JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
PAGES 133–163
1997
Compositional Gradients and Gaps in
High-silica Rhyolites of the Rattlesnake Tuff,
Oregon
MARTIN J. STRECK∗ AND ANITA L. GRUNDER
DEPARTMENT OF GEOSCIENCES, OREGON STATE UNIVERSITY, CORVALLIS, OR 97331-5506, USA
RECEIVED MAY 12, 1995 ACCEPTED AUGUST 22, 1996
The Rattlesnake Tuff of eastern Oregon comprises >99% of highsilica rhyolite glass shards and pumices representing ~280 km3 of
magma. Glassy, crystal-poor, high-silica rhyolite pumices and glass
shards cluster in five chemical groups that range in color from white
to dark gray with increasing Fe concentration. Compositional clusters
are defined by Fe, Ti, LREE, Ba, Eu, Rb, Zr, Hf, Ta, and Th.
Progressive changes with increasing degree of evolution of the magma
occur in modal mineralogy, mineral composition, and partition
coefficients. Partition coefficients are reported for alkali feldspar,
clinopyroxene, and titanomagnetite. Models of modal crystal fractionation, assimilation, successive partial melting, and mixing of
end members cannot account for the chemical variations among
rhyolite compositions. On the other hand, ~50% fractionation of
observed phenocryst compositions in non-modal proportions agrees
with chemical variations among rhyolite compositions. Such nonmodal fractionation might occur along the roof and margins of a
magma chamber and would yield compositions of removed solids
ranging from syenitic to granitic. A differentiation sequence is
proposed by which each more evolved composition is derived from
the previous, less evolved liquid by fractionation and accumulation,
occurring mainly along the roof of a slab-like magma chamber. As
a layer of derivative magma reaches a critical thickness, a new layer
is formed, generating a compositionally and density stratified magma
chamber.
INTRODUCTION
high-silica rhyolite; partition coefficients; rhyolite differentiation; zoned ash-flow tuff; layered convection
The origin, structure, and differentiation processes in
zoned magma chambers have been approached by many
workers through the study of ignimbrites since the
ground-breaking work of Smith (1960) and Smith &
Bailey (1966). Zoned high-silica rhyolite chambers have
been of particular interest, because the high viscosity
of such magmas inhibits crystal–liquid separation and
because extreme trace-element gradients, relative to
major element variations, are not easily reconciled (Hildreth, 1979, 1981; Michael, 1983; Miller & Mittlefehldt,
1984). The generally high trace-element partition coefficients in silica-rich magmas and the presence of accessory phases with exceedingly high partition
coefficients, coupled with low crystal contents, has made
the quantitative modeling of high-silica rhyolites exacting
(see Mahood & Hildreth, 1983; Michael, 1988). In addition to compositional zonation, compositional gaps in
ignimbrites have been documented with implications for
chamber configuration and tapping mechanisms (e.g.
Blake, 1981; Blake & Ivey, 1986; Fridrich & Mahood,
1987).
The Rattlesnake Tuff of southeastern Oregon is composed nearly entirely of high-silica rhyolite that defines
five distinct compositional and mineralogic clusters. The
differentiation of the different rhyolite compositions, using
newly derived partition coefficients, the configuration of
the pre-eruptive chamber, and the origin of the compositional gaps are presented here. The petrogenesis of the least evolved rhyolite and the evolution of
the mafic underpinning of the Rattlesnake Tuff system
∗Corresponding author. Telephone: 541-737-1201. Fax: 541-737-1200.
e-mail: [email protected] or [email protected]
 Oxford University Press 1997
KEY WORDS:
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
Fig. 1. Regional setting and outcrop pattern of the Rattlesnake Tuff in Oregon. Light stipple represents Blue Mountains Province, dense stipple
shows Rattlesnake Tuff; Χ, proposed source area. Dashed lines with numbers are simplified isochrons in millions of years for NW-migrating
silicic volcanism, after MacLeod et al. (1976). Continuous lines indicate faults; _, Cascade composite volcanoes.
have been discussed by Streck (1994) and will be treated
in detail in forthcoming papers.
THE RATTLESNAKE TUFF
The 7·05 Ma Rattlesnake Tuff erupted from the Harney
Basin, a center of Late Miocene silicic magmatism in
southeastern Oregon (Fig. 1). It is part of a northwestward-younging trend of silicic domes and tuffs
(MacLeod et al., 1976) associated with widespread highalumina olivine tholeiite lavas that define the High Lava
Plains. Intermediate compositions are scarce.
The tuff consists of a single cooling unit, 10–30 m
thick, that probably covered an area of ~35 000 km2 and
represents ~280 km3 of magma (Streck & Grunder, 1995).
The tuff typically has few lithic fragments and ranges
from pumice rich, with pumice clasts as large as 60 cm
near the source, to pumice poor with distance from the
vent (Streck & Grunder, 1995). The tuff is remarkable
for spectacular banded pumices and a salt and pepper
matrix of white and gray glass shards (Fig. 2). Exclusively
white tephra occurs in some basal and distal sections or
in rare basal fallout deposits, suggesting that white tephra
represents magma from the top of the magma chamber
(see Smith, 1979). The tuff is fresh and little welded at
many places, facilitating the sampling of individual vitric
pumices. High-silica rhyolite (>75 wt % SiO2) makes up
>99% of the tuff. Dacite pumices are minor (<1 vol. %)
and cognate mafic inclusions, mainly in dacite, are rare
(p0·1 vol. %) (Streck, 1994).
Five distinct high-silica rhyolite compositions are represented by pumice clasts and shards. From most to least
differentiated, these are referred to as Groups A, B, C,
D, and E. The rhyolites are metaluminous to slightly
peralkaline, with molar ratios of alkalis to aluminum of
0·88–1·03 (Table 1). All pumices are crystal poor (0–1·3
wt % crystals) but mineralogically distinct. Group A and
B pumices are white and A is essentially aphyric (Fig. 2,
Table 2). Pumices of the other groups range from beige
134
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
(a)
(b)
Fig. 2. (a) Banded pumice block in the center of picture is 50 cm across and consists of two to three different high-silica rhyolites. Rhyolite
ranges from gray to white with decreasing Fe concentration. (Note also small black dacite pumice at upper left of banded block.) Tuff is glassy
throughout. (b) Typical vitrophyre of Rattlesnake Tuff containing differently colored high-silica rhyolite glass shards under plane-polarized light.
Color variation is related to Fe concentration as indicated by average FeO∗ (n=3±1 SD) of individual shards obtained by electron microprobe
analysis. Shard compositions correspond well to observed high-silica rhyolite pumice clusters. Horizontal field of view is ~3·5 mm.
135
11·72
0·86
0·08
0·10
(2·32)
3·33
5·47
Al2O3
Fe2O3∗
MnO
MgO
CaO
Na2O
K2O
136
37·4
17·9
53
17·2
6·4
1·9
10·9
1·5
Pb
Zn
Ga
V
Cu
Ni
Cr
169
Zr
96
(36·8)
Sr
Y
19
Nb
116
Ba
0·97
99·7
2·5
11·1
0·8
2·4
18·4
79
18·3
1·6
14·0
1·2
2·5
17·8
88
18·9
101
40·3
175
3·4
25
123
0·98
99·6
0·02
5·06
3·89
0·30
n.d.
0·08
0·86
12·07
0·11
77·60
RT165E
n.d.
12·6
2·0
n.d.
17·8
91
21·1
101
40·6
174
2·0
18
123
0·97
99·2
0·01
5·17
3·68
0·28
0·03
0·09
0·89
11·97
0·12
77·77
RT173D
n.d.
12·6
1·4
n.d.
17·9
87
17·2
100
39·3
175
1·1
10
122
0·99
100·2
0·02
5·08
3·81
0·26
0·04
0·08
0·87
11·91
0·11
77·82
RT173E
n.d.
14·3
1·0
n.d.
17·6
88
20·0
100
39·4
175
2·1
39
122
0·98
99·6
0·02
5·53
3·46
0·26
0·05
0·08
0·87
11·96
0·11
77·67
RT173H
0·35
3·19
6·43
0·28 ± 0·02
3·59 ± 0·24
5·32 ± 0·25
31·5
93
18·0
99
17·1
5·6
2·1
9·8
2·2
39 ± 1
19 ± 1
81 ± 14
17·8 ± 0·4
4±2
1·4 ± 0·5
13 ± 1
1·9 ± 0·6
304
5·4
122
91
1·03
99 ± 2
175 ± 4
2±1
25 ± 12
121 ± 3
0·98 ± 0·01
99·3
0·05
0·10
0·07 ± 0·04
0·02 ± 0·01
0·09
0·08 ± 0·01
11·82
12·0 ± 0·1
1·32
0·12
0·87 ± 0·01
76·53
0·11 ± 0·01
RT34D
77·4 ± 0·7
Mean ± 1r
2·4
9·5
1·9
5·8
17·9
106
19·3
96
32·0
308
3·7
132
91
0·97
99·1
0·03
5·95
3·06
0·33
0·09
0·09
1·37
11·77
0·12
77·18
RT34E
0·6
15·0
4·6
1·4
18·0
101
16·9
98
33·0
302
12·5
131
94
1·00
99·4
0·05
5·74
3·36
0·39
0·02
0·09
1·41
11·80
0·11
77·02
RT120A
2
14
7
n.d.
21·0
101
19
89
32·1
258
12
115
94
1·01
97·5
0·05
5·40
3·72
0·26
0·08
0·10
1·39
11·82
0·12
77·06
RT220A
RT219A
1·8 ± 0·8
12 ± 3
4±2
4±2
19 ± 2
102 ± 3
18 ± 2
94 ± 4
32·2 ± 0·6
293 ± 23
8±5
125 ± 8
92 ± 2
1·00 ± 0·03
0·05 ± 0·01
5·9 ± 0·4
3·3 ± 0·3
0·33 ± 0·06
0·07 ± 0·04
0·09 ± 0·01
1·37 ± 0·04
11·8 ± 0·02
0·12 ± 0·01
77·0 ± 0·3
Mean ± 1r
NUMBER 1
98
38·8
180
3·3
40
121
0·96
99·6
n.d.
5·62
3·35
0·30
0·13
0·07
0·86
12·10
0·11
77·44
RT55B
Group B
VOLUME 38
Rb
XRF (p.p.m.)
AI
prn. total
n.d.
0·12
TiO2
P2O5
76·00
SiO2
XRF (wt %)
RT14F
Group A
Table 1: Chemical composition of Rattlesnake Tuff pumices and glass shards
JOURNAL OF PETROLOGY
JANUARY 1997
137
9·09
0·65
2·18
9·50
1·45
0·19
Eu
Tb
Yb
Lu
Eu/Eu∗
29
Nd
Sm
46
Ce
0·42
19·9
La
Co
Ta
3·76
2·05
Hf
Sc
6·75
Th
1·32
8·86
U
4·8
4·61
Cs
As
4·10
Na2O (wt %)
Sb
0·75
3·40
FeO∗ (wt %)
INAA (p.p.m.)
RT14F
Group A
0·20
1·49
9·62
2·23
0·67
8·83
23
49
19·1
0·08
3·96
3·5
1·43
2·15
6·95
9·93
4·65
4·23
3·11
0·74
RT55B
0·18
1·62
10·55
2·30
0·65
9·58
30
54
21·7
0·05
4·05
5·3
1·42
2·28
7·24
9·87
4·48
4·54
3·86
0·80
RT165E
0·18
1·62
10·35
2·24
0·64
9·35
31
55
20·2
0·06
3·90
3·9
1·50
2·14
7·12
9·29
4·43
4·28
3·58
0·76
RT173D
0·19
1·54
10·22
2·17
0·65
9·06
32
51
19·0
0·05
3·89
4·4
1·52
2·15
7·15
9·62
4·77
4·50
3·62
0·78
RT173E
0·18
1·61
10·54
2·22
0·65
9·57
28
49
19·9
0·08
3·93
4·4
1·55
2·16
7·13
9·47
4·94
4·48
3·35
0·78
RT173H
85
47
12·63
1·07
2·53
9·48
1·35
0·24
51 ± 3
29 ± 3
9·3 ± 0·3
0·65 ± 0·01
2·22 ± 0·05
10·1 ± 0·5
1·56 ± 0·07
0·19 ± 0·01
0·18
37·4
0·1 ± 0·2
20 ± 1
3·47
1·68
2·16 ± 0·07
3·9 ± 0·1
9·09
7·1 ± 0·2
1·26
7·15
9·5 ± 0·4
4·7
3·26
4·7 ± 0·2
4·4 ± 0·6
3·07
4·4 ± 0·2
1·46 ± 0·08
1·14
3·53
3·5 ± 0·3
RT34D
0·77 ± 0·02
Mean ± 1r
Group B
0·26
1·38
9·66
2·70
1·22
12·89
45
90
38·9
0·32
3·69
n.d.
1·33
1·75
9·61
7·71
3·52
3·19
3·42
1·22
RT34E
0·24
1·42
9·79
2·68
1·16
13·57
53
97
39·4
0·28
3·84
5·7
1·41
1·81
9·41
7·69
3·62
3·30
3·30
1·29
RT120A
0·26
1·37
9·61
2·64
1·21
13·45
50
99
38·7
0·14
3·77
4·5
1·39
1·71
9·68
7·84
3·63
3·18
3·44
1·24
RT220A
0·31
1·23
8·48
2·14
1·27
12·11
48
88
37·5
0·58
3·97
5·3
1·70
1·62
9·79
7·57
3·17
2·88
1·94
1·27
RT219A
0·26 ± 0·03
1·35 ± 0·07
9·4 ± 0·5
2·54 ± 0·23
1·19 ± 0·08
12·9 ± 0·6
49 ± 3
92 ± 6
38 ± 1
0·3 ± 0·2
3·8 ± 0·2
5·1 ± 0·6
1·4 ± 0·2
1·71 ± 0·07
9·5 ± 0·3
7·6 ± 0·3
3·4 ± 0·2
3·1 ± 0·2
3·1 ± 0·7
1·23 ± 0·06
Mean ± 1r
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
11·77
1·61
0·09
Al2O3
Fe2O3∗
MnO
5·23
K 2O
138
6·5
17·7
1·5
1·9
9·9
0·9
Ga
V
Cu
Ni
Cr
106
16·4
Pb
0·06
n.d.
11·6
4·2
0·1
18·1
103
13·6
85
29·0
372
12·5
491
78
1·03
99·3
1·3
10·1
2·9
3·6
18·2
113
17·4
82
28·3
404
6·8
664
77
0·93
99·2
n.d.
5·81
3·01
0·32
0·04
0·08
1·91
12·06
0·14
76·63
RT80A
RT62A
0·08
6·29
5·5 ± 0·3
18·8
18·0
4·2
3·3
9·1
2·5
18·0 ± 0·3
2±2
3±1
11 ± 1
1·1 ± 0·3
106
16·1
16 ± 2
107 ± 5
26·0
76
86 ± 4
426
29 ± 1
382 ± 19
9±3
67
1201
78 ± 2
0·99
511 ± 144
0·99 ± 0·05
99·5
0·03
2·97
3·5 ± 0·4
0·04
0·62
0·37 ± 0·06
0·06
1·81
11·88
11·8 ± 0·2
0·08 ± 0·01
0·14
1·72 ± 0·2
76·12
76·9 ± 0·2
RT50D
0·13 ± 0·01
Mean ± 1r
n.d.
9·7
4·0
1·8
17·6
115
14·8
79
27·2
433
12·4
1230
67
0·97
99·6
0·02
4·92
3·84
0·45
0·05
0·09
1·93
11·97
0·14
76·59
RT173B
1120
63
RT173I
1140
68
RT173L
4
8·4
6·3
3·7
17·9
107
16·9
71
25·6
432
24·3
1116
81
0·93
99·2
0·01
6·83
2·35
0·65
(0·50)
0·09
1·95
12·17
0·14
75·31
RT4A
3±1
9·0 ± 0·7
5±2
3±1
17·8 ± 0·2
109 ± 5
16 ± 1
75 ± 4
26·3 ± 0·8
430 ± 4
19 ± 6
1161 ± 51
69 ± 7
0·96 ± 0·03
0·02 ± 0·01
6·02 ± 1
3·1 ± 0·8
0·57 ± 0·11
0·06 ± 0·01
0·09 ± 0·01
1·89 ± 0·07
12·0 ± 0·2
0·14 ± 0·00
76·0 ± 0·6
Mean ± 1r
NUMBER 1
Zn
30·5
90
Y
371
Nb
Zr
Sr
80
379
Ba
1·00
99·7
5·43
3·74
0·43
n.d.
0·08
1·64
11·67
0·13
76·82
RT120B
Group D
VOLUME 38
Rb
XRF (p.p.m.)
AI
prn. total
0·01
3·70
Na2O
P2O5
0·35
CaO
n.d.
0·12
TiO2
MgO
77·11
SiO2
XRF (wt %)
RT165A
Group C
Table 1: continued
JOURNAL OF PETROLOGY
JANUARY 1997
139
59
59
14·55
1·47
2·42
8·38
1·28
0·30
Nd
Sm
Eu
Tb
Yb
Lu
Eu/Eu∗
0·32
1·33
8·58
2·32
1·54
14·44
124
51·7
0·17
3·58
4·5
1·17
1·52
10·04
6·80
2·67
2·64
3·62
1·44
RT120B
113
Ce
49·9
0·08
La
Co
1·56
Ta
3·55
9·86
Hf
Sc
6·87
Th
5·2
3·12
U
As
2·80
Cs
1·16
3·68
Na2O (wt %)
Sb
1·37
FeO∗ (wt %)
INAA (p.p.m.)
RT165A
Group C
0·42
1·23
8·22
2·39
1·81
12·26
49·4
112
50·4
0·39
4·05
n.d.
1·28
1·51
10·73
6·93
2·86
2·71
2·76
1·68
RT80A
0·27
1·45
9·98
2·64
1·34
14·60
59·4
107
45·1
0·36
4·34
4·8
1·33
1·81
10·50
7·94
3·18
3·03
3·08
1·44
RT62A
2·41
6·40
10·39
1·34
3·0 ± 0·2
7·1 ± 0·5
10·3 ± 0·4
1·60 ± 0·14
0·40
12·16
1·97
2·17
7·90
1·20
0·48
1·54 ± 0·20
2·44 ± 0·14
8·8 ± 0·8
1·32 ± 0·09
0·33 ± 0·07
53
57 ± 5
14·0 ± 1·1
117
114 ± 7
53·0
0·3 ± 0·2
49 ± 3
3·57
n.d.
3·9 ± 0·4
4·8 ± 0·4
1·16
2·27
2·8 ± 0·2
1·2 ± 0·1
1·57
2·85
3·3 ± 0·4
RT50D
1·48 ± 0·14
Mean ± 1r
Group D
0·44
1·22
7·95
2·23
2·03
14·01
68
130
54·3
0·11
3·76
3·9
1·19
1·34
10·55
6·47
3·16
2·38
3·93
1·69
RT173B
0·46
1·08
7·32
2·12
2·01
13·10
58
117
51·0
0·12
3·65
3·7
1·14
1·29
10·30
5·87
2·49
2·17
3·28
1·69
RT173I
0·45
1·15
7·58
2·18
2·04
13·80
59
118
52·9
0·24
3·57
3·1
1·12
1·38
10·60
6·02
2·44
2·20
3·47
1·72
RT173L
0·47
1·07
7·15
2·06
1·94
12·42
51
113
49·3
0·26
3·73
3·6
1·05
1·40
10·72
6·15
2·33
3·40
2·35
1·71
RT4A
0·46 ± 0·01
1·14 ± 0·07
7·6 ± 0·4
2·15 ± 0·06
2·00 ± 0·04
13·1 ± 0·8
58 ± 7
119 ± 6
52 ± 2
0·2 ± 0·1
3·66 ± 0·09
3·6 ± 0·3
1·13 ± 0·05
1·35 ± 0·04
10·5 ± 0·2
6·2 ± 0·3
2·6 ± 0·3
2·5 ± 0·5
3·2 ± 0·6
1·68 ± 0·06
Mean ± 1r
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
0·49
3·57
5·57
CaO
Na2O
K 2O
140
23·9
7·7
3·1
8·4
4·2
Cu
Ni
Cr
19
V
Ga
15·4
117
Pb
3·4
10·6
4·1
20·9
18·3
106
14·7
75
25·2
469
22·8
1898
64
0·99
98·8
2·9
9·4
3·0
16·8
18·3
117
14·9
75
25·4
460
26·3
1914
63
0·99
99·7
0·03
5·38
3·88
0·63
0·07
0·09
2·34
12·32
0·18
75·08
RT34C
1·0
8·2
3·5
4·4
18·3
108
15·4
74
25·8
464
29·2
1999
62
0·92
99·1
0·08
6·34
2·64
0·63
0·08
0·07
2·07
12·19
0·16
75·73
RT55A
n.d.
10·7
8·3
n.d.
19·9
110
9·7
78
26·7
488
18·4
1839
71
0·88
100·3
0·01
6·75
2·54
0·37
0·10
0·11
2·37
13·04
0·18
74·52
RT140A
0·3
10·0
4·3
3·7
19·0
120
13·1
77
26·8
474
22·5
2031
62
0·95
99·5
0·02
4·59
4·18
0·54
n.d.
0·10
2·23
12·47
0·16
75·70
RT173A
0·3
8·8
4·7
1·9
18·6
113
10·3
76
26·4
457
23·4
1835
64
1·03
99·5
0·01
4·69
4·61
0·50
0·01
0·10
2·18
12·26
0·16
75·47
RT173C
0·07
0·87
2·72
6·20
0·09 ± 0·05
0·53 ± 0·09
3·6 ± 0·8
5·6 ± 0·8
19·0
17·9
2·1
3·3
9·3
3·1
18·8 ± 0·6
9±8
4±2
9±1
2±2
8·9
60
113 ± 5
13 ± 2
29·9
86
76 ± 1
355
26·0 ± 0·6
467 ± 11
24 ± 3
84
843
64 ± 3
0·91
1923 ± 75
0·96 ± 0·05
99·7
0·03
0·09
0·03 ± 0·02
1·67
12·23
12·4 ± 0·3
0·09 ± 0·02
0·14
2·21 ± 0·12
75·98
0·17 ± 0·01
RT17A
75·3 ± 0·4
Mean ± 1r
4·0
10·6
8·9
1·9
17·8
115
15·4
79
27·7
395
13·6
998
77
0·91
99·3
0·01
6·77
2·22
0·43
0·13
0·09
1·87
12·07
0·15
76·24
RT55C
n.d.
10·0
4·3
4·0
17·5
116
15·9
74
26·8
444
25·4
1825
70
0·98
98·9
0·06
6·62
2·84
0·69
0·10
0·09
2·09
12·11
0·16
75·23
RT127A
NUMBER 1
Zn
25·6
76
Y
457
Nb
Zr
Sr
63
1947
Ba
0·96
99·5
0·04
5·77
3·56
0·55
0·15
0·07
2·07
12·19
0·18
75·42
RT34B
Banded
VOLUME 38
Rb
XRF (p.p.m.)
AI
prn. total
0·02
0·12
MgO
P2O5
2·23
0·09
12·42
Al2O3
MnO
0·17
TiO2
Fe2O3∗
75·32
SiO2
XRF (wt %)
RT34A
Group E
Table 1: continued
JOURNAL OF PETROLOGY
JANUARY 1997
141
4·29
0·26
Sc
Co
112
52
12·85
2·56
2·17
7·68
1·16
0·60
Ce
Nd
Sm
Eu
Tb
Yb
Lu
Eu/Eu∗
51·1
4·9
As
La
1·18
1·33
Ta
Sb
5·63
2·30
U
11·17
2·28
Cs
Hf
3·86
Na2O (wt %)
Th
2·00
FeO∗ (wt %)
INAA (p.p.m.)
RT34A
Group E
0·68
1·14
7·65
2·22
2·95
12·76
50
110
51·0
0·50
4·33
5·4
1·28
1·36
11·31
5·69
2·44
2·32
3·90
1·87
RT34B
0·65
1·08
7·51
2·16
2·71
12·26
51
106
49·0
0·59
4·83
4·5
1·34
1·35
11·19
6·12
2·35
2·33
3·71
2·09
RT34C
0·62
1·16
7·56
2·11
2·50
11·75
50
108
49·8
0·50
4·12
3·4
1·08
1·34
11·11
5·47
3·10
2·25
2·48
1·83
RT55A
0·59
1·25
7·99
2·12
2·67
14·16
65
126
53·5
0·22
5·02
4·3
1·16
1·37
10·76
5·33
2·41
2·44
2·42
2·00
RT140A
0·58
1·24
7·97
2·16
2·60
13·56
61
120
51·9
0·14
4·51
4·2
1·14
1·36
11·20
5·7
2·68
2·33
4·04
2·02
RT173A
0·56
1·23
7·91
2·13
2·48
13·39
58
112
51·7
0·13
4·32
4·0
1·17
1·35
11·07
5·67
2·18
2·35
4·50
1·91
RT173C
4·07
0·18
4·5 ± 0·3
0·3 ± 0·2
13·04
1·71
2·33
8·89
1·32
0·39
2·64 ± 0·16
2·15 ± 0·04
7·8 ± 0·2
1·18 ± 0·06
0·61 ± 0·04
48
55 ± 6
13·0 ± 0·8
102
113 ± 7
44·5
4·0
4·4 ± 0·6
51 ± 1
1·28
1·62
1·35 ± 0·01
1·19 ± 0·09
7·11
10·05
5·7 ± 0·2
2·5 ± 0·3
11·1 ± 0·2
2·96
3·50
2·33 ± 0·06
1·48
2·96
3·6 ± 0·8
RT17A
1·96 ± 0·09
Mean ± 1r
Banded
0·42
1·25
8·24
2·22
1·94
14·02
56
93
49·9
0·21
4·85
3·9
1·22
1·53
10·41
6·72
2·60
2·63
2·28
1·71
RT55C
0·52
1·19
7·91
2·06
2·26
13·41
45
126
50·4
2·19
4·18
3·5
1·19
1·41
10·63
5·64
2·70
2·43
2·79
1·82
RT127A
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Ta
Sb
142
51
33
10·58
0·77
2·43
11·26
1·66
0·20
Ce
Nd
Sm
Eu
Tb
Yb
Lu
Eu/Eu∗
0·19
1·50
0·18
1·34
8·83
1·92
0·56
8·54
28
46
17·8
0·17
4·52
4·2
1·48
2·58
8·08
10·80
7·39
4·01
2·30
0·88
AF1C
25·6
70
35
10·45
0·87
2·39
9·99
1·51
0·22
50 ± 4
31 ± 3
9·4 ± 1·1
0·66 ± 0·11
2·18 ± 0·26
9·9 ± 1
1·50 ± 0·16
0·19 ± 0·01
0·21
19 ± 2
4·36
1·75
2·59 ± 0·02
1·53 ± 0·07
4·6 ± 0·1
2·04
8·10 ± 0·04
0·22 ± 0·04
8·05
10·9 ± 0·3
4·1
9·16
6·7 ± 0·8
4·7 ± 1·2
3·34
4·1
4·1 ± 0·3
0·89
3·49
2·4 ± 0·4
0·28
1·42
9·57
2·51
1·31
13·25
53
111
45·0
0·17
3·76
4·0
1·26
1·66
9·36
7·39
3·7
3·11
3·41
1·16
1·39
6·85
2·6
2·97
3·34
0·35
1·36
9·02
2·32
1·66
14·70
62
122
51·8
0·31
4·07
4·7
1·83
1·46
10·15
Glass shard populations
0·88 ± 0·00
Mean ± 1r
1·49
0·37
1·35
8·84
2·39
1·76
14·5
67
135
55·6
0·31
4·10
3·59
2·59
1·55
10·45
7·02
3·2
2·95
3·32
1·79
0·53
1·26
8·36
2·17
2·44
14·23
71
133
54·4
0·51
4·64
2·1
1·41
1·51
11·30
6·13
2·4
3·01
3·40
1·79
0·54
1·25
8·17
2·07
2·42
13·83
68
128
55·8
0·31
4·31
4·5
1·25
1·32
11·15
5·63
2·7
2·87
3·64
0·34
1·38
9·26
2·26
1·45
12·4
48
94
40·1
0·23
4·15
3·7
1·32
1·68
9·02
7·39
3·38
3·14
3·51
1·22
Bulk tuff
NUMBER 1
9·58
2·19
0·65
8·99
32
53
19·0
0·25
4·71
3·8
1·50
2·61
8·15
11·20
5·73
4·45
2·11
0·88
AF1B
Shard matrix sample RT75
VOLUME 38
Rattlesnake Tuff pumice analyses are arranged in groups (A–E) according to clusters observed in scatter diagrams (Fig. 3), plus a group of small pumices (Ø ~1 cm)
from a thin precursor fallout deposit. Major element XRF data are normalized to 100%, volatile free; prenormalization (prn.) totals are listed. AI is molar ratio (Na+K)/
Al. Italicized numbers are values from INAA; values in parentheses are anomalously high and are not used. Sample AF1c is pumice composite of six small pumices.
Fe2O3∗ and FeO∗ are total iron concentrations expressed as Fe2O3 and FeO, respectively.
20·9
La
4·60
2·57
1·61
Hf
0·24
8·07
Th
Co
10·69
U
Sc
6·83
Cs
6·1
3·84
Na2O (wt %)
As
0·88
2·78
FeO∗ (wt %)
INAA (p.p.m.)
AF1A
Fallout
Table 1: continued
JOURNAL OF PETROLOGY
JANUARY 1997
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Table 2: Modal mineral abundances of
representative samples from Rattlesnake
Tuff pumice clusters derived by heavy
liquid mineral separation
Sample:
RT34E
RT165A
RT173B
RT173C
Pumice group:
B
C
D
E
alkali-fsp
0·025
0·736
0·874
0·522
quartz
0·032
0·536
0·022
0·002
cpx
0·007
0·028
0·055
0·062
magnetite
0·027
0·026
0·041
0·043
—
—
0·001
0·007
biotite
trace
—
trace
trace
total min. %
0·091
1·326
0·993
0·636
Glass
99·910
98·674
99·007
99·364
1022
752
1154
Minerals (wt %)
fayalite
total sample wt (g)
477
Accessory minerals
(× = present)
zircon
×
×
×
×
apatite
×
×
×
×
pyrrhotite
×
×
×
chevkinite
×
×
Pumice from cluster A (RT173H) had a total weight of 580 g,
but only eight crystals (feldspar and quartz) were retrieved,
making the pumice ‘aphyric’.
to gray and are sparsely phyric. Phases observed in all
phyric pumices are alkali feldspar, Fe-rich clinopyroxene,
titanomagnetite, quartz and accessory zircon and apatite.
Additional minerals that occur in some pumices are
fayalite, biotite, pyrrhotite, and chevkinite (Table 2).
A rough estimate of the volumetric proportions of the
different rhyolite groups can be made by using the
distribution of gray vs white shards in the matrix, which
roughly groups composition A and B and composition
D and E. This way, a 1:1 proportion is derived which
seems to be a maximum value for the ‘gray magma’ as
white shards dominate distally.
ANALYSIS AND MINERAL
PREPARATION
Mainly unbanded, glassy pumices were selected for analysis. Other analyzed glassy samples included three macroscopically banded pumices and one bulk tuff and its
different shard populations (Table 1). Sampling was done
143
to be representative for regional, local and stratigraphic
chemical variations.
Major-element analyses were made on fused glass disks,
with 5:1 flux to rock powder, and selected trace elements
on pressed powder pellets by X-ray fluorescence (XRF)
at Stanford University using a Rigaku instrument, except
for sample RT220A, which was analyzed at Washington
State University, Washington. Trace element concentrations were determined by instrumental neutron
activation analysis (INAA) at the Radiation Center, Oregon State University, using a 1 MW Triga reactor.
Analytical uncertainties for XRF trace elements, based
on replicate analyses of US Geological Survey (USGS)
standards G2 and AGV1, are: <5% for Nb, Zr, Y, Sr,
Rb, Ga and Zn; <10% for Cu, V and Ba; and <5–15%
for Ni, Cr and Pb. Uncertainties for INAA trace elements
(also based on replicate analyses of in-house standards,
CRBIV and SPGa) are: <5% for Fe, Na, Co, Eu, Hf,
La, Sc, Sm and Yb; <5–10% for Ce, Cr, Lu, Ta, Tb
and Th; <5–15% for Ba, Cs, Nd, Rb and Zn; and <15%
for Ni and U. The range of uncertainties for single
elements is based on the concentration range observed
in standards used as monitors.
Mineral separates were obtained from five representative pumices. After crushing with hammer and
jaw crusher, pumices were sieved. Sieving was done
sequentially with the following mesh size: 2 mm, 1·7
mm, 990 lm, 750 lm, 500 lm, 300 lm, 180 lm, and
106 lm. After first sieving, all material >990 lm was
further crushed in small proportions until all pumice
material passed through 990 lm mesh. Material for glass
separates was handpicked before heavy liquid separation
to insure uncontaminated Br values. Using bromoform
and tetrabromoethane, sequential heavy liquid mineral
separations were done for each size fraction. First glass
was floated, keeping the density of the liquid close to
that of the glass to insure that minerals with attached
glass also sank. The separated glass was weighed. Next,
each size fraction was either separated into individual
phases or into a feldspar–quartz and ‘mafic’ fraction
through handpicking, magnetic or heavy liquid separation. The proportion of fsp:qtz and of timt:cpx was
sometimes visually estimated and recalculated to weight
per cent. The total mode of the pumice (Table 2) represents the sum of the weights of different species in all
size fractions. The agreement between weights of total
mineral yields before and after separation into individual
phases deviated between 1 and 10%, with lower yields
in the separated minerals partly attributable to dissolution
of glass rinds during washing with dilute hydrofluoric
acid after first determination of total mineral yields.
Extrapolation of mineral size distribution indicates that
loss of minerals in the <106 lm fraction is minor.
Mineral separates for INAA analysis were prepared by
hand-picking from the 300–500 lm fraction for feldspars,
JOURNAL OF PETROLOGY
VOLUME 38
106–300 lm for titanomagnetites, and mainly 106–180
lm fraction for pyroxenes. Separates were not powdered.
All selected feldspars were clear and inclusion free. Titanomagnetite separates were superficially 100% clean. In
the case of the pyroxenes, the selected size fraction was
small enough so that most pyroxenes were translucent,
making screening for inclusions feasible. All pyroxene
separates were visually clean at >99%. Mineral separates
were multiply washed in mild acids, distilled water, and
acetone. Br values of <6 ppm for all mineral separates
verify the almost complete removal of heavy liquid residues from mineral surfaces (see Table 4, below). Glass
separates were picked from the 300–500 lm fraction and
separates were completely clean of crystals, except sample
RT165A where very tiny (±10 lm) Fe-oxide(?) crystals
were sparsely but evenly dispersed.
Mineral and glass separates were analysed by INAA
at the Oregon State University Radiation Center using
a 1 MW Triga reactor. Weights of analyzed samples
ranged from 10 to 70 mg. Short activation was performed
at a power level of 50 kW for 5 min and long activation
at 1 MW for 12 h. Counting was done sequentially, three
times after short activation and five times after long
activation using intrinsic germanium and low-energy
photon (LEP) detectors.
One non-welded, glassy bulk-tuff sample was used
for separation of different shard populations. Magnetic
procedures (Frantz magnetic divider) separated loosened
glass shards of the 180–500 lm fraction (weighing ~100
g) into a white shard and a mixed-gray shard fraction.
Using heavy liquids, the white shard fraction was split
into a lighter and heavier fraction, yielding materials for
the first two shard samples analyzed by INAA. Similarly,
splitting the mixed-gray fraction into successively denser
fractions yielded seven shard fractions from which four
were selected for analysis. Therefore, the different shard
populations observed in the matrix of the tuff are thought
to be only approximately represented by the six analyzed
bulk shard samples because of imperfect separation.
Shard separates used for analysis were free of crystals
and lithic fragments. INAA was performed after long
activation at 1 MW for 6 h with corresponding counting
procedures (see above).
Microprobe analyses on minerals were done using a
fully automated Cameca SX-50 electron microprobe at
Oregon State University. For most minerals, beam current was 30–50 nA, accelerating voltage 15 kV and beam
diameter 1–5 lm.
NUMBER 1
JANUARY 1997
systematically with silica. Elements that increase with
silica, that is, those that are enriched compared with the
least evolved compositions (Group E), are Cs, Rb, U,
Th, Ta, Nb, Pb, Y, HREE (heavy rare earth elements),
Sb, and probably Ni (Fig. 4). Depleted elements are Fe,
Ti, Mg, Ca, Ba, Sr, Eu, Zr, Hf, LREE (light rare
earth elements), and Zn. Ga, As, V, Cr, Co, and Mn
concentrations are nearly constant; precise Mn data
(INAA) exist only for pumice glass separates and suggest
a minimum for Group B pumices. Similarly, a minimum
is suggested by Sc and Al and a maximum by Tb.
Na and K concentrations become more variable from
Group A to Group E rhyolites (Fig. 3), whereas total
alkalis (Na2O+K2O) are nearly constant, with an average
of 9·1 wt %. Early post-emplacement ion exchange is
likely to have caused most of this scattering, by increasing
K and reducing Na contents (Fisher & Schmincke, 1984,
p. 328). The samples richest in sodium are presumably
closest to the magmatic concentration. With this assumption, Na2O slightly decreases by ~0·6 wt % and
K2O increases by the same amount from Group E to
Group A. The peralkalinity index [AI = molar (Na+K)/
Al] of the most sodic sample of each group ranges from
0·98 to 1·03, suggesting almost constant alkali–aluminum
balance throughout the compositional range. Alkali mobility did not affect other ‘mobile’ elements (Zielinski,
1982) because neither Na nor K correlate with Cs, Rb,
and U (Fig. 3).
The observed enrichment and depletions are nearly
identical with trends in the Lava Creek and Huckleberry
Ridge Tuffs of the Yellowstone caldera complex (Hildreth
et al., 1984) and similar to those of the metaluminous
Bishop Tuff (Hildreth, 1979). For enrichment trends of
36 elements, the differences between the Rattlesnake and
Bishop Tuffs are mainly enrichment of Na, Mn, Sc, and
Sm, and constant Hf and Zn compared with depletions
of these elements in the Rattlesnake Tuff (Fig. 4). Cl
decreases from Group E to A, probably reflecting degassing during eruption. Comparison of enrichment
trends from Group B to A with the Bishop Tuff reduces
the degree of discrepancy between the two. The slightly
peralkaline Tala Tuff (Mahood, 1981) with constant Si,
Fe, Mg, and Eu, and enriched Zr, Hf, Zn, Sm, and Tb
differs strongly from the Rattlesnake Tuff.
COMPOSITIONAL VARIATION
The Rattlesnake Tuff high-silica rhyolite pumices range
in composition from 74·5 to 77·8 wt % SiO2 (Table 1,
Fig. 3). Despite this narrow range, many elements vary
144
COMPOSITIONAL CLUSTERS
High-silica rhyolite compositions of the Rattlesnake Tuff
bulk pumices cluster in several Groups (A–E). The clusters
are mainly established by La(LREE), Eu, Ba, Ta, Nb,
Zr, Hf, Rb, Cs, Th, U, Ti, and Fe. Pumices within a
cluster are macroscopically the same. The relative position among groups and of individual pumices within
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Fig. 3. Major and trace element variation diagrams for Rattlesnake Tuff pumices and shards. Cs–K2O and U–Na2O diagrams show that
probably limited ion-exchange of Na with K did not affect significantly even mobile elements because clustering of pumices in terms of Cs and
U contents is mainly intact; sh, shard separate.
145
JOURNAL OF PETROLOGY
VOLUME 38
Fig. 3.
146
NUMBER 1
JANUARY 1997
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Similar to the compositional spread between groups, the
differences in the mineral chemistry become wider with
higher degree of differentiation.
MINERALOGY
The mode and mineral chemistry of the high-silica rhyolite groups change progressively (Tables 2 and 3, Figs
5–7). Group A pumice is aphyric and Group B has 0·1
wt % crystals. An abrupt increase in crystal content to 1·3
wt % characterizes Group C. Crystal content decreases to
0·6 wt % in Group E. It is uncertain whether the reversal
in crystal content is an artifact of pumice sampling. The
main modal characteristics with increasing differentiation
are: (1) an increase in the proportion of quartz relative
to feldspar; (2) fairly constant mafic mineral concentration; (3) restriction of fayalite to Groups D and E.
Several flakes of biotite were found in Groups B, D, and
E.
Accessory phases occur only as inclusions in, or adhered
to, major phases; free accessory phases were not observed
even in the smallest size fraction, except for sparse zircons
in the 106–180 lm fraction of Group B. Clinopyroxene,
fayalite, and titanomagnetite are the main hosts for
inclusions, whereas inclusions in feldspar are rare. Crystallization of accessory phases seems to be controlled by
local saturation around mafic phases (see Bacon, 1989).
Accessory phases are zircon and apatite in all groups,
pyrrhotite in Groups C, D, and E, and chevkinite (see
Michael, 1988) in Groups C and E (three grains, altogether). A few small grains of high density or magnetic
phases (sulfides?) were also found. The estimated proportions of zircon to apatite to chevkinite are 10:3:1,
respectively. The absolute concentration of zircon is
estimated to be 0·0003 wt % in Group B, 0·0006 in
Group C and 0·001 in Groups D and E by assuming
that zircon constitutes 1 wt % of the mafic assemblage.
These are upper limits based on a petrographic estimate
of 0·93 vol. % zircon in a clinopyroxene with abundant
zircon and using a density of 4 g/cm3 for bulk mafic
minerals. It follows that zirconium is hosted nearly entirely in the glass.
Fig. 4. Enrichment factors showing enrichment of the average composition of individual pumice clusters relative to the average of least
evolved composition rhyolite Group E (Table 1).
clusters is consistent in most variation diagrams. For most
elements, one standard deviation of the mean of each
group (Table 1) is in the range of the analytical uncertainty, which in some cases is also the difference
between adjacent groups, causing some overlap. Group
C has more internal scatter than the other groups.
White pumice from a basal fallout deposit is similar in
composition to Group A, but has somewhat higher U,
Th, and Ta concentrations (Fig. 3, Table 1). The fallout
is considered precursory to the ignimbrite and probably
represents a slightly more evolved magma composition
than Group A.
The compositional spread of the pumice populations
is similar to that defined by the shard populations (Fig.
3, Table 1). The most evolved shard population, as
indicated by lowest FeO and Eu, and highest Ta, is white
and corresponds in composition to Group A and the
fallout, but is displaced toward the rest of the tuff,
suggesting that shard separation was imperfect. A second
shard population is also white and corresponds well to
Group B compositions. The other four shard separates
are gray; the first two are similar to Group C, which has
considerable range. The two least evolved populations
are similar to Group E, but are displaced consistently
towards Group D, indicating that the population is mixed.
Microprobe analyses of individual glass shards establish
a positive correlation between Fe concentration and
darker color (Fig. 2).
The clustering of the Groups B–E is also observed
in the changing mineral chemistry (Tables 3 and 4).
Titanomagnetites become progressively poorer in Ti,
clinopyroxenes and fayalites more magnesian, alkali feldspars more potassic, and zircons enriched in Hf with
differentiation from E towards A (Tables 3 and 4).
Felsic minerals
Feldspar is anorthoclase in Groups E and D and changes
to Na-sanidine in Groups C and B (Fig. 5). A few euhedral
microcrystals of oligoclase (Ab69–70) were found in Group
A. The compositional range of individual feldspars is
commonly <1 Ab unit, except in some feldspars from
one banded pumice which are anorthoclase (Ab66) with a
more sodic rim (Ab75). The range of feldspar compositions
becomes progressively tighter, with 6 Ab units in Group
C rhyolite and a 2 Ab unit range in Group B rhyolite
147
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
Table 3: Representative microprobe analyses of minerals
Feldspar
Clinopyroxene
Group:
A
C
D
E
B
C
D
Sample:
RT173H RT34E
B
RT165A
RT173B
RT55A
RT34E
RT165A
RT173B
RT173C
rim
ctr
hlf
rim
rim
hlf
ctr
hlf
hlf
49·1
E
SiO2
61·56
65·52
65·71
65·07
64·79
SiO2
49·57
47·84
48·16
Al2O3
23·67
19·27
19·30
20·12
20·5
TiO2
0·17
0·18
0·26
0·23
FeO
0·27
0·28
0·25
0·20
0·33
Al2O3
0·37
0·34
0·37
0·33
BaO
0·22
0·27
0·70
1·40
1·44
FeO∗
20·44
24·13
27·75
26·98
CaO
4·91
0·18
0·28
0·63
1·19
MnO
3·03
2·74
2·4
2·34
Na2O
7·96
5·89
6·24
7·52
8·02
MgO
6·27
4·43
2·37
2·86
K 2O
1·02
7·91
7·08
4·66
3·22
CaO
18·51
18·28
17·59
17·84
Na2O
0·56
0·53
0·54
Total
99·62
99·32
99·56
99·59
99·49
Total
98·92
99·73
99·12
99·24
41·2
0·5
Or
5·9
46·5
42·2
28·1
19·6
Wo
41·7
41·5
41·0
Ab
70·1
52·6
56·4
68·8
74·3
En
19·7
14·0
7·7
9·2
An
24·0
0·9
1·4
3·2
6·1
Fs
38·6
44·5
51·3
49·6
Fs range
37·8–40·3 42–48·5
44·5–52
48–52·8
Ab range 69–70†
51·5–52·8 54–60
67·6–73·7 64·4–75
Titanomagnetite
Fayalite
Biotite
HSR grp:
B
C
D
E
D
E
B
D
E
Sample:
RT34E
RT165A
RT173B
RT173C
RT173B
RT173C
RT34E
RT173B
RT173C
ctr
ctr
hlf
ctr
hlf
hlf
35·57
SiO2
0·07
0·10
0·09
0·10
SiO2
29·85
29·35
SiO2
35·35
37·58
TiO2
12·04
15·51
18·87
20·34
TiO2
0·03
0·04
TiO2
2·40
2·47
2·53
Al2O3
0·47
0·43
0·48
0·56
Al2O3
0·01
0·02
Al2O3
13·40
13·95
13·79
V2O5
0·08
0·11
0·08
0·12
FeO∗
63·18
63·74
FeO∗
25·01
17·02
25·27
Fe2O3
44·82
38·65
31·43
28·45
MnO
5·59
4·85
MnO
0·25
0·59
0·24
FeO
39·71
43·39
46·51
47·92
MgO
1·50
1·28
MgO
7·91
13·25
8·18
MnO
2·07
1·93
1·76
1·72
CaO
0·29
0·35
CaO
n.d.
0·01
n.d.
MgO
0·20
0·15
0·07
0·07
NiO
n.d.
0·03
Na2O
0·37
0·07
0·13
ZnO
0·47
0·42
0·38
0·35
ZnO
0·20
0·16
K2O
9·02
9·59
9·55
H 2O
3·04
3·39
2·85
F
1·45
1·13
1·96
Cl
Total
99·93
100·69
99·67
99·63
Total
100·65
99·82
XUSP
39·6
44·4
54·2
58·4
Fo
3·7
3·2
Fa
88·4
90·0
Te
7·9
6·8
Total
O = F, Cl
TiO2
11·8–12·6 15·2–15·7 18·6–19·2 18–21·4
Fo range
3·7–3·8
range
3·1–3·2
0·02
0·04
99·07
100·11
0·63
0·48
0·84
Total
97·62
98·59
99·27
ZnO‡
0·16
0·16
0·05
BaO‡
0·02
0·63
0·04
MgO
range
148
0·05
98·25
—
12·4–13·2
7·9–8·2
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Table 3: continued
Zircon
HSR grp:
B
C
D
E
Sample:
RT34E
RT165A
RT173B
RT173C
n
9
13
8
13
SiO2
32·28 ± 0·21
31·89 ± 0·35
32·14 ± 0·30
CaO
0·01 ± 0·01
0·02 ± 0·04
0·02 ± 0·02
32·04 ± 0·27
0·06 ± 0·03
TiO2
0·05 ± 0·05
0·11 ± 0·08
0·10 ± 0·1
0·13 ± 0·11
Y2O3
0·32 ± 0·34
0·31 ± 0·53
0·06 ± 0·18
0·01 ± 0·03
ZrO2
65·74 ± 0·68
66·32 ± 0·88
66·57 ± 0·33
66·69 ± 0·31
Gd2O3
0·01 ± 0·01
0·03 ± 0·03
0·01 ± 0·01
0·03 ± 0·03
Yb2O3
0·19 ± 0·12
0·16 ± 0·16
0·04 ± 0·07
0·09 ± 0·05
HfO2
1·36 ± 0·29
1·08 ± 0·16
1·00 ± 0·22
0·89 ± 0·07
ThO2
0·04 ± 0·03
0·08 ± 0·11
0·06 ± 0·07
0·05 ± 0·06
Apatite
HSR grp:
B
C
D
D
Sample:
RT34E
RT165A
RT173B
RT173B
Host:
zircon
magnetite
magnetite
biotite
SiO2
3·29
4·18
3·99
0·39
TiO2
<0·00
0·23
0·22
0·05
Al2O3
<0·00
<0·00
0·27
0·03
FeO
0·51
2·44
1·60
0·60
CaO
48·62
46·68
49·38
55·39
Ce2O3
2·28
3·58
2·22
0·02
Sm2O3
0·70
0·79
0·43
<0·00
Yb2O3
0·19
0·11
<0·00
<0·00
Y2O3
1·96
1·87
0·87
<0·00
P2O5
36·56
33·67
36·70
41·50
F
3·81
2·61
2·87
3·57
Cl
0·21
0·22
0·22
<0·00
total
98·13
96·38
98·77
101·55
O=F,
1·65
1·15
1·26
1·50
96·47
95·23
97·52
100·05
Cl
total
Letters indicate high-silica rhyolite (HSR) pumice group. Analyses at rim, center, and between are indicated by rim, ctr, and
hlf, respectively. †Range is for three andesine grains. ‡Values are poorly constrained. Fe2O3 is calculated by charge balance
and Xusp is mole fraction ulvöspinel. Or, KAlSi3O8; Ab, NaAlSi3O8; An, CaAl2Si2O8; Wo, Ca2Si2O6; En, Mg2Si2O6; Fs, (Fe,Mn)2Si2O6;
Fo, Mg2SiO4; Fa, Fe2SiO4; Te, Mn2SiO4. For zircons, n indicates number of averaged analyses and errors are 1r SDs; analyses
were normalized to 100% before calculating averages and SDs. host, mineral by which analyzed apatite was enclosed.
(Fig. 5). The general decrease in Na and increase in K
is confirmed by the bulk feldspar analyses (Table 4). Ca
and Ba concentrations progressively decrease from Group
E to B, with BaO concentrations commonly twice those
of CaO. Quartz is bipyramidal and crystal faces indicate
only incipient resorption.
149
Mafic minerals
Pyroxenes in the high-silica rhyolites range in composition
from Fe-hedenbergite to Fe-augite (Table 3, Fig. 6).
Individual grains are unzoned with ranges typically <1%
ferrosilite component and, although there is some overlap
between each adjacent group, pyroxenes become
RT173H
A
RT34E
B
Oxides (wt %), MnO (p.p.m.)
TiO2
0·17 ± 0·03
0·12 ± 0·04
Al2O3
12·54 ± 0·05
12·46 ± 0·05
FeO∗
0·70 ± 0·01
1·22 ± 0·01
MnO
800 ± 2
780 ± 2
MgO
n.d.
n.d.
CaO
n.d.
0·40 ± 0·05
Na2O
3·56 ± 0·01
3·21 ± 0·01
K2O
4·8 ± 0·1
5·2 ± 0·1
Trace elements (p.p.m.)
Cs
4·25 ± 0·06
3·24 ± 0·05
Rb
115 ± 2
89 ± 2
Th
9·01 ± 0·05
7·53 ± 0·05
U
4·6 ± 0·1
3·5 ± 0·2
Sr
n.d.
n.d.
Ba
49 ± 8
167 ± 23
La
19·7 ± 0·1
38·3 ± 0·2
Ce
52 ± 0·4
97·8 ± 0·5
Nd
29 ± 1
51 ± 2
Sm
9·16 ± 0·01
13·40 ± 0·01
Eu
0·62 ± 0·01
1·22 ± 0·02
Tb
2·25 ± 0·02
2·71 ± 0·02
Dy
15·0 ± 0·4
15·8 ± 0·4
Yb
10·10 ± 0·05
9·53 ± 0·06
Lu
1·54 ± 0·01
1·45 ± 0·01
Sc
3·75 ± 0·01
3·60 ± 0·01
Zr
170 ± 17
310 ± 16
Hf
6·63 ± 0·05
9·31 ± 0·06
Ta
2·05 ± 0·03
1·75 ± 0·03
Cr
4·9 ± 0·5
4·0 ± 0·5
Co
0·12 ± 0·02
0·45 ± 0·02
Zn
87 ± 8
108 ± 7
W
1·9 ± 0·3
1·4 ± 0·4
As
4·6 ± 0·3
4·3 ± 0·4
Sb
1·48 ± 0·05
1·38 ± 0·05
Cl
483 ± 72
586 ± 78
Br
3·0 ± 0·3
3·1 ± 0·3
Sample:
Group:
Bulk glass
3·88 ± 0·01
4·3 ± 0·1
2·35 ± 0·05
62 ± 2
5·97 ± 0·05
2·8 ± 0·2
n.d.
1120 ± 29
57·2 ± 0·2
119 ± 0·4
63 ± 2
14·30 ± 0·01
1·93 ± 0·02
2·24 ± 0·02
13·3 ± 0·3
7·93 ± 0·06
1·23 ± 0·01
3·44 ± 0·01
389 ± 16
10·40 ± 0·05
1·30 ± 0·02
4·6 ± 0·5
0·12 ± 0·01
108 ± 6
2·4 ± 0·3
4·0 ± 0·3
1·25 ± 0·05
512 ± 82
3·7 ± 0·3
3·74 ± 0·01
2·75 ± 0·06
74 ± 2
6·78 ± 0·05
3·0 ± 0·1
n.d.
357 ± 21
48·2 ± 0·1
113 ± 0·5
58 ± 3
14·00 ± 0·01
1·34 ± 0·02
2·42 ± 0·02
14·9 ± 0·3
8·5 ± 0·05
1·28 ± 0·01
3·32 ± 0·01
304 ± 7
9·55 ± 0·06
1·46 ± 0·03
4·4 ± 0·5
0·18 ± 0·02
97 ± 7
2·0 ± 0·3
4·5 ± 0·3
1·31 ± 0·05
1103 ± 73
5·4 ± 0·3
12·23 ± 0·05
1·65 ± 0·01
850 ± 3
0·06 ± 0·006
0·43 ± 0·05
4·8 ± 0·1
0·17 ± 0·4
12·08 ± 0·05
1·31 ± 0·01
790 ± 2
0·38 ± 0·04
0·50 ± 0·05
RT173B
D
0·14 ± 0·03
RT165A
C
150
n.d.
(2·6 ± 0·6)
0·07 ± 0·02
24·8 ± 0·9
0·20 ± 0·02
n.d.
n.d.
5680 ± 34
4·86 ± 0·08
7·7 ± 0·2
n.d.
0·27 ± 0·01
6·22 ± 0·03
0·052 ± 0·010
n.d.
0·22 ± 0·06
0·030 ± 0·006
0·098 ± 0·002
n.d.
0·40 ± 0·02
0·05 ± 0·01
2·1 ± 0·3
0·09 ± 0·02
<10
1·9 ± 0·6
<3·3
0·11 ± 0·01
7·5 ± 0·1
6·36 ± 0·01
20·19 ± 0·08
0·24 ± 0·003
18 ± 0·5
n.d.
n.d.
n.d.
RT165A
C
n.d.
(3·6 ± 1·0)
0·14 ± 0·04
12·3 ± 0·8
<0·3
n.d.
162 ± 13
11000 ± 44
3·98 ± 0·09
4·4 ± 0·2
n.d.
0·22 ± 0·01
9·60 ± 0·04
0·061 ± 0·015
n.d.
0·18 ± 0·03
<0·045
0·049 ± 0·003
n.d.
0·48 ± 0·03
0·06 ± 0·02
5·9 ± 0·4
0·17 ± 0·02
<7·5
5·5 ± 0·6
<4·2
0·11 ± 0·02
5·1 ± 0·1
8·12 ± 0·01
21·30 ± 0·09
0·23 ± 0·004
6·5 ± 0·7
n.d.
n.d.
n.d.
RT173B
D
n.d.
(2·6 ± 0·4)
0·20 ± 0·04
8·3 ± 0·9
<0·2
n.d.
246 ± 16
12300 ± 37
4·00 ± 0·06
3·9 ± 0·2
n.d.
0·12 ± 0·008
12·20 ± 0·05
<0·14
n.d.
0·09 ± 0·02
0·016 ± 0·005
0·051 ± 0·003
n.d.
0·29 ± 0·06
0·06 ± 0·02
7·0 ± 0·5
0·12 ± 0·02
<13
<3·6
<3·6
0·15 ± 0·02
4·1 ± 0·1
8·76 ± 0·01
21·17 ± 0·11
0·23 ± 0·003
8·7 ± 0·8
n.d.
1·09 ± 0·12
n.d.
RT173C
E
NUMBER 1
n.d.
(5·7 ± 0·5)
0·13 ± 0·04
30 ± 1
<0·3
n.d.
n.d.
2580 ± 41
6·5 ± 0·1
10·2 ± 0·2
n.d.
0·11 ± 0·01
4·58 ± 0·04
0·066 ± 0·019
n.d.
0·18 ± 0·05
0·018 ± 0·004
0·058 ± 0·003
n.d.
0·27 ± 0·03
0·11 ± 0·04
9·4 ± 0·4
0·39 ± 0·03
<15
3·3 ± 0·6
<4·2
0·31 ± 0·03
7·9 ± 0·2
5·82 ± 0·01
20·0 ± 1·6
0·24 ± 0·005
10 ± 1
0·14 ± 0·02
n.d.
n.d.
RT34E
B
Bulk feldspar
VOLUME 38
1246 ± 92
4·0 ± 0·3
2·37 ± 0·05
66 ± 2
5·74 ± 0·05
2·44 ± 0·15
n.d.
1640 ± 28
53·7 ± 0·2
123·0 ± 0·5
59 ± 2
13·80 ± 0·01
2·17 ± 0·02
2·20 ± 0·02
12·5 ± 0·3
7·75 ± 0·07
1·19 ± 0·01
3·73 ± 0·01
424 ± 16
10·40 ± 0·05
1·32 ± 0·02
6·3 ± 0·5
0·16 ± 0·02
101 ± 6
1·3 ± 0·3
3·9 ± 0·3
1·17 ± 0·04
3·9 ± 0·1
4·30 ± 0·01
12·42 ± 0·05
1·77 ± 0·01
860 ± 3
0·05 ± 0·005
0·62 ± 0·06
0·14 ± 0·04
RT173C
E
Table 4: Instrumental neutron activation analyses of bulk minerals and glass
JOURNAL OF PETROLOGY
JANUARY 1997
RT34E
B
151
n.d.
0·89 ± 0·08
26·40 ± 0·05
2·45 ± 0·01
2·3 ± 0·6
19·2 ± 0·5
5000 ± 20
n.d.
n.d.
n.d.
<2
<6·6
n.d.
n.d.
51·5 ± 0·2
186 ± 2
227 ± 5
74·6 ± 0·1
7·38 ± 0·08
8·72 ± 0·09
69 ± 3
37·4 ± 0·4
6·99 ± 0·06
385 ± 0·01
586 ± 120
19·6 ± 0·4
0·38 ± 0·08
19 ± 6
3·1 ± 0·2
n.d.
994 ± 52
12 ± 1
<6·6
<1·6
(4·8 ± 0·6)
n.d.
n.d.
18 ± 3
<8
n.d.
n.d.
534 ± 0·5
1210 ± 4
667 ± 12
146 ± 0·2
10·4 ± 0·1
15·6 ± 0·2
97 ± 4
52·2 ± 0·4
9·07 ± 0·05
467 ± 0·01
1300 ± 160
34·3 ± 0·6
0·82 ± 0·14
47 ± 7
6·6 ± 0·3
n.d.
1170 ± 87
23 ± 2
<4·8
<3·3
(4·7 ± 1)
RT173B
D
n.d.
0·85 ± 0·09
24·50 ± 0·07
2·70 ± 0·01
5·5 ± 1·1
17·2 ± 0·7
6000 ± 20
n.d.
RT165A
C
n.d.
n.d.
<3·3
<8·4
n.d.
n.d.
48·0 ± 0·2
190 ± 3
192 ± 7
66·2 ± 0·1
7·78 ± 0·12
8·41 ± 0·14
63 ± 5
36·8 ± 0·3
6·43 ± 0·06
361 ± 0·01
810 ± 120
15·0 ± 0·5
0·60 ± 0·19
42 ± 7
4·3 ± 0·2
n.d.
1000 ± 81
92 ± 2
<8·1
<1·6
(4·8 ± 0·6)
n.d.
0·92 ± 0·08
28·80 ± 0·09
2·21 ± 0·01
2·3 ± 0·4
18·4 ± 0·7
4800 ± 20
n.d.
RT173C
E
n.d.
n.d.
0·62 ± 0·17
<1·8
n.d.
n.d.
8·49 ± 0·06
19·3 ± 0·7
7·9 ± 1·6
1·52 ± 0·01
0·16 ± 0·02
0·23 ± 0·05
n.d.
0·88 ± 0·08
0·15 ± 0·02
23·30 ± 0·05
272 ± 65
7·0 ± 0·2
2·45 ± 0·09
19 ± 2
5·8 ± 0·1
76 ± 6
3520 ± 32
<1·6
<0·9
0·36 ± 0·09
(2·2 ± 0·3)
11·0 ± 0·3
0·66 ± 0·06
80·4 ± 0·1
2·02 ± 0·01
n.d.
<0·19
62 ± 1
n.d.
RT34E
B
n.d.
n.d.
2·7 ± 0·2
2·9 ± 0·2
n.d.
n.d.
47 ± 0·1
116 ± 1
79 ± 3
20·50 ± 0·02
1·33 ± 0·03
2·66 ± 0·08
16 ± 3
10·9 ± 0·1
1·81 ± 0·02
24·50 ± 0·05
2570 ± 80
77·2 ± 0·2
3·12 ± 0·09
23 ± 2
5·57 ± 0·09
78 ± 6
3180 ± 29
1·3 ± 0·2
<1·4
0·28 ± 0·07
(2·1 ± 0·3)
14·1 ± 0·2
0·65 ± 0·05
78 ± 0·1
1·89 ± 0·01
n.d.
<0·21
155 ± 2
n.d.
RT165A
C
Bulk titanomagnetite
n.d.
n.d.
0·74 ± 0·15
<2·4
n.d.
n.d.
19·9 ± 0·1
51·4 ± 0·9
37 ± 7
9·31 ± 0·02
0·88 ± 0·04
1·33 ± 0·09
n.d.
2·29 ± 0·09
0·30 ± 0·02
25·70 ± 0·05
446 ± 71
13·7 ± 0·3
4·68 ± 0·13
23 ± 2
6·0 ± 0·1
73 ± 9
3130 ± 38
1·3 ± 0·2
<1·8
0·73 ± 0·1
(3·1 ± 0·3)
17·0 ± 0·4
0·87 ± 0·09
76·5 ± 0·2
1·74 ± 0·01
n.d.
<0·29
177 ± 2
n.d.
RT173B
D
n.d.
n.d.
1·46 ± 0·13
1·8 ± 0·1
n.d.
n.d.
27·7 ± 0·1
69·1 ± 0·7
62 ± 2
14·00 ± 0·01
1·48 ± 0·03
1·94 ± 0·06
n.d.
8·62 ± 0·08
1·47 ± 0·02
26·80 ± 0·03
3210 ± 22
93·8 ± 0·2
4·90 ± 0·07
20 ± 2
5·34 ± 0·07
80 ± 4
3150 ± 22
0·79 ± 0·16
<1·1
0·25 ± 0·06
(0·8 ± 0·2)
19·4 ± 0·2
0·84 ± 0·04
72·2 ± 0·1
1·59 ± 0·01
n.d.
<0·05
97 ± 1
n.d.
RT173C
E
Analytical error given is a measure for the ‘distinctness’ of the analyzed peak. <, maximum possible concentration for the element which was undetected at a
confidence level of 3r; this is therefore an upper limit and the actual concentration may be less. n.d., not detected. Bromine values in parentheses are reported
to indicate removal of almost all Br-containing heavy liquid residues from mineral surfaces. Feldspar samples consisted of ~70–100 feldspar grains in samples
RT173C, -173B, and -165A, and ~40 grains in sample RT34E. Analyzed clinopyroxene separates consisted of ~100–300 grains. Each titanomagnetite separate
consisted of >150 grains.
Oxides (wt %), Na2O (p.p.m.)
TiO2
n.d.
Al2O3
0·73 ± 0·08
FeO∗
21·30 ± 0·06
MnO
2·99 ± 0·01
MgO
8·0 ± 1·2
CaO
18·8 ± 0·8
Na2O
6100 ± 20
K2O
n.d.
Trace elements (p.p.m.)
Cs
n.d.
Rb
n.d.
Th
<3
U
<14
Sr
n.d.
Ba
n.d.
La
36·4 ± 0·3
Ce
152 ± 4
Nd
215 ± 9
Sm
81·6 ± 0·1
Eu
6·08 ± 0·09
Tb
13·2 ± 0·2
Dy
99 ± 4
Yb
47·1 ± 0·6
Lu
8·78 ± 0·08
Sc
496 ± 0·01
Zr
474 ± 120
Hf
8·8 ± 0·5
Ta
4·9 ± 0·2
Cr
58 ± 8
Co
4·5 ± 0·3
V
n.d.
Zn
1240 ± 90
W
1170 ± 2
As
<4·8
Sb
<1·8
Br
(2·4 ± 0·6)
Sample:
Group:
Bulk clinopyroxene
Table 4: continued
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
Fig. 7. Titanomagnetite compositions represented as molar proportions on the FeO–Fe2O3–TiO2 ternary plot. End members of
magnetite series are: Usp, ulvöspinel; Mag, magnetite. Capital letters
indicate high-silica rhyolite groups. Bracket encloses range of Group
D (RT173B). Open symbols indicate mineral separate analyses.
Fig. 5. Rattlesnake Tuff feldspar compositions represented on the
feldspar ternary diagram. Each solid symbol represents one analysis.
Capital letters indicate high-silica rhyolite group. Tie-line to hypothetical sanidine composition corresponding to plagioclases Ab70 in
sample RT173H was drawn with information from Fuhrman & Lindsley
(1988) at 1 kbar and 750°C. Ternary at bottom shows all alkali feldspars
in relation to the single feldspar field at 1 kbar, 750°C and 825°C from
Fuhrman & Lindsley (1988).
progressively poorer in Fe, and more Mn rich, with
higher degree of evolution of the liquid. INAA data
for pyroxene separates are in excellent agreement with
microprobe data, particularly for Mn (Table 4).
Fayalites occur only in the two least evolved rhyolite
compositions. The compositional range of individual
phenocrysts and within pumice clasts is no more than
0·1 Fo units (Table 3). As for pyroxenes, Mn and Mg
increase with degree of evolution, as indicated by slightly
higher concentrations of MgO and MnO in Group D
vs E (Fig. 6).
Titanomagnetite is the only Fe–Ti oxide found in
Rattlesnake Tuff high-silica rhyolites. Titanomagnetite
compositions become progressively poorer in Ti from
Group E pumice to Group B pumice (Tables 3 and 4,
Fig. 7). Titanomagnetite compositions overlap those for
Groups E and D and are distinct for Groups C and B.
Fig. 6. Clinopyroxene and fayalite variation diagrams and pyroxene compositions projected onto the pyroxene quadrilateral and olivine
compositions projected onto the Fo–Fa tie-line; each symbol represents one analysis except crosses on ternary projection indicate analyses of
mineral separates. Capital letters indicate high-silica rhyolite groups.
152
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Mineral separate analyses correspond well except that
TiO2 is systematically about 1 wt % lower.
Biotite in trace amounts was found in Groups B, D,
and E (Table 3). Arguments favoring a phenocrystic
magmatic origin are the relative large size (~0·5–2 mm)
and the unaltered appearance of the flakes. On the other
hand, the flakes have anhedral crystal form and euhedral
apatite inclusions in the biotite have more SiO2, FeO
and REE, and less CaO, P2O5 and F, compared with
inclusions in other mafic phenocrysts of the same pumice
clast (Table 3). MgO and MnO concentrations are higher
and FeO∗ lower in Group D than in Group B or E
(Table 3). Fluorine contents of all biotites are high.
ESTIMATION OF INTENSIVE
PARAMETERS: P, T, H 2O AND f O2
Evidence for P, T, H2O, and f O2 for the Groups E–A are
consistent with a single magma chamber that was zoned
to some degree. Normative feldspar components of the
bulk composition of pumices indicate minimum pressures
of ~1 kbar when compared with the position of the
water-saturated minimum on the Ab–Or–Qz ternary
diagram (Fig. 8). Water-undersaturated minima are displaced to greater Or at constant Qz ( Johannes & Holtz,
1990), so pressure estimates should not be strongly affected. A small gradient in pressure is suggested by slight
displacement of Group E compositions to higher pressure
relative to other groups. The scatter within groups subparallel to the Ab–Or tie-line is probably due to slight
alteration (see above).
Water-undersaturation can be deduced from the position of the most sodic samples to the right of the watersaturated minimum combined with the K-enrichment
trend from Group E to A in liquids and feldspars. Liquids
of a K-enrichment trend otherwise would fall to the left
of the minima (Tuttle & Bowen, 1958). Furthermore, an
increase in water from Group E to Group A is consistent
with the observed decrease in crystal content (see Hildreth, 1979). The presence of oligoclase in Group A
suggests an evolution to subsolvus conditions, although
the corresponding sanidines were not found.
For compositions B–E, minimum temperatures are
~800–830°C at 1 kbar pressure because the feldspars
plot on or above the line dividing the single-feldspar
from the two-feldspar field at 750°C, 1 kbar and below
the dividing line at 825°C (Fuhrman & Lindsley, 1988).
Temperature estimates using the formulation of zircon
saturation of Watson & Harrison (1983) progessively
decrease from 880°C in Group E to 840°C in Group B.
Group A yields T of 795°C, although it lacks zircon.
More oxidizing conditions with higher degree of evolution are recorded by an increase of 0·12 log units for
the fayalite–magnetite–quartz assemblage in Group D
Fig. 8. Qz–Ab–Or ternary. Normative pumice compositions (dots for
E and ovals for others) and corresponding feldspars (bars). Scatter
within each group is probably due to limited post-depositional ionexchange. Crosses indicate minima for 0·5, 1, 3, and 5 kbar pressure
at water-saturated conditions from Tuttle & Bowen (1958). Arrow
indicates direction of shift of minima in water-undersaturated conditions
after Johannes & Holtz (1990). Capital letters indicate high-silica rhyolite
group.
compared with Group E, using the FMQ buffer of
Carmichael et al. (1974). Groups C and B contain quartz
and titanomagnetite but no fayalite, consistent with a
further increase in f O2. The progressive increase in f O2 is
in accord with an increase in Mg/Fe of clinopyroxene
and fayalite with degree of evolution of the magma (see
Wones & Eugster, 1965; Grunder & Mahood, 1988). F
data from electron microprobe analysis range from 0·05
wt % in Group E to 0·04 wt % in Group A but are
within analytical error. Cl data from INAA indicate a
decrease from 1250 to 480 p.p.m. We doubt that this is
a magmatic signal; it is more probably an expression of
degassing during eruption.
PARTITION COEFFICIENTS
153
Partition coefficients (D) were determined from mineral–
glass pairs for feldspars, clinopyroxenes, and titanomagnetites for each phyric group (E–B). The determined
values are largely within ranges of previous rhyolite to
high-silica rhyolite studies (Fig. 9). Partition coefficients
of manganese into mafic silicates and oxides of the
Rattlesnake Tuff are among the highest reported (see
Carmichael, 1960, 1967; Deer et al., 1978, 1982; Mahood,
1981; Novak & Mahood, 1986; Warshaw & Smith, 1988).
Sb partition coefficients are among the first reported.
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
Fig. 10. Selected compound partition coefficients.
constant (Fig. 10), consistent with a dominant crystal
chemical control on partitioning (see Mahood & Stimac,
1990). The slight decrease in compatibility of Eu relative
to Ca suggests a decrease in available Eu2+, which in
turn would be consistent with increasingly oxidizing
conditions with differentiation. For the remaining elements, there is no obvious correlation between feldspar
composition and partitioning.
The high partition coefficients for Ba are real because
the samples are crystal poor, from a single magma
chamber, alkali feldspar is the only carrier of Ba (see
Mahood & Stimac,1990), and concentrations are within
limits of Henry’s Law (Long, 1978). D Cr and D Co are
near unity and are significantly higher compared with
values for Sc and Mn (Fig. 9), indicating that high values
for Cr and Co are unlikely to be due to mafic mineral
inclusions.
Clinopyroxene
Fig. 9. Mineral–glass partition coefficients for (a) alkali feldspars; (b)
clinopyroxene; (c) titanomagnetite. Shaded squares represent values
compiled from the literature (Long, 1978; Leeman & Phelps, 1981;
Mahood & Hildreth, 1983; Wörner et al., 1983; Nash & Crecraft, 1985;
Stix & Gordon, 1990; diamonds in (b): Sisson (1991; assuming D Zr =
D Hf ), in (c): Mahood & Stimac, 1990.
Feldspars
Rb, Ba, Sr, La, and Ce partition coefficients in feldspars
progressively increase and those for Eu decrease with
degree of evolution, that is, from Group E to Group B
composition. Compound partition coefficients of Ba/K
[=(Ba/K)fsp/(Ba/K)glass], Rb/K and of Eu/Ca are nearly
Concentrations of Sc, Mn, and Cr in clinopyroxene, and
their partition coefficients, increase with evolution of the
magma, as do those of Zn and Ta, albeit with more
scatter. It is likely that the increase in partitioning of Sc
and Mn is a response to an increase of Fe3+/Fe2+ in the
melt, making incorporation of Sc, Mn, and Zn (and Mg)
over Fe3+ favorable. The nearly constant compound
partition coefficients of Sc/Fe∗ and Mn/Fe∗ (Fig. 10) is
consistent with partitioning of Sc, Mn, and Zn into the
M1 site (Carbonin et al., 1991; Gallahan & Nielsen,
1992). Partitioning of Mn and Mg in fayalite mimics
clinopyroxene, indicating that fayalite composition is also
affected by the increase in Fe3+ in the melt.
HREE patterns of clinopyroxenes are flat, within analytical uncertainty, and do not indicate a downwarp of
154
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Table 5: Mineral–glass partition coefficients of Rattlesnake Tuff high–silica rhyolites
Feldspar
Clinopyroxene
Titanomagnetite
Sample:
RT34E
RT165A
RT173B
RT173C
RT34E
RT165A
RT173B
RT173C
RT34E
RT165A
RT173B
RT173C
HSR grp:
B
C
D
E
B
C
D
E
B
C
D
E
Cs
0·040
0·025
0·060
0·084
Rb
0·34
0·34
0·20
0·13
Th
0·03
2·65
0·08
U
0·40
0·12
0·97
Sr
13·1
15·91
0·25
0·74
10·5
Ba
19·55∗
9·82
6·95∗
La
0·170
0·101
0·070
0·074
0·95
11·08
0·90
0·89
0·22
0·98
0·35
Ce
0·104
0·068
0·037
0·032
1·55
10·71
1·56
1·54
0·20
1·03
0·43
0·56
4·22
11·50
3·60
3·25
0·15
1·36
0·59
1·05
1·01
Nd
0·52
Sm
0·008
0·019
0·015
0·009
6·09
10·43
5·22
4·80
0·11
1·46
0·65
Eu
3·75
4·64
4·97
5·62
4·98
7·76
3·82
3·59
0·13
0·99
0·46
0·68
Tb
0·024
0·022
0·027
4·87
6·45
3·89
3·82
0·08
1·10
0·59
0·88
Yb
0·019
0·026
0·023
Lu
0·012
0·023
Sc
0·016
0·030
Dy
0·014
6·27
6·51
5·19
5·04
0·012
4·94
6·14
4·72
4·75
0·09
1·28
0·29
1·11
0·013
6·06
7·09
5·68
5·40
0·10
1·41
0·24
1·24
0·014
Zr
137·8
140·7
111·9
96·8
1·07
6·47
7·38
7·47
7·18
1·53
4·28
1·51
1·91
0·88
8·45
1·15
7·57
9·02
Hf
0·029
0·042
0·046
0·028
0·95
3·59
1·88
1·44
0·75
8·08
1·32
Ta
0·063
0·034
0·046
0·045
2·80
0·56
0·29
0·46
1·40
2·14
3·60
3·71
Sb
0·22
0·08
0·09
0·13
0·26
0·21
0·58
0·21
11·48
12·06
9·20
9·90
32·59
32·78
28·98
31·19
Mn
0·013
0·023
0·008
0·010
38·33
34·18
28·82
25·70
25·90
23·92
20·47
18·49
Cr
2·35
0·48
1·29
1·11
14·50
10·68
4·13
6·67
4·75
5·23
5·00
3·17
Co
0·87
0·50
1·42
0·75
10·00
36·67
25·83
26·88
12·89
30·94
50·00
33·38
Zn
XRF whole-rock data were used to calculate Sr partition coefficients in cases where Sr could be determined in feldspar
separates. ∗Ba values of glass of 34E and 173C are whole-rock data; for 173C corrected for minerals.
the pattern as observed in clinopyroxenes from other
high-silica rhyolites [ionprobe spot analyses by Sisson
(1991)]. Because Hf concentrations are low, zircon inclusions are unlikely to be dominating the HREE contents
of clinopyroxene separates and the nearly flat HREE
pattern is interpreted to reflect intrinsic clinopyroxene
REE partitioning. Nonetheless, Th and LREE to MREE
partition coefficients for Group C clinopyroxenes and
D Hf for most groups (Fig. 9, Table 5) are higher than
published data, indicating some influence of accessory
phase inclusions.
Titanomagnetite
As in clinopyroxene, concentrations and partition coefficients for Mn, Zn, and Mg in titanomagnetite indicate
an increase with higher degree of evolution of the pumice.
Unlike clinopyroxene, Ta and Sc concentrations
progressively decrease, and D Ta decreases, whereas D Sc
is about constant at seven (Table 5, Fig. 9). Zn and Mn
substitute for Fe2+, whereas Ta5+, with its high charge,
is likely to primarily substitute for Ti, as supported by
nearly constant Ta/Ti compound partition coefficient
(Fig. 10).
The range of partition coefficients for REE, U, Th,
Zr, and Hf of titanomagnetite separates is probably
variably affected by trace phase inclusions which easily
escape detection in opaque titanomagnetites. The trace
element contribution of apatite or glass inclusions is
negligible based on undetectable Ca and Na in the
titanomagnetite samples (Table 4). The samples which
are least affected by zircon or other inclusions and
whose partition coefficients are probably true partition
155
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
coefficients are from Group B and D, as indicated by
low Th, flat REE patterns and low Hf and Zr values.
DISCUSSION
The magma chamber
Similar to previous studies of ignimbrites, the characteristics of the Rattlesnake Tuff are consistent with
reconstruction of a pre-eruptive magma chamber that
was compositionally and mineralogically zoned (see Fig.
14 below). The chamber resided at shallow crustal levels
and was tapped from the chemically highly evolved, more
water-rich, cooler top to deeper less-evolved, phyric and
hotter levels. The high-silica rhyolite chamber was underlain by dacite and more mafic magma, as indicated
by the presence of quenched mafic inclusions and dacite
pumices that are a mixture of mafic magma and the
least-evolved rhyolite (Streck, 1994). The near-liquidus
magma resided at ~0·1 GPa and magma temperatures
ranged from ~880 to ~800°C at the top, consistent with
normative pumice data, single-feldspar compositions, and
Zr geothermometry.
Unusually well documented in the Rattlesnake Tuff
is the existence of compositionally and mineralogically
distinct clusters of pumices. Similar clustering, but defining only two high-silica rhyolite compositions, has been
described from the Grizzly Peak, Mt Jefferson, Rainer
Mesa, and Bandelier Tuffs (Fridrich & Mahood,1987;
Boden, 1989; Hervig & Dunbar, 1992; Cambray et al.,
1995). The Rattlesnake Tuff has five high-silica rhyolite
clusters. We interpret the pre-eruptive magma to have
been stacked in distinct cells in the chamber, with the most
evolved and least dense magma at the top. Calculated dry
densities for Groups A–E increase progressively from
2358 to 2394 kg/m3, at 800°C and 0·1 GPa (after Lange
& Carmichael, 1990) and the density gradient is not
strongly affected by an 80°C temperature increase in the
deeper magma. On the other hand, the density gradient
would be accentuated by concentration of water towards
the roof of the chamber.
Any differentiation scenario for the Rattlesnake Tuff
magma chamber must account for the range of composition from Groups E to A and also for the absence
of some compositions, that is, the existence of clusters
separated by gaps.
Fig. 11. Enrichment rates for selected elements are displayed by
plotting enrichment factors between rhyolite composition E and D
(D/E), composition D and C (C/D), composition C and B (B/C), and
between composition B and A (A/B) (Data from Table 1.)
is, the rate of enrichment through the suite. Enrichment
factors for adjacent rhyolites are plotted according to the
degree of evolution to evaluate enrichment rates for key
elements defining the pumice clusters (Figs 4 and 11).
Ba and Zr enrichment rates decrease continuously with
higher degrees of evolution (Fig. 11). Eu enrichment rates
are constant during the early stages of differentiation but
decrease during the last evolutionary stage (B to A),
whereas the enrichment rate of Eu/Eu∗ stays constant
throughout. La enrichment rates during the first two
increments are close to one (constant concentration)
but decrease rapidly with continued differentiation. The
decrease in enrichment rates for La and Zr could indicate
that an LREE-enriched phase, or phases, and zircon
become more important as evolution proceeded.
Enrichment rates for incompatible elements, such as
Cs, Rb, U, Th, and Ta, are fairly constant but increase
during the last evolutionary step (Fig. 11), accounting for
about half of the maximum variation for these elements
(Fig. 4). The apparent increased incompatibility during
evolution step B to A of these elements is unlikely to be
due to reduced bulk partition coefficients but could be
explained by a reduced mass ratio of daughter to parent
liquid, that is, a higher degree of fractionation.
Crystal fractionation
Petrogenesis of the high-silica rhyolites
Successive enrichment
Because of systematic compositional and mineralogical
changes from Group E to A, it is reasonable to consider
that the different rhyolites lie along a liquid line of
descent. Whatever processes acted must account for the
successive enrichments and depletions in the groups, that
156
The differentiation of the compositional range was
modeled in terms of Rayleigh crystal fractionation in
steps using successively evolved groups as the daughter
liquid, for example, deriving Group D from E, Group
C from D, etc. (Fig. 12). The goal was to determine the
viability and the amount of crystal fractionation necessary
to explain observed gradients. Bulk distribution coefficients (Table 6) were calculated using modal and
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Table 6: Bulk mineral partition coefficients used
for crystal fractionation models in Fig. 12
Bulk partition cofficients
Sample:
34E
165A
173B
173C
HSR grp:
B
C
D
E
Cs
0·01
0·01
0·05
0·07
Rb
0·09
0·19
0·17
0·10
Ba
5·38
8·83
8·64
5·70
Eu
1·45
2·66
4·59
4·97
Sc
12·52
3·13
6·52
9·93
Ta
0·48
0·07
0·20
0·33
Zn
10·56
0·90
1·70
3·08
Mn
10·64
1·20
2·44
3·77
Cs, Rb, and Ba could not be determined in pyroxene and
titanomagnetite and their partition coefficients are assumed
to be zero.
Without fayalite and without trace phases. DEu values for all
magnetites are DEu in 34E, as this is the lowest. DEu for cpx
of 165A is the one from 173B, because cpx of 165A is affected
by LREE phase. DTa for cpx of 34E is exchanged with value
for 165A owing to uncertainty of high Ta at high W in cpx of
34E.
Fig. 12. Crystal fractionation model. Modeled compositional step is
plotted vs the amount of fractionation (cumulative) required to best fit
observed concentrations in daughter liquid given bulk partition coefficient based on observed modal abundances and element concentrations in minerals (Table 6). Open symbols indicate incompatible
elements, filled symbols indicate compatible elements. Selected elements
are independent of accessory trace phases.
partition coefficient data (Tables 2 and 5). To exclude
the effect of the accessory phases zircon, apatite and
chevkinite, elements were used that are insensitive to
their presence, such as Cs, Rb, Ta, Ba, Sc, Zn, and Mn.
Eu is included for the first three steps where behavior of
Eu and Eu/Eu∗ are coupled (Fig. 11).
If all elemental variation could be explained by the
same amount of crystal fractionation, then the lines in
Fig. 12 would coincide. Parallel lines indicate that the
required amount of fractionation is the same but deviated
in an earlier modeling step. If the bulk partition coefficient
is close to unity, then differences of <10% are irrelevant,
such as deviations between Mn and Zn in models D to
C and C to B. On the other hand, when the bulk partition
coefficient deviates considerably from unity, then even a
10% discrepancy is likely to be significant, such as the
difference in fractionation dictated by Ba and Mn in
modeling E to D.
As in Hildreth’s (1979) study of the Bishop Tuff, modal
fractionation cannot account for the chemical trends of
the Rattlesnake Tuff because of large discrepancies in
the amount of fractionation, consistent with enrichment
rates of elements hosted in glass or feldspars, namely Cs,
Rb, and Ba, compared with those elements hosted in
mafic minerals, such as Sc, Mn, and Zn.
To evaluate non-modal crystal fractionation, we undertook the following steps. First, the amount of crystal
fractionation for each step is estimated by assuming that
157
bulk partition coefficients of Rb and Cs are those based
on the modal proportions (Table 6); this assumption can
be made because bulk partition coefficients for Rb and Cs
are largely insensitive to changing mode of an assemblage
consisting of quartz, alkali feldspar, clinopyroxene, and
titanomagnetite. Using these values of fractionation in
combination with elemental variations, appropriate bulk
partition coefficients for trace elements are calculated.
Once the bulk partition coefficients are known, the proportions of fractionating minerals can be calculated using
derived partition coefficients (Table 5). For this purpose,
Ba and Eu are used as tracers for feldspar and Sc, Mn,
and Zn for a best combination of clinopyroxene and
titanomagnetite. This procedure yields the major element
composition of the bulk assemblage being removed from
the magma (Table 7). The major element variations of
the calculated daughters are compared with the observed
compositions and quartz is added to the assemblage to
balance silica. Addition of quartz slightly affects the
calculated bulk partition coefficients by diluting them.
With adjustment of the mode, an optimal model is
then obtained that closely fits major and trace-element
variations (Table 7, Fig. 13).
Accessory phase proportions were determined in a
similar way by using Zr and Hf as a tracer for zircon,
LREE as tracers for an LREE-enriched phase, in this
case allanite, and MREE to adjust apatite (Fig. 13).
Allanite instead of chevkinite was used as the LREEenriched phase because more partitioning data are
available. Addition of chevkinite would reduce the required amount of such a phase, as D LREE is greater.
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 1
JANUARY 1997
Table 7: Results of crystal fractionation models with non-modal mineral proportions
Model E to D
Total frac. %:
Model D to C
7
SiO2
TiO2
13
Model B to A
13
27
Solid
Melt
Av. D
Solid
Melt
Av. C
Solid
Melt
Av. B
Solid
Melt
Av. A
62·7
76·5
76·0
73·3
76·67
76·9
76·4
77·2
77·0
76·7
77·2
77·4
0·87
Al2O3
Model C to B
19·8
0·12
11·9
0·14
12·00
0·79
0·04
13·7
11·8
0·13
11·8
0·65
11·3
0·05
11·9
0·12
11·8
0·51
11·1
-0·02
12·1
0·11
11·9
FeO∗
4·13
1·80
1·68
3·31
1·45
1·48
3·44
1·19
1·22
3·47
0·4
0·77
MnO
0·15
0·09
0·09
0·08
0·09
0·08
0·09
0·08
0·09
0·1
0·09
0·08
MgO
0·10
0·09
0·06
0·01
0·07
0
0·03
0·00
0·07
0·03
0·08
0·06
CaO
1·26
0·48
0·57
0·52
0·58
0·37
0·27
0·39
0·33
0·18
0·39
0·28
Na2O
7·14
3·35
3·10
5·11
2·81
3·5
3·65
3·49
3·30
3·38
3·27
3·59
K 2O
3·86
5·74
6·02
3·16
6·46
5·5
4·14
5·72
5·9
4·54
6·41
5·32
Mineral (%)
fsp
92·5
qtz
0
67
58
57
28·5
37·4
cpx
38·6
3·5
0·5
0·6
0·4
timt
4
4
4
4
zirc
0·2
0·13
0·2
0·15
ap
0
0·04
0·05
0·9
alla
0·02
0·05
0·13
0·12
Trace elements in
removed solids†
Model:
E to D
D to C
C to B
B to A
Bulk:
syenite
granite
high-Si granite
high-Si granite
Th
1·7
2·4
5·6
U
2·0
1·6
2·7
6·3
2·9
Sr
183
79
46
11
Ba
7460
3360
1150
280
La
31
61
118
62
Ce
62
123
248
141
Sm
5·1
6·6
13·9
Eu
10·8
5·4
3·5
12·1
1·7
Yb
10·3
7·0
11·4
10·1
Lu
1·77
1·2
1·9
1·7
Nb
6·6
7·5
5·1
6·1
Ta
Zr
0·4
940
0·37
668
0·27
766
Hf
15·0
11·6
17·6
Sc
14·0
3·9
5·1
Zn
175
132
144
0·23
450
13·2
3·9
114
Total frac. %, amount of crystals fractionated; solid, bulk composition of removed minerals; melt, modeled daughter
composition; av. X, composition of group averages (Table 1); min.%, mineral proportions of removed solid; fsp, alkalifeldspar; qtz, quartz; cpx, clinopyroxene; timt, titanomagnetite; zirc, zircon; ap, apatite; alla, allanite.
†Calculated with modeled bulk partition coefficients and observed daughter compositions.
158
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Fig. 13. Modeling of bulk partition coefficients to test non-modal fractionation (see text). Χ, bulk partition coefficients consistent with proportions
of crystal fractionation based on enrichments of Cs and Rb in successive high-silica rhyolite daughters D, C, B, and A indicated by capital letters
( Table 1). Β, bulk partition coefficients based on best proportions of observed major phases: alkali-feldspar, clinopyroxene, titanomagnetite, and
quartz. Α, bulk partition coefficients with addition of accessory phases. Partition coefficients for single phases are those from Table 5 with some
exceptions. For all models, D Zr,Hf values for titanomagnetite are from RT34E. For model C to B, D REE values of clinopyroxene and titanomagnetite
are from RT34E. In addition, all D values are 0·01 for quartz; D Sr =10 for feldspar where not determined; D Th,U = 0·01 for all major phases;
D Sr,Ba = 0·01 for clinopyroxene, titanomagnetite, zircon, and allanite; D REE values are interpolated where not available; D Zr,Hf = 0·3 for clinopyroxene
(Sisson, 1991); D Y = D Tb for all phases where no DY available; D Nb = D Ta for all phases; and D Zn = D Mn for feldspar. Partition coefficients for
zircon and allanite are based on values for those phases from the Bishop Tuff ( Mahood & Hildreth, 1983), except Rattlesnake Tuff zircons were
used for D Zr,Hf . Apatite partition coefficients were taken from Mahood & Stimac (1990). Allanite instead of chevkinite was used as LREE-enriched
phase owing to limited D values for chevkinite. Mn values of Groups A–E are values determined from glass separates by INAA.
The total amount of crystal fractionation from composition E to A is 51% and, overall, the proportion of
feldspar in the modeled fractionation assemblages is
greater than in the mode, and the proportions of
clinopyroxene and titanomagnetite are lower than in the
mode.
Not well reproduced are bulk partition coefficients for
HREEs for step D to C, which could be due to using
higher D HREE for zircon than would actually be the case
in Rattlesnake Tuff zircons (Fig. 13). Bulk partition
coefficients for Th, U, Ta, and Nb are also not well
reproduced and require bulk partition coefficients that
are mainly lower for Th and U and higher for Ta and
Nb.
Assimilation, mixing and progressive partial melting
Contamination of the magma by assimilation of components derived from the wall rock may have contributed
to the compositional variation in the Rattlesnake Tuff,
and it could reduce the amount of fractionation, as
modeled above, by as much as 1/5 (DF of 10%), but
contamination does not govern the trace element distribution. For the major element composition of the
159
JOURNAL OF PETROLOGY
VOLUME 38
magma not to be affected and without increasing the
amount of fractionation during AFC processes, the assimilant has to have a very similar bulk composition
(DePaolo, 1981). Thus, any assimilant similar to Rattlesnake Tuff rhyolites would represent easily fused components from wall rocks, which would be enriched in
incompatible elements such as Cs and Rb. On the other
hand, Ta, among others, is likely to be housed in residual
phases and so not be contributed from wall rocks. There
is a modest increase in Rb/Ta, Cs/Ta, and U/Ta from
Group E to A, indicating that ~15% of the concentrations
of Rb, Cs, and U are unaccounted for by fractionation
alone (that is, 20, 0·7, and 0·7 p.p.m., respectively). Rb/
Ta, Cs/Ta and U/Ta are constant during fractionation
(assuming the same bulk partition coefficient of zero). In
addition, the agreement between the amount of fractionation indicated by the very incompatible elements
and Ba, which is compatible, precludes substantial assimilation. Easily fused components of wall rock are likely
also to contain more Ba than Group A and B, in which
case the calculated proportion of fractionation, neglecting
assimilation, would be too low, whereas the incompatible
elements would indicate the opposite.
Banded pumices of the Rattlesnake Tuff clearly indicate
mixing; such macroscopically observable mixing is likely
to have occurred during eruption in the vent or along
the interface between different rhyolites. As a whole,
nonlinear trace element trends for the suite (Fig. 3)
preclude simple mixing as the mechanism for compositions between A and E.
Progressive partial melting can be excluded on similar
grounds as argued by Hildreth (1979, 1981) and Mahood
(1981). Problems with progressive partial melting include:
(1) extremely low Ba concentration in the most evolved
rhyolite A; (2) 60-fold Ba depletion compared with twofold Rb increase over a 2·5 wt % silica range; (3)
enrichment in HREE, but depletion in LREE in A
compared with E yielding a cross-over of the REE
pattern.
An evolutionary model and origin of
compositional clusters
Crystal fractionation of an assemblage richer in feldspar
plus quartz and poorer in mafic minerals than the mode
is considered the dominant process required to generate
the compositional variation of the Rattlesnake Tuff. Fractionation, however, cannot account for the existence of
compositional clusters. Mineral compositions cluster as
well (e.g. minerals of Group D overlap those from E),
indicating crystallization from discretely different batches
of magma. So although eruption dynamics may contribute to compositionally discontinuous sampling of a
magma chamber, we believe the compositional clusters
NUMBER 1
JANUARY 1997
reflect the existence of compositionally segregated magma
before eruption.
The site of crystal fractionation depends on the geometry and heat loss of the magma chamber. Deep parts
of the chamber are unlikely fractionation sites for rhyolites
based on many magma chamber models indicating that
rhyolites are underlain by mafic and hotter magmas (e.g.
Hildreth, 1981; Wörner & Schmincke, 1984; Fridrich &
Mahood, 1987; Bacon & Druitt, 1988), which is also the
case for the Rattlesnake Tuff (Streck, 1994). Sidewall
crystallization is widely proposed for differentiation in
magma chambers (e.g. Turner & Gustafson, 1981; Spera
et al., 1984; de Silva & Wolff, 1995). The area available
for sidewall crystallization will rival the area of the
chamber roof when the chamber has a 1:4 ratio of height
to width (assuming a cylindrical shape). If heat loss is
greatest through the roof, then roofward crystallization
is likely to be increasingly important in flatter disk-shaped
chambers. Tank experiments simulating boundary-layer
crystallization processes indicate that the roof can be the
main crystallization site when cooling occurs primarily
from the roof downward and will only affect the uppermost layers (Baker & McBirney, 1985). Flat cylindrical
shapes for large silicic bodies have been proposed based
on diameters and downdrop of calderas (Smith, 1979;
Spera & Crisp, 1981) and based on comparison of
voluminous tuffs with intrusive:extrusive ratios (de Silva
& Wolff, 1995).
We propose for the Rattlesnake Tuff magma chamber
that a significant part of the crystallization took place
along the roof of the chamber that started as a homogeneous batch of least-evolved rhyolite E (see Bacon &
Druitt, 1988). Rhyolite D was generated from E (Fig. 14)
and because the derivative daughter liquid was lighter,
it stayed more or less where it was generated. Marginal
crystallization would contribute liquids that rise to become part of the differentiating cap. The liquid contributions to the layer are from the underlying parent as
well as from crystallization of the layer itself and convection of the layer homogenizes them (Rayleigh numbers
calculated for the Rattlesnake Tuff magmas are greater
than critical for layers over 1·5 m thick). Such accumulation proceeds until the layer reaches a critical
thickness, at which point a new layer initiates at the top
of the chamber (Turner & Campbell, 1986). The critical
thickness is reached when the layer cannot be stirred.
In such a model, the daughter liquid is a mixture of
fractionated liquids whose overall mode of minerals removed could be more felsic and less mafic than of the
observed mode.
The first derivative rhyolite D becomes the next parent
composition and the process of crystallization at the roof
repeats itself (Fig. 14). Sidewall crystallization may still
160
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
same process happens several times then a compositionally zoned rhyolitic magma body, as recorded in
the Rattlesnake Tuff rhyolites, can be generated (Fig.
14).
Alternatively, compositional layers may be due to punctuated accumulation of fractionated liquids. Such a process might occur in a chamber where thermal input
fluctuates, so that the chamber undergoes stages of crystallization and thus accumulation of a new layer of
differentiate. Repetition of the process could then
generate a series of progressively more evolved compositions with gaps. Such a model could work in concert
with the one above.
CONCLUSION
Fig. 14. Proposed differentiation model. Each more evolved liquid is
generated from the previous less-evolved liquid mainly through derivative liquids obtained from boundary-layer fractionation processes
along the top and sides of the magma chamber. From top down: (1)
is stage after first differentiation interval producing composition D from
E; (4) represents last evolution stage generating composition A from
B; (5) shows hypothetical high-silica rhyolite portion of pre-eruptive
Rattlesnake Tuff magma chamber and crystallized granitoid margins;
shading indicates chemical zonation within both, and the thickness of
the crystallized margins reflects the amount of crystallization consistent
with the chemical models.
contribute to maintenance of previous layers of differentiate. Intra-layer convection would help preserve compositional layers (see Turner & Campbell, 1986). If the
161
Chemical data for glassy pumices and shards from the
Rattlesnake Tuff [280 km3 dense rock equivalent (DRE)]
cluster into groups recording 4–5 distinct and progressively more evolved, high-silica rhyolite magmas.
Crystal content decreases from ~1 to 0 wt % towards
the most evolved composition. With degree of evolution
of the magma, alkali feldspars (anorthoclase to sodic
sanidine) become more potassic, Fe-rich clinopyroxene
more magnesian, and titanomagnetite more iron rich.
The pre-eruptive, high-silica rhyolite cap of the Rattlesnake Tuff magma chamber is interpreted to have been
stratified with several compositionally homogeneous layers, increasingly more differentiated towards the top.
Partition coefficients were determined for feldspars,
clinopyroxenes, and titanomagnetites. They mainly fall
within the range of previously published data and change
progressively with degree of differentiation for elements
whose partitioning behavior correlates closely with crystal
chemical changes.
Non-modal crystal fractionation models can account
almost completely for the observed chemical gradients
among high-silica rhyolites requiring a cumulative fractionation amount of ~50% from the least to the most
evolved composition. Modeled fractionated mineral assemblages have more feldspar and quartz and a smaller
proportion of mafic minerals than the mode and they
range from syenite to high-silica granite, suggesting that
voluminous masses of common plutonic rocks are left at
depth during differentiation of large-volume rhyolites.
The non-modal crystal fractionation process is envisioned to take place at the roof and upper walls of a
slab-shaped rhyolitic top of a larger magma chamber.
Less dense and more evolved liquids are generated and
are mixed into a new homogeneous rhyolite layer. The
new rhyolite layer, shielding the parent rhyolite from the
roof, itself fractionates while sidewall differentiates from
the parent are still contributed. When the daughter
layer reaches a critical thickness, a new layer of roof
JOURNAL OF PETROLOGY
VOLUME 38
differentiates is initiated, repeating the process. This way,
each more evolved rhyolite layer becomes the parent
composition for the next more evolved rhyolite, leading
to a density and compositionally stratified high-silica
rhyolite magma column. Compositional gaps to the previous parent would be generated and maintained by
mixing within layers in the column.
ACKNOWLEDGEMENTS
The INAA Team at OSU, Roman Schmitt, Art Johnson,
Jack Higginbotham, Brian Dodd, and Mike Conrady,
provided scientific and financial support which are highly
appreciated. Joel Sparks is thanked for assistance with
obtaining excellent XRF analyses. H.-U. Schmincke is
thanked for making the microprobe facility at GEOMAR
available for additional shard analyses. This study was
supported by Chevron and GSA–Penrose grants to M.J.S.
and National Science Foundation Grant EAR-9220500
to A.L.G. Final preparations were done while the first
author was supported by a postdoc stipend through
the Graduiertenkolleg ‘Dynamik globaler Kreisläufe’ at
GEOMAR. Gail Mahood, Stephen Blake, and Shanaka
de Silva are thanked for their thoughtful reviews, which
helped considerably to improve the manuscript; additional comments and editorial assistance throughout the
review process by Marjorie Wilson are also appreciated.
REFERENCES
Bacon, C. R., 1989. Crystallization of accessory phases in magmas by
local saturation adjacent to phenocrysts. Geochimica et Cosmochimica
Acta 53, 1055–1066.
Bacon, C. R. & Druitt, T. H., 1988. Compositional evolution of the
zoned calcalkaline magma chamber of Mount Mazama, Crater
Lake, Oregon. Contributions to Mineralogy and Petrology 98, 224–256.
Baker, B. H. & McBirney, A. R., 1985. Liquid fractionation, part III:
Geochemistry of zoned magmas and the compositional effects of
liquid fractionation. Journal of Volcanology and Geothermal Research 24,
55–81.
Blake, S, 1981. Eruptions from zoned magma chambers. Journal of the
Geological Society, London 138, 281–287.
Blake, S. & Ivey, G. N., 1986. Magma-mixing and the dynamics of
withdrawal from stratified reservoirs. Journal of Volcanology and Geothermal Research 27, 153–178.
Boden, D. R., 1989. Evidence for step-function zoning of magma and
eruptive dynamics, Toquima caldera complex, Nevada. Journal of
Volcanology and Geothermal Research 37, 39–57.
Cambray, F. W., Vogel, T. A. & Mills, J. G., Jr, 1995. Origin of
compositional heterogeneities in tuffs of the Timber Mountain
Group: the relationship between magma batches and magma transfer
and emplacement in an extensional environment. Journal of Geophysical
Research 100, 15793–15805.
Carbonin, C., Dal Negro, A., Ganeo, S. & Piccirillo, E. M., 1991.
Influence of magma composition and oxygen fugacities on the crystal
structure of C2/c clinopyroxenes from a basalt–pantellerite suite.
Contributions to Mineralogy and Petrology 108, 34–42.
NUMBER 1
JANUARY 1997
Carmichael, I. S. E., 1960. The pyroxenes and olivines from some
Tertiary acid glasses. Journal of Petrology 1, 309–336.
Carmichael, I. S. E., 1967. The iron–titanium oxides of salic volcanic
rocks and their associated ferromagnesian silicates. Contributions to
Mineralogy and Petrology 14, 36–64.
Carmichael, I. S. E., Turner, F. J. & Verhoogen, J., 1974. Igneous
Petrology. New York: McGraw–Hill.
Deer, M. A., Howie, R. A. & Zussman, J., 1978. Single-chain Silicates.
New York: John Wiley.
Deer, M. A., Howie, R. A. & Zussman, J., 1982. Orthosilicates. New
York: John Wiley.
De Silva, S. L. & Wolff, J. A., 1995. Zoned magma chambers:
the influence of magma chamber geometry on sidewall convective
fractionation. Journal of Volcanology and Geothermal Research 65, 111–118.
DePaolo, D. J., 1981. Trace element and isotopic effects of combined
wallrock assimilation and fractional crystallization. Earth and Planetary
Science Letters 53, 189–202.
Fisher, R. V. & Schmincke, H.-U., 1984. Pyroclastic Rocks. Heidelberg:
Springer.
Fridrich, C. J. & Mahood, G. A., 1987. Compositional layers in the
zoned magma chamber of the Grizzly Peak Tuff. Geology 15, 299–303.
Fuhrman, M. L. & Lindsley, D. H., 1988. Ternary-feldspar modeling
and thermometry. American Mineralogist 73, 201–215.
Gallahan, W. E. & Nielsen, R. L., 1992. The partitioning of Sc, Y,
and the rare earth elements between high-Ca pyroxene and natural
mafic to intermediate lavas at 1 atmosphere. Geochimica et Cosmochimica
Acta 56, 2387–2404.
Grunder, A. L. & Mahood, G. A., 1988. Physical and chemical models
of zoned silicic magmas: the Loma Seca Tuff and Calabozos Caldera,
southern Andes. Journal of Petrology 29, 831–867.
Hervig, R. L. & Dunbar, N. W., 1992. Cause of chemical zoning in
the Bishop (California) and Bandelier (New Mexico) magma chambers. Earth and Planetary Science Letters 111, 97–108.
Hildreth, W., 1979. The Bishop Tuff: evidence for the origin of
compositional zonation in silicic magma chambers. In: Chapin, C.
E. & Elston, W. E. (eds) Ash-flow Tuffs. Geological Society of America,
Special Papers 180, 43–75.
Hildreth, W., 1981. Gradients in silicic magma chambers: implications
for lithospheric magmatism. Journal of Geophysical Research 86, 10153–
10192.
Hildreth, W., Christiansen, R. L. & O’Neil, J. R., 1984. Catastrophic
isotopic modification of rhyolite magma at times of caldera subsidence, Yellowstone Plateau volcanic field. Journal of Geophysical
Research 89, 8339–8369.
Johannes, W. & Holtz, F., 1990. Formation and composition of H2Oundersaturated granitic melts. In: Ashworth, J. R. & Brown, M.
(eds) High-temperature Metamorphism and Crustal Anatexis. London: Unwin
Hyman, pp. 87–104.
Lange, R. J. & Carmichael, I. S. E., 1990. Thermodynamic properties
of silicate liquids with emphasis on density, thermal expansion, and
compressibility. Mineralogical Society of America, Reviews in Mineralogy
24, 25–64.
Leeman, W. P. & Phelps, D. W., 1981. Partitioning of rare earth
elements and other trace elements between sanidine and coexisting
volcanic glasses. Journal of Geophysical Research 86, 10193–10199.
Long, P. E., 1978. Experimental determinations of partition coefficients
for Rb, Sr, and Ba between alkali feldspar and silicate liquid.
Geochimica et Cosmochimica Acta 42, 833–846.
MacLeod, N. S., Walker, G. W. & McKee, E. H. 1976. Geothermal
significance of eastward increase in age of upper Cenozoic rhyolitic
domes in southeastern Oregon. Second United Nations Symposium on the
Development and Use of Geothermal Resources, San Francisco, CA, Proceedings,
162
STRECK AND GRUNDER
RATTLESNAKE TUFF RHYOLITES
Vol. 1 Washington, DC: US Government Printing Office,
pp. 465–474.
Mahood, G. A., 1981. Chemical evolution of a Pleistocene rhyolitic
center: Sierra La Primavera, Jalisco, Mexico. Contributions to Mineralogy
and Petrology 77, 129–149.
Mahood, G. A. & Hildreth, W., 1983. Large partition coefficients for
trace elements in high-silica rhyolites. Geochimica et Cosmochimica Acta
47, 11–30.
Mahood, G. A. & Stimac, J. A., 1990. Trace-element partitioning
in pantellerites and trachytes. Geochimica et Cosmochimica Acta 54,
2257–2276.
Michael, P., 1983. Chemical differentiation of the Bishop Tuff and
other high-silica magmas through crystallization processes. Geology
11, 31–34.
Michael, P., 1988. Partition coefficients for rare earth elements in mafic
minerals of high-silica rhyolites: the importance of accessory mineral
inclusions. Geochimica et Cosmochimica Acta 52, 275–282.
Miller, C. F. & Mittlefehldt, D. W., 1984. Extreme fractionation in
felsic magma chamber: a product of liquid-state diffusion or fractional
crystallization. Earth and Planetary Science Letters 68, 151–158.
Nash, W. P. & Crecraft, H. R., 1985. Partition coefficients for trace
elements in silicic magmas. Geochimica et Cosmochimica Acta 49, 2309–
2322.
Novak, S. W. & Mahood, G. A., 1986. Rise and fall of a basalt–
trachyte–rhyolite magma system at the Kane Springs Wash Caldera,
Nevada. Contributions to Mineralogy and Petrology 94, 352–373.
Sisson, T. W., 1991. Pyroxene–high silica rhyolite trace element partition coefficients measured by ion microprobe. Geochimica et Cosmochimica Acta 55, 1575–1585.
Smith, R. L., 1960. Ash flows. Geological Society of America Bulletin 71,
795–842.
Smith, R. L., 1979. Ash-flow magmatism. In: Chapin, C. E. & Elston,
W. E. (eds) Ash-flow Tuffs. Geological Society of America, Special Paper 180,
5–27.
Smith, R. L. & Bailey, R. A., 1966. The Bandelier Tuff: a study of
ash-flow cycles from zoned magma chambers. Bulletin of Volcanology
29, 83–104.
Spera, F. J. & Crisp, J. A., 1981. Eruption volume, periodicity, and
caldera area: relationships and inferences on development of compositional zonation in silicic magma chambers. Journal of Volcanology
and Geothermal Research 11, 169–187.
Spera, F. J., Yuen, D. A. & Kemp, D. V., 1984. Mass transfer rates
along vertical walls in magma chambers and marginal upwelling.
Nature 310, 764–767.
Stix, J. & Gordon, M. P., 1990. Variations in trace element partition
coefficients in sanidine in the Cerro Toledo rhyolite, Jemez
Mountains, New Mexico: effects of composition, temperature, and
volatiles. Geochimica et Cosmochimica Acta 54, 2697–2708.
Streck, M. J., 1994. Volcanology and petrology of the Rattlesnake AshFlow Tuff, eastern Oregon. Ph.D. Thesis, Oregon State University,
Corvallis, pp. 1–185.
Streck, M. J. & Grunder, A. L., 1995. Crystallization and welding
variations in a widespread ignimbrite sheet; the Rattlesnake Tuff,
eastern Oregon, USA. Bulletin of Volcanology 57, 151–169.
Turner, J. S. & Campbell, I. H., 1986. Convection and mixing in
magma chambers. Earth-Science Reviews 23, 255–352.
Turner, J. S. & Gustafson, L. B., 1981. Fluid motions and compositional
gradients produced by crystallization or melting at vertical boundaries. Journal of Volcanology and Geothermal Research 11, 93–125.
Tuttle, O. F. & Bowen, N. L., 1958. Origin of granite in the light of
experimental studies in the system NaAlSi3O8–KAlSi3O8–SiO2–H2O.
Geological Society of America, Memoir 74, 153 pp.
Warshaw, C. M. & Smith R. L., 1988. Pyroxenes and fayalites in the
Bandelier Tuff, New Mexico: temperatures and comparison with
other rhyolites. American Mineralogist 73, 1025–1037.
Watson, E. B. & Harrison, T. M., 1983. Zircon saturation revisited:
temperature and composition effects in a variety of crustal types.
Earth and Planetary Science Letters 70, 295–304.
Wones, D. R. & Eugster, H. P., 1965. Stability of biotite: experiment,
theory, and application. American Mineralogist 50, 1228–1272.
Wörner, G. & Schmincke, H.-U., 1984. Petrogenesis of the zoned
Laacher See tephra. Journal of Petrology 25, 836–851.
Wörner, G., Beusen, J.-M., Duchateau, N., Gijbels, R. & Schmincke,
H.-U., 1983. Trace element abundances and mineral/melt distribution coefficients in phonolites from the Laacher See Volcano
(Germany). Contributions to Mineralogy and Petrology 84, 152–173.
Zielinski, R. A., 1982. The mobility of uranium and other elements
during alteration of rhyolitic ash to montmorillonite: a case study
in the Troublesome formation, Colorado, U.S.A. Chemical Geology
35, 185–204.
163