis. Geochim. Cosmochim. Acta 34, 307-322. Ruegg, N. R. (1976) Characteristicas de distribuicao e
teor de elementos tracos dosados em rochas basalticas
de bacia do Parana. Naturalia 2, 23-45.
Shaw, D. M. (1970) Trace element fractionation during anatexis. Geochim. Cosmochim. Acta 34, 237-243.
Smith, M. R. and Schmitt, R. A. (1981) Petrogenesis of continental flood basalts from the Siberian Plat
form and a model of mantle evolution. Lunar Planet.
Sci 12, 1011-1013.
Stosch, H. G., Carlson, R. W. and Lugmair, G. W. (1980) Episodic mantle differentiation: Nd and Sr isotopic evidence. Earth Planet. Sci. Lett. 47, 263-271.
Sun, S.-S., Nesbitt, R. W. and Sharaskin, A. Y. (1979) Geochemical characteristics of mid-ocean ridge basalts. Earth Planet. Sci. Lett. 44, 119-138.
Thompson, G., Humphris, S. and Schilling, J. G. (1983) Petrology and geochemistry of basaltic rocks from
Rio Grande Rise, South Atlantic: Deep Sea Drilling Project Leg 72, Hole 516F. Initial Rep. Deep Sea
Dri.. Pro/. 72,457-466.
Trace element and Sr-Nd isotopic constraints 189
APPENDIX
Distribution coefficients used in trace element models
Olivine Orthopyroxene Clinopyroxene Plagioclase
Sr
Ba
La
Ce
Nd
Sm
Eu
Gd
Th
Dy
Yb
Lu
Se
Hf
Th
0.009
0.009
0.0056
0.0056
0.0058
0.0078
0.0086
0.010*
0.011
0.012
0.042
0.053
0.37
0.022
0.0
0.017
0.013
0.022
0.024
0.033
0.054
0.054
0.091
0.11*
0.15
0.34
0.42
1.4
0.051
0.0
0.12
0.026
0.105
0.125
0.287
0.477
0.562
0.595 0.61 *
0.622
0.601
0.560
3.8
0.121
0.0
2.7**
0.23
0.21
0.12
0.081
0.067
0.10**
0.063
0.059*
0.055
0.059*
0.060
0.044*
0.01
0.0
* Values which are interpolated or estimated from several sources. ** Estimated values for D
Sr and DEu in plagioclase at higher f02 values(see text).
Sources of data:
Sr, Ba: Philpotts and Schnetzler (1970), Arth (1976). REE, Hf: Fujimaki et al. (1984), Arth (1976). Sc: Lindstrom (1976), McKay and Weill (1977).