Melting relations of some lavas of R6union Island, Indian Ocean

M I N E R A L O G I C A L M A G A Z I N E , SEPTEMBER
~ 9 7 I, VOL. 38, PP. 3 4 4 - 3 5 2
Melting relations of some lavas of R6union Island,
Indian Ocean
C. E. T I L L E Y
Department of Mineralogy and Petrology, University of Cambridge
R. N. T H O M P S O N
AND W. J. W A D S W O R T H
Department of Geology, University of Manchester
AND B. G. J. U P T O N
Department of Geology, University of Edinburgh
SU MMAR Y. The one-atmosphere experimental melting and crystallization relations of some basic lavas
of R6union Island, Indian Ocean, are investigated. They comprise products of the extinct volcano
Piton des Neiges and the active volcano Piton de la Fournaise. New data and interpretations are
presented for the I939 oceanite and its contained glass, originally described by Lacroix (I939a). The
liquidus v. iron enrichment curves for these R6union lavas fall about 2o ~ above those of the lavas
of Kilauea, Hawaii. The primitive liquid reaching the surface contained about IO % MgO.
THE island of R6union is built of lavas of two shield volcanoes, Piton des Neiges
(3069 m) and Piton de la Fournaise (263I m). The former is extinct and deeply eroded,
exposing a complex of intrusions ranging from picrite to quartz syenite. The volcanic
succession revealed in Piton des Neiges is of an older series dominated by olivine basalts,
referred to as the Oceanite Series, and a younger series ranging from feldsparphyric
basalts through intermediate lavas to trachytes, known as the Differentiated Series.
The most recent study of the geology of this volcano is given by Upton and Wadsworth
(I966, 1967).
Lacroix, in a series of contributions (I936, x939a, I939b), has devoted attention in
particular to the still-active volcano Piton de la Fournaise, with publication of
numerous chemical analyses of its basalts. More recently, age determinations and
Sr isotope ratio measurements have been carried out by McDougall and Compston
(1965). According to these investigations the STSr/S6Sr values for the basalts and an
oceanite from R6union fall within an extremely narrow range, from o.7o4~ to o'7o46,
and no difference exists between the ratios for rocks of Piton des Neiges and Piton
de la Fournaise, indicating that the source region for both these volcanoes was very
homogeneous in its STSr abundance.
Melting experiments have been carried out on oceanites and basalts from both
Piton des Neiges and Piton de la Fournaise. Some of these rocks have already been
described and analysed (Upton and Wadsworth, I966), but in other cases hitherto
unpublished analyses are presented. In addition to this recently collected material,
9 Copyright the Mineralogical Society.
MELTING
u~ u~ t-.~,o o , , ~ - o ~
o r-oko~
o o ~o
RELATIONS
OF LAVAS
I
I
345
~
~
~
~
~ 0
~
~
0 ~
~
-~P
~
"0
ox.r ~ o x ~ n x t t - o
~ o
o~
~:o ~
~:u ' ~.
~
~.~
~
~
~=~.o
~"or
~,~
..=~
~
~
~r~
.~,.cJ
Z
r.~
unen~o
u~ rn~.o o ~ o o~
r
0.~o ~ o ..,,r
~ e>
~o
'~ ~ ~
Z~,O
ge
o
~-.7
q- ~ q-oh h h q-~b b b
o
~
~
~
oo
e~
~,~
~r
~ ~,.~ ~'g~
~'~'~
t~
E
~
r . o ~o ~o on ~..oo ko
r~e~ o
~
en
03
So
oxtc~r-.oxe~ u 3 o u30~e~
~Z~ ~ 1 7 6 1 7 6 1 7 6 187 6 1 7 6
u~rn f ~ o x ~ ~ ~ - ~ r - o ~
0 ~
?
oo
~
q ' h ,~ q - / - ~ b ~ / n b
e~et
=
<
"~
b
3
~ . ~
11;'~,]~'11
I I I I 1
I
IIIIIIIIII
1
IIIIIIII11
.o ~ y ,
o
~
oo
,q.o t.-. t-.~D o ~ v ~
9~~176
,,~
o
~
~
~J-~
O~
o
r~
I1"~
o
~ g.~
:''~176176176
8
mm
b h , b G / - - r q-;~ b 6
~ o ~.~
,-,~ o ^ o +1
~o
,.~o ~-o o o c 7 c ~
!~
~ ~.-~
346
C. E. TILLEY, R. N. THOMPSON, W. J. WADSWORTH, AND
two samples of basalts (verging on oceanites) from Lacroix's original collection of the
products of the I939 eruption of Piton de la Fournaise have also been studied. We
are greatly indebted to Dr. F. Conqu6r6 of the Department of Mineralogy of the
Museum National d'Histoire Naturelle, Paris, for forwarding us these two specimens.
FIGS. I and 2 : FIG. I (left). Crushed fragments of I939 Piton de la Fournaise basaltic pumice in oil
(n = 1.58). Note abundant fragments of olivine phenocrysts, white or grey with high relief. Halfcrossed nicols • I289 FIG. 2 (right). Oil mount of powder from the bottom of the bottle containing the
pumice fragments: shown in fig. I. Note the total absence of olivine phenoerysts. The powder was
formed by abrasion of the vesiculated interior of the pumice fragments. Half-crossed nicols • ] 289
The liquid mount has n = I '5 8.
The chemical data are presented in table I. The rock types range from true oceanites
(picrite basalts) from both volcanoes, to basalts, principally from Piton des Neiges,
and include also a basaltic hawaiite from the Differentiated Series (Re 332). The
range in silica percentage is from SiO2 43"57 to 49"77, and in iron enrichment, defined
as the ratio ( F e O + F e 2 0 3 ) / ( M g O + F e O + F e 2 0 3 ) , from o'316 to o'754. These values
relate to whole rocks and groundmass or glassy fractions separately analysed.
The 'basalt' specimens of Lacroix are in effect oceanites of the eruption of I939
at Piton de la Fournaise. One of these rocks (875G), erupted on 6 January, was
analysed by Raoult (Lacroix, I939a, p. 4o8). The second sample (i93%) was also
erupted in I939, but the date not further specified. Analysis of the first specimen
(875(3) showed I5.2I % MgO and 45"84 % SiO2 and is quoted here in table I no. I.
Both samples are of identical appearance, consisting of two parts: light-brown coloured
glass, finely vesiculated and containing scarce fine olivines, and irregular seams o f a
dark slaggy material, which forms the borders of larger vesicles. These seams are not
vesiculated and prove to be composed of olivines wrapped in dark glass.
The impression received from Lacroix's account is that this analysis, with I5-2I %
MgO, refers to a completely glassy basalt and this conception has been repeated in
the literature (Wadsworth, I967). The analysed material must, however, have contained olivine phenocrysts in part from the slaggy borders of the larger vesicles. In
the two photomicrographs presented here (figs. I and 2), the first is of a mount of the
B. G. J. U P T O N ON M E L T I N G R E L A T I O N S OF L A V A S
crushed fragments o f the Piton de la
F o u r n a i s e oceanite (I93%) in oil (n =
1.58) showing a b u n d a n t olivine p h e n o crysts; the second (fig. z) is o f a m o u n t
representing the p o w d e r f o u n d in the
b o t t o m o f the bottle c o n t a i n i n g the
p u m i c e fragments o f the oceanite. This
p o w d e r is totally free f r o m olivine
p h e n o c r y s t s a n d is f o r m e d by a b r a s i o n
o f the friable vesiculated interior o f the
pumice fragments, in which process
glass shards are separated f r o m olivine
crystals. D u r i n g the shaking o f the
bottle, the olivines segregate f r o m the
shards in the a b r a d e d material, owing
to their shape a n d density. I t seems
evident t h a t Lacroix, in describing his
a n a l y s e d r o c k as a glass, m u s t have
studied such an a b r a d e d fraction o f
the friable pumice a n d d e e m e d it to be
wholly representative o f the analysed
material.
9
9
9
332
9
9
~31~6r~
~
347
[7
14gm
grog ~ : x or, F
'?,,o,o?,3,
N
-~50
1939
v
40
50
v
~
30
v
~a
20
10
9 20
Ft~. 3. FMA (FeO§
plot of experimentally studied lavas of R6union.
Points refer to analyses so designated in table I. Re
332 has been shown with a different symbol from the
other specimens to emphasize that it comes from the
Differentiated Series of Piton des Neiges. In addition
av. F represents the plot of the average of 33 basalts
from Piton de la Fournaise, calculated by Upton and
Wadsworth (1966). The lines marked 875G exp and
193% give the trends of the partially analysed
samples, based on their iron-magnesium ratios.
TABLE I I
SiO2
Al~O3
FelOn
FeO
MnO
MgO
CaO
Na20
K20
TiO~
P2On
H~O+
H20--
REg
II4gm
AvF
REg
I 14gm
AvF
48"56
14"38
2'34
8"94
0-20
7"18
11"44
2"65
0"77
3"o2
o'29
0'07
nil
48"73
14"44
3"76
7"88
o.I8
7"37
11'30
z'39
0"64
2'37
o'39
0"35
0"03
47"98
13"92
2"60
9"o5
o'13
7"67
I 1.6I
2'34
o-89
2-98
o'25
0.80
O'2I
99'84
99"83
IOO'43
Glass from oceanite 193% (table I). Iron enrichment o-61x.
Groundmass of 196I oceanite (Re II4), Piton de la Fournaise (Upton and Wadsworth,
1966). Iron enrichment o'6t2.
Average of 33 basalts from Piton de la Fournaise (Upton and Wadsworth, I966). Iron
enrichment o.6o~.
348
C. E. TILLEY, R. N. T H O M P S O N , W. J. W A D S W O R T H , A N D
The material of fig. 2 must, in fact, have the composition o f REg, a pure glass
separated from I93% with only 7"~8 % o f M g O (table I, REg). I f an analysis o f such
a composite rock is to be representative, sampling o f a large mass would be required.
Di
332
~l~331gm(53)
(6o)I[3 ,'%
REg . /
E.,J
av.N (54)
bjso (ss)
~|14arn (S|)
YY
~331 (2:114 (31)
OI
Hy
FI~. 4- Normative diopside, hypersthene, and olivine (recalculated to Ioo) plotted for the R6union
lavas. In addition are plotted the averages F (cf. fig. 3) and N; the latter being the average of 4 basalts
of the Oceanite Series of Piton des Neiges (Upton and Wadsworth, I966). The plot of the Kilauea
Series, dashed line joining solid circles, is drawn for comparison. The lines 331--33Igmand I I4-I I4g~
join the compositions of oceanites and their respective groundmasses. Figures in brackets give the
total normative feldspar content for these lavas.
As anomalous results were found in the melting work using Lacroix's analysis for
our crushed material, we have carried out melting experiments on a sample actually
reanalysed for MgO, FeO, and F%O3 (table I, analysis 875G exp). This material
proved even richer in olivine than is represented by the material shown in fig. I. The
melting data are in accord with the iron enrichment o f this reanalysed sample (875 G
exp, fig. 8).
B. G. J. U P T O N O N M E L T I N G
RELATIONS
OF LAVAS
349
Some of the chemical characters of the rock types experimentally studied are
revealed in figs. 3 to 7. Fig. 3 shows the (FeO+FezOz)-MgO-alkalis relationship;
the trend extends from the oceanite (Re 33I) to the basaltic hawaiite (Re 332). The
plot of the glass of the I939 oceanite (REg) is close to that of the crystalline groundmass of the oceanite (Re I I4) and also to that of an average composition of 33 basalts
of Piton de la Fournaise (table II). No glass more basic than that of REg has been
recognized.
Fig. 4 portrays the normative
26
331~ I
I
I
I
I
I
diopside-hypersthene-olivine relationships of the suite, which are also
24
114
compared with a typical Kilauea,
|
Hawaii, series. The straight lines 331-22
875Gexp
33Igm and II4-II4g m join the whole
rock and groundmass compositions of
2O
the oceanites Re 331 (27) and Re 114
1939p
18
(3I). These contain abundant phenocrystic olivine, which was removed
from them before the chemical analyses O ~
of the groundmasses were made. The
increase in total normative feldspar of
the successions--shown in brackets
12
against the points on the diagram-indicate that olivine is the main conI0
134
,~,ok
stituent removed in the differentiation
course. The rock type Re 332, which
8
~
ov.F
/X
114gin
/
is shown as a plotted point near the
diopside-hypersthenejoin, although an
alkali rock, crystallized in an oxidizing
4
I
I
I
I
I
I
/
environment, is rich in magnetite and
43
44
45
46
47
48
49
50
shows a high content of Fe208 (7"89
S~O2
%), which is responsible for the nor5. MgO-SiO~ plot of the experimentally
mative hypersthene. If the Fe2Oz FIG.
studied lavas, with the MgO values of 875 G exp
content is adjusted to 2-o %, the
and I93% shown as lines across the main trend.
analysis becomes nepheline-normative
(2-34 %). This change is indicated on fig. 4 by the arrowed line from the point directed
to the diopside-olivine join.
The role of olivine as the prime extract in the differentiation of the oceanites is
emphasized further in the plot of the MgO-SiO2 variation (fig. 5). Remote from this
trend is the point representing the basaltic hawaiite, where plagioclase is the dominant
separating phase. The anomalous position of Re 332, relative to the basalts, is also
revealed in fig. 6, which plots SiO2 versus total alkalis. The SiO2-iron enrichment
trend is shown in fig. 7- It is clear that, as a whole, the Oceanite Series of Piton des
Neiges and the lavas of Piton de la Fournaise occupy a transitional position between
the typical tholeiites of Hawaii and alkali olivine basalts; as becomes apparent, too,
\
350
C. E. T I L L E Y , R. N. T H O M P S O N , W. J. W A D S W O R T H , A N D
from inspection of fig. 4, where they are compared with the Kilauea (Hawaiian)
Series.
The results of the melting runs on the thirteen rocks are set out in table III, and the
liquidus temperatures related to iron enrichment in fig. 8. The experiments were
0.80
. . . .
,
,
,
. . . . . .
E1~3~
,
I.....
171
0
I'
r
LIMIT ~%~
ov F.~(~REg
3 4 3 ~ ~
.
,,O/
69
070
C ) ~s~
~o6o
,'~
O~,so
L
,'o
sio~
,',
.'8
/
I
I
RE
g,
gm
"~
0 5o
o
0 40
9m
~pl93~
b,j
,~
~
t~s
i(~
,'9
03_43(3
,0
~3
"k)U4
I
44
I
V
I
45
I
I
46
,
47
I
48
I
,
49
I
50
Si O~
Fins. 6 and 7: FIo. 6 (left). SiOs-alkalis plot of the Rdtmion lavas with the trend of the Hawaiian
tholeiite limit indicated. Fro. 7 (right). SiOz iron enrichment (FeO+Fe~Os)/(MgO+FeO+Fe~O3)
plot of the R6union lavas.
TABLE III. Results of melting experiments on lavas of Rdunion
Spec, no.
Iron
enrichment
Re 331
Re 114
875G exp
I93%
Re 343
Re 134
Re I71
Re 168
Re 33Igm
Re 69
REg
Re 18o
Re 332
o'3In
o'35~
0.379
0'397
o'44o
0"564
0"580
o-582
o'63~
0'643
O"611
0"697
0"754
n of glass
ol (I52Z~
ol (147o~
ol (I443~
ol (1422~
ol (137o~
ol (t24o~
ol (1zzo~
ol (1234~
ol (I19o~
ol (I193~
ol (1178~
ol (1192~
!ot (1185~
pl
pl
pl
pl
pl
pl
pl
pl
pl
pl
pl
pl
ol
(I165~
(116o~
(116o~
(I I60~
(II67~
(1174~
(1169~
(1 I6O~
(1179~
(117o~
(II65~
(1192~
(I14O~
cpx (II50~
cpx (115o ~
cpx (I157 ~
cpx (I I60 ~
cpx (1156~
cpx (1138~
cpx (1145 ~
cpx (I I4 o~
cpx (1179 ~
cpx (I153 ~
cpx (II65 ~
cpx (1145 ~
cpx (II18 ~
1"657
1"643
1'65z
I'647
1"64z
1.613
1.6IO
1.6o8
1'595
~-6o6
I'602
1"6o4
1.618
carried out in an argon atmosphere below I23O ~ and in air above this temperature.
Capsules of silver-palladium alloy (AgToPd~0) were used in runs below I I5o ~ and
of platinum at higher temperatures (sealed when run in air). All runs were for one
hour, except for some lower-temperature experiments, which required several hours
to equilibrate. The succession of phases separating was olivine-plagioclase-clinopyroxene, except for Re I8O, where olivine and plagioclase crystallized at the same
time, and for Re 332, where plagioclase was the highest temperature phase, followed
by olivine at 45 ~ lower temperature. The most mafic 'aphyric' basalts studied were
B. G. J. UPTON ON MELTING RELATIONS OF LAVAS
35[
Re I34, with 2"0 % olivine phenocrysts and Re I7I , with o . 1 % olivine phenocrysts
(3ooo point modal analyses).
Fractionation is clearly dominated by an olivine control, strikingly revealed in the
upper (accumulative) curve of fig. 8. Plagioclase plays a significant role only at lower
1550
I
I
I
I
I
I
I
I
I
(~331
1500
1450
' ~ ~
1400
875O e x
p
KI P b ~ C ) ~ C ) 343
I.U 1350
m
ZD
~ 1300
I
I,LI
D.
IE
t.U 1250
iI
1200
RE~gg
(~,,~31
~ g~ ~
~
/
1150
1100
IO5O
0"300
I
I
I
I
I
I
I
I
I
0.350 0-400 0.450 0'500 0 '550 0.600 0.650 0-700 0.750 0"800
FeO+FezO3 //~AgO+FeO+Fe203
FIG. 8. Plot relating liquidus temperatures and iron enrichment for the experimentally investigated
lavas of Rdunion. The trend is compared with that of the Kilauea Series (Thompson and Tilley,
1969).
temperatures. Rocks Re 180 and Re 332 depart significantly from the iron enrichment
v. temperature curve defined by the more marie rocks. This is no doubt related to the
prominent role of plagioclase as a liquidus phase in these two lavas, with factors other
than iron enrichment also controlling the liquidus temperatures. The primitive liquid
reaching the surface would appear to have an iron enrichment near o-56 (i.e. near
Re I34) where there is a change of slope in the liquidus v. iron-enrichment curve.
352
C . E . "I-ILLEY Ex AL. ON R E U N I O N L A V A S
This w o u l d c o r r e s p o n d to a liquid containing a b o u t IO % M g O , which is in fact near
the primitive low-pressure liquid for the K i l a u e a Series, H a w a i i (iron e n r i c h m e n t
~'~o'53, M g O ~ I o %). A t greater d e p t h it is likely that an even m o r e basic liquid,
with higher olivine content, w o u l d have existed, b u t this does n o t a p p e a r to have r e a c h e d
the surface.
Acknowledgements. We are greatly indebted to Dr. F. Conqu6r6 who has sent us from Paris two
of the critical oceanite pumices studied by Lacroix from the I939 eruption of Piton de la Fournaise,
and to Dr. J. Esson and J. H. Scoon for providing analytical data on these specimens.
REFERENCES
LACROI• (A.), I936, Le Volcan actif de 17le de la Rdunion et ses produits. Gauthier-Villars, Paris.
I939a. Compt. Rend. Acad. Sci. Paris, 209, 4o5-8.
I939b. Ibid. 42I-3.
McDoUGALL (I.) and COMPSTON(W.), I965. Nature, 207, 252-3.
TI-IOMVSON(R. N.) and TILLEY(C. E.), I969. Earth Planet Sci. Letters, 5, 469-77.
UPTON (B. G. J.), I967. Proc. Geol. Soc. No. I637, 54-6.
and WADSWORTH(W. J.), I966. Bull. Volc. 29, 7-24.
WADSWORTH(W. J.), I967. Proc. Geol. Soc. No. I637, 53-54.
-
-
-
-
-
-
[Manuscript received I4 July I97O]