EARTH AND PLANETARY SCIENCE LETTERS 10 (1970) 7-11.

NORTH-HOLLANDPUBLISHINGCOMPANY

Rb, Sr, Y, Zr AND Nb IN SOME OCEAN

FLOOR BASALTIC ROCKS

J R. CANN
School of Envlronmental Sciences, Unwerszty of
East Anglta, Norwtch NOR 88C, England

Recewed 6 October 1970

Rewsed version recewed 20 October 1970

X-ray fluorescence measurements have been made of Rb, St, Y, Zr and Nb m 35 samples of ocean floor basaltic
rocks from four dafferent areas of the oceans Samples include fresh, weathered and metamorphosed basalts and
dolentes. Comparison Is made with K and TI determinations on the same samples Rb, K and, to a lesser extent, Sr
are affected by processes of ocean floor weathering and metamorplusm, but abundance of Y, Zr, Nb and Tl seem to
be unaffected even by severe secondary processes High correlations are found between Y and T1, and Zr and T1 for
both fresh basalts and the group of samples as a whole. These correlations define a narrow composmon space m
terms of these elements for ocean floor basaltic rocks even when they are severely altered It is possible that th~s
property might be used to identify altered ocean floor basalts and to distinguish them from volcamcs from other
sources

1. Introduction culated for different adopted values for the USGS

standards by multiplying by an appropriate factor.
Rb, Sr, Y, Zr and Nb have been measured by X-ray Rock types chosen for analysis cover a broad range
fluorescence techniques m 35 samples of basaltic of the chfferent varieties of basaltic rocks so far recov-
rocks from four different areas of ocean floor. ered from the ocean floor. There are aphyrlc, ohvme
The analyses were made using a Pluhps PW 1410 phync and plagloclase phync fresh basalts, with some
spectrometer and a peak-to-background ratio method fresh dolentes, and weathered basalts of different
The method was checked agaxnst results obtained on degrees of ocean floor weathering. Metamorphosed
some of the rocks using the method of Nornsh and basalts include spflltes, spdltlsed basalt glasses, urahte
Chappell [ 1]. Agreement between the two methods amptubohtes, and amplubohtes m wluch plagtoclase
was excellent. Full details of analytical techniques has suffered secondary replacement by muscovite.
used are avmlable on request. Precision is estimated to
vary from 50% at 2ppm to 10% at 10 ppm and 5% at
100 ppm Accuracy depends on the standards used Table 1
for cahbratlon, wbach were the USGS standard rocks Concordant values measured and adopted for this work for
USGS standard rocks.
W l , G2, GSP1, AGV1, and BCR1. Values quoted by
Flanagan [2] and Flelscher and Stevens [3] were Rb Sr Y Zr Nb
adjusted to gwe internal agreement, taking into ac-
count values for Y measured on two of the rocks by W-1 23 185 20 95 70
neutron activation [4]. The concordant values finally G-2 180 480 10 295 9.0
GSP-1 245 235 28 495 12 0
adopted are hsted m table 1. Y and Nb values m tlus
AGV-1 72 655 18 245 12 0
table are somewhat lower than many prexaous meas- BCR-1 49 325 32 170 12 5
urements Indicate. The results of table 2 can be recal-
 8 J R Cann, Ocean floor basalttc rocks

Table 2
Analytical results quoted in p p m Values for K and T~ are taken from major element analyses Key to symbols * not analysed,
B basalt, D dolerlte, F fresh, A aphync, P porphyritic, (ol) ohvme, (pl) plagloclase, M metamorphosed, S spdlte, S(G) spllmsed
basalt glass, U amphlhohte, W weathered, R plagloclase replaced by muscovite

Sample no Rb Sr Y Zr Nb K T1 Petrography

4519 11 10 145 20 85 14 2600 7200 DFP(ol)

4519 25 16 195 25 105 18 4100 9200 DFA
4519 45 8 250 23 105 * 4400 8600 BFA
4519 46 13 195 22 100 * 3500 8300 BFA
4519 54 10 240 22 100 19 * * BFA
4519 55 8 200 24 95 14 4200 8200 BFA
4520 1 4 135 21 75 10 2250 7400 BFA
4520 4 7 140 23 80 * 2150 7400 BFA

5106 1 0 22 38 110 55 850 9500 BMS(G)

5106 2 0 90 40 125 40 500 10500 BMS
5106 7 0 65 36 140 35 350 11100 BMS
5106 12 10 15 37 125 55 850 10700 BMS(G)
5106 18 0 110 37 130 30 350 11000 BM
5111 2 15 145 29 100 4 0 1350 9200 BFA
5111 5 30 135 28 90 10 2250 9100 BFA

5610 2 3 5 109 25 60 4 0 4150 6500 BMU

5610 3 4 5 150 22 60 7 5 6650 6100 BMU
5623 1 8 5 80 29 64 2 5 6800 7100 BW
5978 9 9 5 110 16 52 7 0 4250 4700 DW
5985 10 3 0 110 23 62 7 5 1500 5800 BMU
5985 22 3 0 165 25 100 12 2500 8400 BMU
5985 25 48 260 23 100 8 0 14400 6400 DMUR
5985 28 22 130 26 65 8 0 9950 7100 BMUR

6213 1 ] 0 88 16 41 20 850 3200 DFP(ol)

6216 6 05 88 23 63 25 650 5700 BFP(pl)
6216 28 25 125 29 72 25 900 7400 BFA
6221 3 05 125 34 100 4 0 600 9000 DFA
6224 2 65 125 27 63 55 3300 7200 BW
6224 3 70 115 23 60 65 2750 5800 BW
6231 30 86 20 50 70 850 5400 BFA
6243 2 30 135 25 79 70 1750 6800 BFA
6243 4 30 135 23 80 75 1650 7200 BFA
6244 30 113 20 58 50 1650 5400 BFA
6245 30 125 19 57 70 1650 5300 BFP(pl)
6252 10 12 5 171 29 122 65 * * BW

Table 2 cont~uns p e t r o g r a p h i c characteristics o f the m e n I d e n t i f i c a t i o n n u m b e r . S t a t i o n s 4 5 1 9 and 4 5 2 0

rocks set o u t m c r y p t i c f o r m and e x p l a i n e d in the are o n the crest o f the Mld-AtlanUc Radge at 4 5 ° N
a c c o m p a n y i n g legend. S u p p l e m e n t a r y r e f o r m a t i o n [5,6] Major e l e m e n t analyses are available for all
a b o u t t h e s e rocks, including m o r e d e t a i l e d p e t r o - e x c e p t 4 5 1 9 54. S t a t i o n s 5106 and 5111 are o n the
graptuc d e s c n p t t o n s and m a j o r e l e m e n t analytical crest o f the Carlsberg Radge in the I n d i a n Ocean at
data, can be o b t a m e d f r o m r e f e r e n c e s [ 5 - 1 1 ] . 5½°N [7,8]. Stations with numbers between 5610
The sample n u m b e r s in table 2 consist o f a f o u r and 5985 are f r o m P a l m e r Ridge in the NE A t l a n t i c
figure Discovery s t a t i o n n u m b e r , f o l l o w e d b y a s p e c t - [9,10] S t a t i o n s w i t h n u m b e r s b e t w e e n 6213 and
 J R Cann, Ocean floor basaltte rocks 9

6221 are from the Alula-Fartak fracture zone m the m the fresh basalts, are very much reduced when all
Gulf of Aden [12]. Malor element analyses have been of the samples are considered
completed, but not so far pubhshed, for the rocks Sr is less affected by secondary processes Changes
from this area. Statmns with numbers from 6224 to in its abundance are largely controlled by changes in
6245 are from the median valley and its surrounding abundance of the Ca sites into winch it substitutes
lulls In the centre of the Gulf of Aden [ 11 ] Station Extreme ocean floor weathering leads to depletion of
6252 is on the flanks of the Carlsberg Ridge at Ca [15], and Sr can be expected to fall under these
2 ° 46 4'N, 59 ° 51 7'E, 3833-3718m conchtlons, though none of the analysed basalts are
sufficiently weathered to show this to any great de-
gree Greenscinst facies metamorpinsm, too, often
2 Results leads to a depletion of Ca sites in the rocks, partic-
ularly when the pressures of oxygen and carbon di-
The results of the analyses are set out in table 2, oxide are low, and this is accompanied by a decrease
together with petrographic characteristics and values m Sr The analyses of the splhtlsed basalt glasses
for K and TI taken from the major element analyses 5106 1 and 12 [8] show tins effect well The result is
of these rocks ]5,6,8,10,11 and unpubhshed] that the high correlations shown by Sr with Rb, Nb,
Table 3 contains two correlation matrices calculated K and Zr for the subset of fresh basalts are all re-
from the data of table 2, one for all samples analysed duced when all the samples are considered. Sr is not
(the upper right half of the table) and the other for so strongly affected by secondary processes as Rb and
the subset of all fresh basalts (the lower left half of K. but is more affected than Nb and Zr.
the table) Two matrices have been calculated because The other four elements considered here, Y, Zr,
there is a difference m the behavlour of two groups of Nb and TI are interesting in that their abundance
the elements which is brought out by tins division seems to be httle if at all affected over the range of
Rb and K, and to a lesser extent Sr, are affected by secondary processes that have affected these rocks.
processes of weathering and metamorphism, so that Table 2 shows that even m the rocks subject to pro-
their abundances are controlled to some extent by found secondary chemical changes, such as the meta-
the post-volcanic history of the rock rather than by morphosed basalt glasses 5106 1 and 12 and the
effects associated with its igneous petrogenesls, whale weathered basalts 5623 1 and 6224.2, the values of
Y, Zr, T1 and Nb apparently preserve their original Y, Zr, Nb and T1 are comparable m every way with
abundances through complex post-volcanic processes. those of the fresh basalts In these altered rocks some
Both Rb and K have previously been shown to elements have changed their abundances by a factor
become enriched in ocean floor basalts dunng ocean of 10 or more [8], and anytinng of tins magmtude
floor weathering [ 13,14], and tins effect is well seen would be easily detectable m Y, Zr, Nb and T1 In-
here in such samples as 5623 1,5978 9, 6224.2 and stead, their nearly constant abundances in, for exam-
6224 3. Both elements too apparently become deplet- ple, the rocks of haul 5106, winch were presumably
ed m greenschlst facies rocks (5106.1,2,7,12), though onginally comagmatlc, though now of very contrast-
they are much less affected by ampinbohte facies ed mineralogical compositmns [8], suggest that any
metamorpinsin (5610 2,3, 5985 10,22) This is proba- changes cannot have exceeded about 10%.
bly an effect of the relative availabilities of atomic This point is brought out again from an examina-
sites suitable for K and Rb in the different mineral tion of table 3 Some pairs of these four elements, in
assemblages. When K is enriched by secondary pro- particular Zr and T1, and Y and TI, are very Inghly
cesses, as for example in 5985.25 and 28 where pla- correlated in the fresh basalts (lower left hand half of
gioclase is extensively replaced by muscovite [ 10], the table) When all of the samples are considered
Rb also becomes enriched, ranging up to nearly (upper right hand half of the table) these correlatmns
50 ppm Such effects lead to the high correlatmn are maintained or even Increased, while the correla-
between Rb and K not only m the fresh basalts, but tmns with the elements that are affected by second-
also for all of the samples, whale correlations with ary processes are dlmimshed from their previous posi-
other elements, such as Nb, Sr and Zr, winch are high tive values, and often become negative. Thus consid-
 10 J R Cann, Ocean floor basaltic rocks

Table 3
Correlatlon matrix for elements listed m table 2 The upper rtght hand half of the matrix was calculated from the results for all
samples, the lower left hand half from the results for fresh basalts and dolerltes only (those with F m the column labelled Petro-
graphy m table 2)

Rb Sr Y Zr Nb K TI

Rb 1 +0 61 - 0 25 +0 03 +0 34 +0 89 - 0 16
Sr +0 73 1 - 0 32 +0 11 +0 62 +0 56 - 0 07
Y -0.15 +0 10 l +0 74 - 0 34 - 0 25 +0 85
Zr +0 58 +0 80 +0 52 1 +0 11 - 0 14 +0 94
Nb +0 90 +0 82 - 0 23 +0 49 1 +0 26 0.00
K +0 84 +0 92 - 0 10 +0 64 +0 83 l -0.21
T1 +0 43 +0 67 +0 75 +0 95 +0 29 +0 51 1

pore Zr eratIon of abundances o f Z r , Nb, T1 and Y and of

their correlations leads to the conclusion that over
lr30i
the range of secondary processes sampled no substan-
.--I--. tial change of abundance takes place. At the m o m e n t
II

0
too few samples have been measured to check this
100 conclusion statistically, but work is in progress which
will enable this to be done
The high posltwe correlations observed between
.."I. , i,,

50 some of these element pairs are themselves highly

interesting Fig 1 shows two of them graphically,
those between Y and TI, and Zr and TI, for the
33 samples for which TI values are available Inspec-
0 SO00 10000 ppm T, tion of the diagrams, with their approxamate error
bars, indicates that most of the observed devaatton
from a linear relationship could be caused by analyt-
pp rn Y ical vanablhty. There is an almost complete one to
one correspondence between values for one element
l,o
• • and those for another over the suite of rocks exam-

30
÷ ined.

• • o*
.o

o°
•i • • .° • • 3. Discussion
20 i

A full discussion of the petrogenetlc ]mphcations

of the results and of the correlation matrix would
10
unbalance this paper A model can be constructed
which accounts for the vanatlon seen, and the high
0 I I , , and low correlations observed, In terms of a homo-
0 5000 10000 ppmr,
geneous mantle involved in repeated partial melting,
in which the vanables are the number of times partial
Fig 1. Relationship between Zr and T1, and Y and TI in 33 of melting has occurred, the degree of partial melting
the samples for which TI values are available Error bars are and the extent of crystal fractlonatlon during sohdifi-
approximate only. cation of the melt This model will be the subject of
 J R Cann, Ocean floor basalttc rocks 11

another paper. Doubtless other vahd models can also 4519 and 4520 for trace element analysis, and Mr.
be constructed to account for the data. C.K. Winter for help m making the determinations.
However, there is one feature of the results that
deserves further comment here If, as these results
m&cate, T1, Zr, Y and Nb contents remain unchanged References
m ocean floor basalts throughout processes of weath-
ering and metamorphism, and ff ~t ~s correct to gener- [1] K Nornsh and B W Chappell, X-ray fluorescence spec-
ahse from these samples that there is a very close trography, m Physical methods m determinative miner-
correlation world-wide between Zr and T1, and Y and alogy, J Zussman, ed (Academic Press, 1967)
[2] F J Flanagan, U S Geological Survey standards - II
T l m ocean floor basalts, then tlus might be a very First compdatlon of data for the new U.S G.S rocks,
powerful method of identifying ocean floor basaltic Geochma Cosmochlm. Acta 33 (1969) 81
rocks as such even after the operation of profound [3] M. Flelscher and R.E. Stevens, Summary of new data
secondary processes. Optuohte complexes and similar on rock samples G-1 and W-l, Geochlm Cosmochtm
occurrences of volcamcs m mountain belts have fre- Acta 26 (1962) 525
[41 J -G SchxUlng,Sea floor evolution rare earth evidence,
quently been considered to represent fragments of Phd. Trans Roy. Soc A (in press).
ocean crust caught up m orogemc movements Iden- [5] I D Mug and C E Tdley, Basalts from the northern
tification of the pdlow lavas and dyke rocks m such part of the rift zone of the Mid-Atlantic Ridge, J
sequences as ocean floor basalts has depended m the Petrol 5 (1964) 409
[6] I D. Mutt and C.E. Tdley, Basalts from the northern
past on companson of major element contents or on
part of the Mid-Atlantic Ridge II The Atlantis collec-
contents of alkah elements, both of which are very tions from 30°N, J. Petrol. 7 (1966) 193.
sensitive to change by the secondary processes that [7] J.R Cann and F J Vine, An area on the crest of the
often accompany the incorporation of such volcamc Carlsberg Ridge, petrology and magnetic survey Phd
matenal m a mountain belt. The more such rocks Trans Roy. Soc A 259 (1966) 198
[8] J R Cann, Splhtes from the Carlsberg Ridge, Indian
have been altered, the less certain ~t is by these crit-
Ocean J Petrol 10 (1969)1.
eria that they started as ocean floor basalts However, [9] J.R Cann and B M Funnell, Palmer Ridge a section
the defimtlon here of a very restricted composition through the upper part of the ocean crust9 Nature 213
space which is strongly resistant to secondary (1967) 661
changes, m terms of T1, Zr and Y, might allow recog- [10] J.R Cann, Petrology of basement rocks from Palmer
Ridge, N.E. Atlantic Phd Trans Roy. Soc A (m
ration of original ocean floor basalts even when they
press).
have become highly altered Invest~gatlon of the T1, [ 11 ] J R Cann, Petrology of basalts dredged from the Gulf
Zr, Y space occupied by basalts from other sources of Aden, Deep Sea Res 17 (1970) 477
wdl be necessary before ttus is certain, as pubhshed [12] A S Laughton, R B Whltmarsh and M T Jones, The
data is very fragmentary, b u t such data as there is evolution of the Gulf of Aden, Phil Trans Roy Soc
A 267 (1970) in press
suggests that the geologically ~mportant &stmctlon
[ 13] J A Phdpotts, C C Schnetzler and S.R Hart, Sub-
between calc-alkalxne volcanlcs and ocean floor bas- manne basalts some K, Rb, Sr, Ba, rare earth, H20 and
alts might be made on this basis [16]. It will certainly CO2 data bearing on thetr alteration, modification by
be mterestlng to pursue this further, and work along plagloclase and possible source materials Earth Planet
these hnes is proceeding. Scl Letters 7 (1969) 293
[ 14] S R Hart and A J Nalwalk, K, Rb, Cs and Sr relation-
ships m submarine basalts from the Puerto Rico Trench
Geochlm Cosmochma Acta 34 (1970) 145
Acknowledgements [151 D H Matthews, Weathered basalts from Swallow Bank,
an abyssal hdl m the NE Atlantic, Phil Trans Roy Soc
I would like to thank Professor C.E. Tdley and Dr. A (m press)
[ 16] S.R. Taylor and A J R. Wlute, Trace element abund-
I.D. Muir for prow&ng samples of basalt from hauls
ances m andesltes, Bull Vole 29 (1966) 177