Full text of DOI:10.1007/s004100000211

Contrib Mineral Petrol .2001) 140: 554±566
DOI 10.1007/s004100000211
F. Bea á A. Arzamastsev á P. Montero á L. Arzamastseva
Anomalous alkaline rocks of Soustov, Kola: evidence of mantle-derived
metasomatic ¯uids affecting crustal materials
Received: 5 May 2000 / Accepted: 29 September 2000 / Published online: 1 December 2000
Ó Springer-Verlag 2000
Abstract The intrusive complexes of Gremiakha- solute composition of the ¯uid duringthe in®ltration
Vyrmes and Soustov represent the two extremes of the metasomatism. Our results, usingthe LREE abundances
Early Proterozoic alkaline plutons of Kola, predomi- and the Sr and Nd isotope composition of a Gremiakha-
nantly composed of feldespathoidal syenites. Gremia- Vyrmes pegmatite as the starting solute composition of
kha-Vyrmes rocks .zircon age: 1,884‹6 Ma) have the ¯uid, and the mode and mineral trace-element and
trace-element and isotope signatures .87Sr/86Srt ꢀ 0.704, isotope composition of a common Kola gneiss as rep-
ꢀNdt ꢀ )3±1.3) compatible with an ultimate mantle resentative of percolated materials, indicate that the
origin. Soustov syenites .zircon age: 1,872‹8 Ma) are ¯uid would have acquired a signature closely matching
totally di€erent and show an acute crustal imprint. They Soustov's, even in the case of Nd isotopes, if the gneiss
have sodaline and analcite instead of nepheline, contain age is 2.9 Ga, near its real age. This model is still a mere
a plethora of REE-HFSE-rich accessories, and are workinghypothesis that needs further re®nements, but
characterised by elevated contents of F, Cl, REE, Y, Th, may represent a reasonable explanation of the genesis
U, Zr, Hf, Nb, Ta, Sn, Be, Li, Rb, Tl, Pb and Cs, neg- of anomalous alkaline rocks with high 87Sr/86Srt and
ative Eu anomalies, K/Rb ꢀ 190±160, Nd/Th ꢀ 3, and ꢀNdt ³ 0, either saturated or undersaturated, which are
87
Nb/Ta ꢀ 12, with extremely high Sr/86Srt .>0.720) dicult to understand in terms of magmatic fractiona-
and, at the same time, relatively high ꢀNdt .ꢀ)1.6±1.7). tion/contamination.
In this paper, we explore the idea that the anomalous
features of Soustov syenites can be explained if we as-
sume they are derived from a metasomatic agent, ini-
tially an H2O-CO2 supercritical ¯uid released by alkaline
ma®c magmas, that was profoundly contaminated dur-
ingpercolation through crustal materials. As percola-
tion advanced, the bulk composition of the ¯uid solute
changed from alkali halides and carbonates to a silica-
Introduction
In the Kola Peninsula, the Early Proterozoic .ꢁ1,900 Ma)
episode of alkaline magmatism produced numerous
small complexes, with feldspathoidal syenites often
undersaturated alkaline melt. When the ¯uid cooled to a
temperature of ꢁ550±600 °C, it reached the point at
predominating.Kukharenko et al. 1971; Mitrofanov
et al. 1995). The syenites from these complexes share
which vapor and melt were no longer miscible and split
similar major-element and Nd isotope compositions
into two components, a vapour phase and a Cl- and
F-rich silica-undersaturated silicate melt that crystallised
.ꢀNdt ꢀ )3±2), but show stunningdi€erences in trace
elements and Sr isotopes .Tables 1, 2). In most cases,
they have 87Sr/86Srt ꢀ 0.704 and element ratios consis-
to produce Soustov syenites. To study this process, we
have developed a numerical method for modellingthe
tent with an ultimate mantle origin. In a few cases,
however, such those of Soustov and the plutono-volca-
nic complexes of Tominga .Batieva et al. 1983) or
Poritash mountain .Balashov 1996), the syenites and
trachytes have notably high 87Sr/86Srt .up to 0.720) and
distinctive trace-element patterns, with low K/Rb, low
Nd/Th, low Nb/Ta, negative Eu anomalies, etc., which
reveal an acute crustal imprint. These features, especially
the simultaneously high 87Sr/86Srt and 143Nd/144Ndt,
cannot have been caused merely by the bulk incorpo-
ration of crustal materials into mantle-derived melts.
F. Bea .&) á P. Montero
Department of Mineralogy and Petrology,
Fuentenueva Campus, University of Granada,
18002 Granada, Spain
e-mail: fbea@goliat.ugr.es
Tel.: +34-95-8246176; Fax: +34-95-8243368
A. Arzamastsev á L. Arzamastseva
Geological Institute, Kola Science Center,
14 Fersman St, Apatity, Murmansk region, 184200 Russia

555
Table 1 Major- and trace-element composition of Soustov and Gremiakha-Vyrmes syenites. Associations: *Foidolitic; **ultrama®c;
***silica-saturated peralkaline
Soustov
1
Gremiakha-Vyrmes
2
3
4
5
6
7
2*
3*
5**
6***
7*
14*
15**
16**
Major-element composition .%)
SiO2
TiO2
63.27
0.17
Al2O3 19.72
61.21 57.23 62.30
0.23 0.29 0.14
19.17 19.51 18.90
57.48 56.04 56.20
0.51 0.29 0.28
15.71 17.66 20.32
52.58 55.80
0.85 0.40
22.75 21.90
57.53 63.79
1.02 0.48
15.62 16.22
55.68 56.39
0.50 0.32
16.24 19.98
4.88 2.33
2.79 3.72
0.59 0.41
0.17 0.14
3.10 1.56
7.70 7.51
5.54 5.15
0.11 0.11
59.23
60.86
1.08
16.05
4.04
5.23
0.45
0.24
3.21
6.13
3.95
0.27
0.52
17.40
2.28
2.74
0.22
0.11
2.53
6.87
4.40
0.19
Fe2O3
FeO
MgO
MnO
CaO
Na2O
K2O
P2O2
F
1.22
1.58
0.12
0.06
0.96
6.55
6.15
0.03
0.16
0.36
0.17
2.23
2.45
0.12
0.13
1.21
7.14
5.54
0.03
0.11
0.29
0.33
2.44 1.09
3.74 1.25
0.46 0.03
0.18 0.06
2.13 0.67
7.33 7.63
5.25 5.64
0.05 0.03
0.29 0.28
0.39 0.41
0.93 1.64
4.27
5.09
1.51
0.26
2.81
2.90
3.32
0.72
0.15
3.06
2.11
2.92
0.19
0.16
1.32
7.70
6.23
0.05
2.58
2.63
0.77
0.09
1.87
10.39
5.19
0.11
1.42
2.84
0.40
0.10
1.12
8.97
5.53
0.08
3.63
6.58
0.52
0.28
2.73
6.10
3.88
0.22
2.35
3.32
0.13
0.13
1.16
7.03
4.97
0.12
6.58 10.09
5.06
0.09
0.39
0.47
0.24
4.58
0.05
0.21
0.31
0.71
0.32 £ 0.06 £ 0.06 £ 0.06 £0.06 £ 0.06 £0.06 £ 0.06 £ 0.06
Cl
H2O
0.42 £ 0.03 ³30.03 £ 0.03 £0.03 £ 0.03 £0.03 £ 0.03 £ 0.03
1.72
0.32
1.35
1.54
0.17
2.56 1.95
0.11
0.89
Total 100.52 100.18 100.21 100.07 100.47 100.09 99.94 100.14 99.97
99.68 99.90
99.91 99.60 100.00
99.06
Trace-element composition .ppm)
Li
47
276
99
325
5.88
14.0
171
173
0.76
0.83
1.32 120
1.35
3.51
7.19 12.9
124 152
37.0
62.3
159
12.6
1258
27.3
4.18
13.95 15.8
0.80
27.8
8.45
38.5
115
207
20.6
67.0 106
11.5 18.3
1.35 2.42 0.92
9.44 14.31 5.78
1.60 2.32 1.01
9.99 13.18 5.73
105
302
69
247
51
273
8.13
8.78 11.1
86.1 217
150
3.16
1.12
36.6
2.02
8.96
27.5
286
32.7
45.5
138 161
9.09 13.5
151
217
4.26 10.7
73
369
4.1
81.0
14.4
76.2
7.07 12.7
4.10 20.4
89.9 116
0.95 1.28
3.39 1.58
4.9
16.5
3.2
14.8
0.25
0.38
Rb
Cs
Be
Sr
Ba
Sc
23.7
0.45
0.53
103
6870
10.02
1.84
7.14 23.2
84.9
0.48
3.95
164
434
2.07
1.19
5.13
5.07
130
200
0.25
1.12
2.57
0.63
2.21
6.74
8.75 4.12
13.3 5.64
83.5
172
0.31
3.72
755
2059
2.37
23.2
23.1
7.43
9.95
13.1
107
50.5
5.00
32.3
2.71
106
0.62
2.04
198
578
1.51
1.43
12.3
3.56
2.72
8.73
56.8
33.3
7.80
47.4
1.73
33.8
1.37
0.69
0.04
0.23
2.78
1.74
1.99
0.20
0.30
114
6899
14.3
0.69
10.43
2.37
6.24
8.22
19.2
425
112
0.64
4.89
7.99 17.6
608
396
1240
1586
221
223
129
11960
9.98
291
2.96
0.99
0.94 0.03
5.37 0.47
3.62 1.90
V
16.5
37.1
1.24
3.16
0.53
9.86
1.12
4.85
5.75
211
24.7
4.69
4.66
0.33
Cr
Co
Ni
Cu
Zn
Ga
Y
Nb
Ta
Zr
Hf
Mo
Sn
Tl
Pb
U
Th
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
0.47
2.61 34.8
8.99 0.54
1.97
5.14
11.4
141
37.9
62.2
3.36
7.38
31.01
149
35.5
42.0
123
9.94
783
17.4
3.07
6.44
13.3
247
25.6
1.87
5.55
8.90
87.7
42.9
4.58 3.99
7.98 4.76
10.7 10.1
3.52
34.7
47.9
120
69.7
48.7 34.2
8.33 9.72
201
32.7
26.2
56.5
5.01
361
7.51
0.91
3.99
0.77
11.4
2.55
11.8
43.7
86.3
8.9
30.4
5.46
0.78
4.46
0.75
4.60
1.00
2.63
0.37
1.83
0.22
38.8 32.1
79.3 32.9
24.47
5.75
4.48
0.28
21.9
0.69
0.92
0.36
0.07
2.83
0.11
0.33
8.78
22.3
3.24
15.2
3.25
6.14
2.70
0.32
1.46
0.23
0.53
0.07
0.43
0.07
7.22 18.9
9.28 54.9
1.11 2.64
58.3 322
161
13.6
1137
25.7
3.28 2.01
4.66
0.77 0.55
32.7 20.05
8.39 3.18
41.1 12.33
173
311
31.6 11.9
39.9
7.06
67.8
6.92
363
8.05
367
15.3
542
15.4
95.3
3.42
81.2
2.77
1141 1879
29.09 43.5
41.1
3.54
0.44
0.28
0.14
2.35
0.34
0.90
1.60
1.87
7.66
1.52
1.06
0.72
0.62
0.05
2.31
0.14
0.45
6.14
5.63
6.47 15.9
1.03 0.89
8.96 28.819 24.9
4.02
18.74 35.1
112
222
25.3
92.4
16.5
2.14
12.1
1.72
9.52
1.89
4.63
0.73
4.32
0.71
6.15 10.5
15.7
1.32
0.83 0.21
8.74 0.37
0.14 1.07
6.98 4.28
0.130 0.92
0.07
3.01
2.20
1.55
0.26
3.61
0.88
3.76
7.49
9.3
45.7
148
248
23.3
73.1
11.1
1.19
8.74
1.39
7.76
1.47
3.70
0.55
3.06
0.42
12.4
1.19
2.65 1.70
15.1 33.5
38.3 76.9
4.78 9.55
19.1 36.1
3.70 6.49
1.34 1.41
2.88 4.56
0.41 0.54
1.99 2.62
0.37 0.44
0.85 0.92
0.14 0.12
0.90 0.76
0.16 0.11
60.7
118
132
236
23.0
76.5
13.2
1.61
10.1
1.66
9.09 27.7
20.4 61.4
2.66 7.36
10.8 27.4
5.51
10.4
24.9
3.34
15.3
3.34
31.1
81.2
9.91
41.1
8.60
15.5
2.34
11.1
2.21
0.78
1.67
0.21
0.92
0.15
0.36
0.05
0.34
0.06
2.58
4.62
2.03
0.27
1.20
0.21
0.44
0.06
0.34
0.06
1.37
3.38
0.42
1.97
0.34
0.74
0.09
0.53
0.08
5.53
2.70
0.35
1.66
0.29
0.69
0.10
0.52
0.09
2.29
6.53
0.89
4.51
0.80
1.91
0.25
1.62
0.26
9.91
2.21
6.26
0.98
5.52
0.84
2.17
6.10
0.90
4.91
0.69
2.77 1.25
7.26 3.16
1.06 0.41
5.49 2.13
0.76 0.26
Tm
Yb
Lu
Rather, they suggest a complex open-system interaction generated. For this purpose, we have compared the
between old crustal materials and mantle-derived alka- mineralogy, geochemistry and Sr and Nd isotope geol-
line ¯uids, either a melt or a supercritical hydrous ¯uid. ogy of two massifs, Soustov and Gremiakha-Vyrmes
The aim of this report is to decipher the mechanism.s) .Fig. 1). They are presumably coeval and represent the
by which such anomalous syenites could have been two extremes of the compositional spectrum of the Early

556
Table 2 Sr and Nd isotope of Soustov .SO) syenites, the tonolitic gneiss used for modeling .SU-19) and Gremiakha-Vyrmes .GV) rocks.
Identi®cation numbers are the same as in Table 1
Rb Sr
.ppm) .ppm)
87Rb/86Sr 87SR/86Sr 87SR/86Sr
.1,880 Ma) .1,880 Ma) .ppm) .ppm)
eSr
Nd
Sm
147Sm/144Nd 143Nd/144Nd 143Nd/144Nd eNd
.1,880 Ma)
SO-1
SO-2
SO-3
SO-4
SO-5
SO-6
SO-7
276
325
302
247
273
217
369
130 6.262
0.885483 0.716066
0.872685 0.721358
0.759940 0.720905
0.954790 0.718197
0.975213 0.720107
0.797171 0.718302
0.791956 0.723346
0.717163 0.709318
0.707162 0.702372
0.712491 0.704088
0.735494 0.705246
0.725758 0.702288
0.723143 0.705092
0.744119 0.703438
0.708997 0.703319
0.795050 0.703163
0.722267 0.703518
0.704752 0.703921
0.704558 0.703865
0.704311 0.703891
0.704345 0.703946
196
271
265
226
253
227
299
100
30.4
67.0 11.5
106 18.3
39.9
92.4 16.5
76.5 13.2
73.1 11.1
4.02 16.6
5.46 0.1086
0.511481
0.511464
0.511574
0.511446
0.511625
0.511469
0.511312
0.511997
0.511841
0.511745
0.511430
0.511558
0.511783
0.511709
0.511712
0.512330
0.511635
0.511862
0.511788
0.511797
0.511837
0.510138
0.510178
0.510293
0.510124
0.510287
0.510182
0.510178
0.510148
0.510272
0.510217
0.509927
0.510228
0.510152
0.510143
0.510261
0.510049
0.510158
0.510159
0.510104
0.510130
0.510127
)1.304
)0.521
1.724
)1.588
1.609
)0.437
)0.522
)1.109
1.324
0.248
)5.432
0.462
)1.039
)1.198
1.098
)3.059
)0.917
)0.897
)1.980
)1.469
)1.517
171 5.593
608 1.443
83.5 8.745
86.1 9.429
217 2.915
425 2.536
760 0.290
812 0.177
755 0.311
198 1.118
266 0.867
103 0.667
164 1.504
0.1040
0.1036
7.06 0.1069
0.1082
0.1040
0.0917
0.1494
SUB-19 76.1
GV-1
GV-2
GV-3
GV-4
GV-5
GV-6
GV-7
GV-8
GV-9
GV-10
GV-11
GV-12
GV-13
49.7
81.0
76.2
79.7
23.7
84.9
1.13
25.5
45.9
10.8
27.4
17.1
15.3
41.1
19.1
9.62 0.1268
2.21 0.1235
5.51 0.1215
3.05 0.1075
3.34 0.1319
8.60 0.1265
3.70 0.1173
42.0
)0.06
39.8
16.3
14.6
12.3
17.4
23.2
22.4
22.7
23.5
89.9 1240 0.210
76.0
59.1
10.4
65.3 3.396
247 0.693
976 0.031
16.1 0.026
7.60 2.32 0.1844
35.6
29.7
7.04 0.1194
6.77 0.1377
0.14
0.94 0.21 0.1361
58.6 13.1 0.1348
37.6 8.59 0.1382
4.20 783 0.016
4.29 841 0.015
Proterozoic alkaline magmatism of Kola, highly anom- rated association forms a body in the north of the massif
alous in the case of Soustov and derived from nearly composed of peralkaline granite and syenites, rarely
pure mantle melts in the case of Gremiakha-Vyrmes. We feldspathoidal, with a ma®c mineralogy consisting either
have veri®ed that both massifs are the same age by of aegirine-arfvedsonite, aegirine-aenigmatite, or aegi-
datingthem with the single-zircon stepwise-evaporation
207Pb/206Pb and whole-rock Rb-Sr methods. To model
the variations of trace elements and isotope ratios during
in®ltration metasomatism, we have developed a nu-
merical method enablingus to propose a percolation
mechanism capable of producingthe anomalous chem-
ical and isotopic signature of the Soustov rocks.
rine-richterite.
The Soustov intrusion
The Soustov Intrusion lies in the vicinity of the southern
contact of the giant Devonian agpaitic complex of
Khibina, within the axial zone of the Imandra-Varzuga
volcanic belt .Fig. 1). It is a lens-shaped 2 ´ 10-km body
whose contacts are roughly parallel to the southern
boundary of Khibina. The Soustov Intrusion is com-
posed almost exclusively of syenites that, in most cases,
are feldspathoidal and peralkaline. Non-feldspathoidal
syenites are sparse, and a narrow rim of quartz syenite
forms the contact zone with the country rock.
Geological setting
The Gremiakha-Vyrmes Complex
The Gremiakha-Vyrmes Complex .Fig. 1) is located
within the Kola-Norwegian part of the Central Kola
terrane, inside the N-S fault zone separatingPechenga
from the Imandra-Varzuga regions. It comprises three
rock associations characterised by the presence of ultr-
abasic rocks, foidolitic basic and intermediate rocks, and
peralkaline silica-saturated and oversaturated rocks.
The ultrabasic association is composed of peridotites,
pyroxenites, gabbros and anorthosites with minor
feldspathoid-bearingsyenites and monzodiorites. The
foidolitic association consists of melteigites, ijolites,
juvites and nepheline syenites. The foidolites form a
strati®ed body, with alternatinglayers of melanocratic
and leucocratic varieties. Nepheline syenites, either ag-
paitic or miaskitic, de®ne a net of veins cross-cuttingthe
foidolites. They usually consist of foyaites that may lo-
cally grade into foyaitic pegmatites or alkaline syenites
.Kukharenko et al. 1971). The peralkaline silica-satu-
Analytical methods
Whole-rock major-element determinations were performed by
X-ray ¯uorescence after fusion with lithium tetraborate. Typical
precision was better than ‹1.5% for an analyte concentration of
10 wt.%. Trace-element determinations were done by ICP-mass
spectrometry .ICP-MS) with a Perkin Elmer Elan-5000 spectrom-
eter usingRh as internal standard. Precision was better than ‹2%
and ‹5% for analyte concentrations of 50 and 5 ppm, respectively.
Samples for Sr and Nd isotope analyses were digested in a clean
room usingultra-clean reagents and analysed by thermal ionization
mass spectrometry .TIMS) in a Finnigan Mat 262 spectrometer
after chromatographic separation with ion-exchange resins.
Normalisation values were 86Sr/88Sr=0.1194 and 146Nd/144Nd=
0.7219. Blanks were 0.6 and 0.09 ngfor Sr and Nd, respectively.
The external precision .2r), estimated from the analysis of ten
replicates of the standard WS-E .Govindaraju et al. 1994), was

557
Fig. 1 Location and schematic maps of Gremiakha-Vyrmes .GV) calculated from the model of Stacey and Kramers .1975). Mass
and Soustov .SO). The giant Palaeozoic massifs of Khibina .Kh) fractionation was corrected by multiplyingby Ö.207/206). Standard
and Lovozero .LV) are also shown
errors for each step were calculated accordingto the formula:
SE=2*r/Ön. However, the 2r con®dence interval for the ®nal age is
given by [X ) t.0.025)r/Ön, X+t.0.025)r/Ön], where X and r are the
average and standard deviation of measured steps, n the number of
steps, and t.0.025) is the upper .0.025) point of the t distribution for
n±1 degrees of freedom.
better than 0.003% for 87Sr/86Sr, and 0.0015% for 143Nd/144Nd.
87Rb/86Sr and 147Sm/144Nd were directly determined by ICP-MS
.Montero and Bea 1998), with a precision, estimated by analysing
ten replicates of the standard WS-E, better than 1.2% and 0.9%
.2r) respectively.
Major-element analyses of minerals were obtained by wave-
length dispersive analyses with a Camebax SX-50 electron mi-
croprobe usingsynthetic standards. Acceleratingvoltage was 20 kV
and beam current was 15 nA. Coecients of variation were close to
‹1%, ‹2.5%, and ‹5% for 10 wt.%, 1 wt.%, and 0.25 wt.%
analyte concentrations, respectively. REE analyses of accessories
were done by electron microprobe usingsynthetic standards.
Precision was close to ‹4% for 1 wt.% conc. and ‹10% for
0.25 wt.% conc. REE analyses of major minerals were performed
by LA-ICP-MS with a precision close to ‹5% and ‹10% for
50 ppm and 1 ppm, respectively .Bea et al. 1996).
Zircons were dated with the single-grain stepwise-evaporation
207Pb/206Pb method developed by Kober .1986). Lead was collected
on the ionization ®lament for 20±30 min and then analysed in ®ve
blocks with seven scans per block. Data acquisition was performed
in dynamic mode .peak hopping), using a secondary electron
multiplier .SEM) as detector with the 206±204±206±207±208 mass
sequence. The mass-ratio 204/206 was monitored to detect and,
if necessary, correct for common lead, whose composition was
Isotopic dating
Previous age determinations of the Gremiakha-Vyrmes
Complex range from 1,865‹45 Ma .Puskharev 1990) to
1,930‹52 Ma .Kukharenko et al. 1971). A recent Pb/Pb
isochron in apatites yielded 1911‹87 Ma .Savatenkov
et al. 1998). For this study, we analysed ten zircon grains
with the Kober method, six from a syenite of the
foidolitic association and four from a syenite of the
peralkaline association, which yielded a mean age of
1,884‹6 Ma .Table 3; Fig. 2). Fourteen samples from
the three main facies, includingperidotites, ijolites,
nepheline syenites and peralkaline granitoids, ®t well
into a 87Sr/86Sr vs 87Rb/86Sr regression line with a slope
correspondingto 1,870 Ma and an intercept on the
y

558
Table 3 Single-zircon stepwise
evaporation data of zircons
from Gremiakha-Vyrmes and
Soustov
Zircon
204Pb/206Pb
207Pb/206Pb
204Pb/206Pb
.corrected)
Error .2s%)
Age .Ma)
Gremiakha-Vyrmes
1
2
3
4
5
6
7
8
9
10
0.000643
0.000647
0.000658
0.000658
0.000646
0.000355
0.000653
0.000242
0.000397
0.000662
0.123233
0.114835
0.115645
0.115269
0.115250
0.116216
0.114416
0.115420
0.116373
0.114885
0.115119
0.28
0.34
0.21
0.13
0.22
0.29
0.29
0.16
0.59
0.41
1,877‹5
1,890‹6
1,884‹4
1,884‹3
1,899‹4
1,871‹6
1,886‹5
1,888‹3
1,878‹It
1,882‹8
0.124358
0.124127
0.124115
0.124908
0.118926
0.123944
0.118530
0.119959
0.123763
Mean
1,884‹6
Soustov
0.000019
0.000035
0.000087
0.000071
0.000459
1
2
3
4
5
0.113938
0.114748
0.115871
0.115378
0.120418
0.113959
0.114558
0.114979
0.114696
0.114228
0.82
0.36
0.43
0.28
0.35
1,863‹15
1,873‹7
1,879‹8
1,875‹5
1,868‹7
Mean
1,872‹8
axis of 87Sr/86Sr=0.7038 .Table 2; Fig. 2). We thus
consider the zircon age of 1,884‹6 Ma to represent the
best estimate for the crystallisation age of the Gremia-
kha-Vyrmes Complex.
The only previous chronological data of Soustov
were a discordant zircon age of 1,745 Ma and a
87Rb/86Sr vs 87Sr/86Sr regression line with poorly ®tting
samples that yielded 2,000 Ma, with 87Sr/86Sr2,000
Ma
showinga high dispersion, from 0.701 to 0.719 .Batieva
et al. 1983). We have analysed ®ve zircon grains which
yielded a mean age of 1,872‹8 Ma, slightly younger
than in the case of Gremiakha-Vyrmes .Table 3; Fig. 2).
In the Sr isotope evolution diagram .Table 2; Fig. 2),
seven whole-rock samples ®t into a regression line with a
slope correspondingto 1860 Ma and an intercept on the
y axis of 87Sr/86Sr=0.7215, a value notably higher than
in the case of Gremiakha-Vyrmes.
Our zircon data show, therefore, that the Gremiakha-
Vyrmes and Soustov bodies formed within a narrow
interval of ꢁ12 Ma, so that we can safely assume they
belongto the same magmatic episode.
Mineralogy and petrography of syenites
Gremiakha-Vyrmes
The three rock associations of Gremiakha-Vyrmes
include syenites. Those belonging to the foidolitic asso-
ciation are either agpaitic or miaskitic foyaites, with
nepheline as the only feldspathoid and albite as the
dominant feldspar. The ma®c assemblage is composed
of tschermakitic aegirine-augite, hastingsitic amphibole,
and annitic biotite. Amphibole and biotite usually
replace early crystals of clinopyroxene. The accessory
assemblage is composed of Mn-rich ilmenite, titanite,
apatite, zircon, rare baddeleyite, abundant crystals of a
Fig. 2 Single-zircon stepwise 207Pb/206Pb ages and Sr isotope
evolution diagram for Soustov and Gremiakha-Vyrmes rocks.
Note that the age is virtually identical, but Soustov rocks have a
much higher initial 87Sr/86Sr

559
Si-Ti-Fe-Ba mineral .benitoõte?), barite, and rare but
conspicuous celestine. The only REE-saturated minerals
found under the SEM are scarce crystals of metamictic
Ca-poor allanite.
The syenites from the ultrama®c association are
orthoclase-rich pulaskites and akerites, infrequent sye-
nogabbros, with rare or no nepheline. The ma®c asso-
ciation is formed of fayalite, ferropargasite, ferrian
hedenbergite and annitic biotite. The accessory assem-
blage comprises abundant magnetite that forms simp-
lectitic intergrowths with ferrian hedenbergite, ilmenite,
apatite, zircon and rare barite. So far, we have found no
REE-saturated minerals.
The syenites from the saturated peralkaline associa-
tion have very rare or no feldspathoids. The major
mineral assemblage is composed of albite, microperthite,
aegirine and alkaline amphibole, either riebeckite, arfv-
edsonite, or richterite, sometimes accompanied by
aenigmatite. The accessory assemblage consists of
ilmenite, apatite, zircon and rare barite. As REE-satu-
rated accessories, they have rare crystals of a primary-
appearingTi-LREE-Ca mineral .knopite?) and more
abundant crystals of a secondary-appearingLREE-Ca
carbonate, sometimes with elevated Sr contents .khan-
neshite?).
Fig. 3 Total-alkalis vs silica diagram of Soustov and Gremiakha-
Vyrmes rocks. The dashed rectangle encloses the syenites of both
massifs
.AI=mol Na2O+K2O/Al2O3) close to 1. Soustov
syenites are always silica-undersaturated, but Gremia-
kha-Vyrmes syenites, especially those related to the
peralkaline granitoids, can also be saturated or slightly
Soustov
Soustov syenites, in contrast to Gremiakha-Vyrmes oversaturated .Table 1).
syenites, have sodalite and analcite, probably secondary,
Apart from the notorious di€erences in halogens, the
instead of nepheline. The dominant minerals are alkali syenites of both massifs show a marked contrast in their
feldspars, albite and microperthite accompanied by trace-element composition .Table 1). Soustov syenites
subordinate amounts of aegirine, katoforite, annitic are approximately ®ve to ten times richer in LREE,
biotite, and Ti-rich andraditic garnet. Secondary mus- HREE, Y, Th, U, Zr, Hf, Nb, Ta, Sn, Be, Li, Rb, Tl, Pb
covite is also found locally, replacingmicroperthite. The and Cs, but considerably poorer in Ba and, to a lesser
accessory assemblage is by far more abundant and var- extent, in Sr, V and Sc. Their chondrite-normalised REE
ied than in the Gremiakha-Vyrmes syenites. It consists patterns decrease steeply from La to Sm, have a
of ¯uorite, magnetite, titanite, apatite, zircon, minor pronounced negative Eu anomaly, and then decrease
baddeleyite, allanite, LREE-Ca silicates .kaynosite?), gradually from Gd to Lu .Fig. 4).
LREE-Nb-silicates .nacareniobsite?), LREE-Al-sili-
In contrast, the chondrite-normalised REE patterns
cates .allumobrittolite?), Y-HREE-silicates .thalenite?), of the Gremiakha-Vyrmes syenites decrease moderately
LREE-carbonates .parisite and bastnaesite), thorite, from La to Sm, have a very small or no Eu anomaly,
pyrochlore, cheralite, chalcopyrite, galena and barite. either positive or negative .except in the case of those
Most of these minerals, especially the REE-rich acces- related to the ultrama®c association, which have a
sories, have textures indicatingcrystallisation during
late-magmatic conditions.
strongpositive Eu anomaly), decrease from Gd to Er
and are either ¯at or show a progressive enrichment
from Tm to Lu .Fig. 4).
In addition to the di€erences in absolute abundances,
there is also a remarkable contrast in some meaningful
elemental ratios. In Soustov, Nd/Th it is close to 3
.Fig. 5), a value very common in crustal rocks .Bea and
Montero 1999; Bea et al. 1999). The syenites and related
Geochemistry
Major and trace elements
The syenites of both massifs have a similar major- rocks of Gremiakha-Vyrmes, in contrast, have Nd/
element composition, except for the much higher F and Th>15, as usually happens in mantle-derived rocks. In
Cl but slightly lower P2O5 and TiO2 of the Soustov Soustov, the syenites have K/Rb ꢀ 140±190 .Fig. 5), a
syenites .Table 1). In both cases, they consist of inter- range rarely found in rocks other than highly di€eren-
mediate rocks with total alkalis in the range of 10±16 tiated crustal leucogranites. Gremiakha-Vyrmes sye-
wt.% .Fig. 3), Na2O>K2O, and an agpaitic index nites, on the other hand, always have a K/Rb>350

560
Fig. 6 ꢀNd1,880
vs ꢀSr1,880
diagram of Soustov and
Ma
Ma
Gremiakha-Vyrmes massifs and their host rocks. Gremiakha-
Vyrmes rocks plot into a mixinghyperbola pointed towards their
host gneisses, indicating bulk assimilation. Soustov syenites, in
contrast, de®ne a weak positive correlation and do not show
Fig. 4 Chondrite-normalized REE patterns of syenites. Note the
highest REE contents and negative Eu anomaly characteristic of evidence of bulk mixingwith their host rock. Data from Table 2;
Soustov rocks
for Gremiakha-Vyrmes rocks, data were taken from Savatenkov
et al. .1998)
.Fig. 5), reaching exceptionally high values in the sye-
nites associated with the peridotite-gabbro complex. The vation. The initial 87Sr/86Sr is around 0.7038 and the
average Nb/Ta ratio is close to 12 in Soustov and 22 in ꢀNd1,880 Ma ranges from )3 to 1.3 .Table 2). Both
Gremiakha-Vyrmes, nearly matchingthe average crustal
and mantle values, respectively .Green 1995).
parameters are negatively correlated, de®ning a para-
bolic trend towards the host gneisses .Fig. 6) that
We can therefore conclude that, whereas Gremiakha- strongly suggests bulk contamination with crustal
Vyrmes syenites have a trace-element signature charac- materials.
teristic of mantle-derived magmas, Soustov syenites Soustov syenites have a Nd isotope composition
have many trace-element features characteristic of similar to that of Gremiakha-Vyrmes, with ꢀNd1,880 Ma
crust-derived magmas.
Sr and Nd Isotopes
between )1.6 and 1.7, but are highly enriched in radio-
genic Sr, with a 87Sr/86Sr1,880 Ma from 0.7160 to 0.7233.
In contrast to Gremiakha-Vyrmes, Soustov syenites
show a weak positive correlation between ꢀSr1,880 Ma and
ꢀNd1,880 Ma .Fig. 6). Remarkably, each of these two
The Sr and Nd isotope composition of Gremiakha- parameters is well correlated, with a de®nite group of
Vyrmes rocks is totally compatible with a mantle deri- elements and meaningful interelemental ratios.
Fig. 5 Nd-Th and K-Rb ratios
of rocks from Soustov and
Gremiakha-Vyrmes. Whereas
Gremiakha-Vyrmes rocks have
mantle-like values, Soustov
have values typical of conti-
nental, highly di€erentiated
rocks .see text)

561
mantle, with little contribution from the crust. None-
theless, this supposition lacks consistency with the trace-
element and Sr isotope data, so that in the sections
below we shall look for an alternative explanation.
The correlation between ꢀNd1,880 Ma and ꢀSr1,880 Ma
in Soustov syenites is not negative, but weakly positive,
and does not show the mixingtrend towards their host
gneisses depicted by Gremiakha-Vyrmes rocks .Fig. 6).
It is also important to remark that the 87Sr/86Sr1,880 Ma
of Soustov syenites are higher than in any other Kola
rock, except in a few similar alkaline rocks. The isotopic
signature of Soustov syenites, therefore, could be pro-
duced neither by assimilation nor by partial melting
of crustal materials, requiringinstead an open-system
process able to cause a selective enrichment in radio-
genic Sr.
Given the coincidence in space and time, we accept as
a startinghypothesis that alkaline ultrama®c magmas
originating in the mantle and crystallising within the
crust, such as those of the Gremiakha-Vyrmes Complex,
released a metasomatic agent which ®rst percolated
through crustal materials and, once it was profoundly
modi®ed by reaction with these materials, produced the
silica-undersaturated alkaline magmas of Soustov. In
the paragraphs below we shall explore this idea.
Fig. 7 ꢀSr1,880 Ma and ꢀNd1,880 Ma vs selected trace elements and
interelemental ratios. The good correlation between Sr1,880 Ma and
Rb, Cs, K/Rb and Eu/Eu* is possibly caused by the simultaneous
destruction of biotite and monazite. The correlation between
ꢀNd1,880 Ma and Sm, Mg, Ni, and P is related to the destruction of
amphibole and apatite .see text)
Numerical modelingof trace-element and isotope ratio
variations duringin®ltration metasomatism
The percolation of the metasomatic ¯uid, either a melt
or an H2O-CO2 supercritical ¯uid, would have fenitised
the percolated crustal materials, causingthe destruction
of some pre-existingminerals and the precipitation of
new ones. Duringthis process, chemical elements would
have been leached from the protolith into the ¯uid and/
or precipitated from the ¯uid into the newly formed
minerals, the net balance dependingon the composition
and relative proportion of the ¯uid and minerals in-
volved.
ꢀSr1,880 Ma bears a good positive correlation with ele-
ments such as Rb, Cs, Tl, Be, Pb, U, Th and the LREE,
and negative with K/Rb and Eu/Eu* .Fig. 7). ꢀNd1,880 Ma
has an excellent positive correlation with the MREE,
Cr, Ni, Mg, Ti, Nb, P and Nb/Ta .Fig. 7).
The variation in the concentration of a given trace
element in the ¯uid with an in®nitesimal increment of
the percolation process is:
Discussion: a model for the genesis of Soustov syenites
To discuss the petrogenesis of the Gremiakha-Vyrmes
Complex in detail is beyond the scope of this paper. For
our purposes, it is enough to consider that all the
available evidence points to variably fractionated man-
tle-derived melts that experienced limited contamination
with crustal materials.
dꢂClW † ˆ CsdWs ‡ CpdWp;
ꢂ1†
where W, Ws and Wp, and Cl, Cs, and Cp are the masses
and the element concentrations of the liquid, the phases
that dissolve, and the phases that precipitate, respec-
tively. It follows that:
The presence of crustal components in Soustov sye-
nites is much more evident than in the case of Gremia-
kha-Vyrmes, as revealed by the high proportion of
radiogenic Sr, the crustal values of Nd/Th, K/Rb and
Nb/Ta, and the shape of the chondrite-normalised REE
patterns .Fig. 4). The presence of mantle components is,
however, more dicult to recognise. Given the low
values of the ꢀNd1,880 Ma .from )1.6 to 1.7), one would
expect, in theory, a signi®cant part of the Nd currently
present in Soustov syenites to have derived from the
dW ˆ dWs ‡ dWp;
W s ˆ TRs;
ꢂ2†
ꢂ3†
ꢂ4†
W p ˆ TRp;
where T is the mass of protolith metasomatised in each
incremental step, considered as a constant, and Rs and
Rp are respectively the amounts of dissolved and pre-
cipitated substances per each T mass unit of protolith.

562
TransformingEq. 1 we obtain:
centration of the element in the dissolvingphases remain
constant.
W dCl ‡ CldW ˆ CsdW s ‡ CpdW p;
ꢂ5†
SubstitutingEq. 2 in Eq. 5 and simplifyingwe then have:
Applications to Soustov: startingassumptions
À
Á
dWs
W
dWp
W
dCl ˆ ꢂCs À Cl†
‡ Cp À Cl
;
ꢂ6†
UsingEq. 10 we have checked whether mantle-derived
¯uids in®ltratingrocks, such as those composingthe
Archean crust of Kola, could acquire a geochemical
signature similar to Soustov syenites. Owing to the na-
ture of the in®ltration metasomatism, there are more
variables in Eq. 10 than we can determine, especially the
modal fraction of phases that dissolve and precipitate.
Similarly, it is unreasonable to expect the in®ltrated
rocks to be homogeneous in age, initial isotope com-
position, nature, modal fraction and composition of
minerals. Therefore, the calculations that follow are not
intended to be a quantitative model of how contami-
nation duringin®ltration metasomatism has proceeded,
but an attempt to determine whether we can simulate the
chemical and isotopic anomalies of Soustov syenites
startingfrom a set of reasonable assumptions.
Dividingboth members of Eq. 6 by Cl, and substituting
Eq. 3 and Eq. 4 we see that:
À
Á À
Á
Cp À Cl d RpT
dCl ꢂCs À Cl† dꢂRsT†
ˆ
‡
;
ꢂ7†
Cl
Cl
W
Cl
W
Equation 7 is integrable if T is expressed as a fraction of
the initial mass, W0, of metasomatic agent. In this case
we have
À
Á
Z
Z
Z
C
W 0‡Ws
W 0‡Ws
l dCl ꢂCs À Cl†
RsdW
W
RpdW
W
Cp À Cl
ˆ
‡
;
Cl0
W 0
W 0
Cl
Cl
Cl
ꢂ8†
ꢂ9†
and therefore
ꢀ
ꢁ
ꢀ
ꢁ
À
Á
Cl
Cl0
Cs
Cl
ln
ˆ
À 1 lnꢂ1‡TRs† ‡ ꢂK À 1†ln 1 À TRp
;
For the ¯uid solute, we assumed a composition sim-
ilar to that of the pegmatites of Gremiakha-Vyrmes
.Table 4). For the solid, we assumed a composition
similar to a sample of a common biotite- and amphibole-
bearingTTG gneiss of Kola whose whole-rock chemical
and isotopic composition, as well the trace-element
composition of major and accessory minerals, has been
thoroughly studied .Table 4). Based on the detailed de-
scriptions of mineralogical changes during fenitisation
of quarztfeldspathic rocks .McKie 1966; Kresten and
Morogan 1986; McLemore and Modreski 1990; Sindern
and Kramm 2000), we also assumed that the fenitisation
process caused the total destruction of biotite and
hornblende to form aegirine, alkali-amphibole and
where K is the solid/liquid partition coecient for pre-
cipitatingphases. Cl/Cs, however, does not represent a
partition coecient because the metasomatic agent and
protolith are not in chemical equilibrium. If Rs=0, the
above expression is fully equivalent to Rayleigh's
equation for fractional crystallisation. If there is more
than one phase that precipitates or dissolves, Eq. 9 can
be expressed as:
ꢀ
ꢁ
ꢀ
ꢁ
ꢂ
ꢃ
Xi
Xj
À
Á
À
Á
Cl
Cl0
Csi
Cl
ln
ˆ
À 1 ln 1‡TRis
‡
Kj À 1 ln 1 À TRjp
;
1
1
ꢂ10†
which can be used as longas the rates of precipitation
and dissolution, the partition coecients, and the con- K-feldspar, and the partial albitization of the plagio-
Table 4 Parameters used in numerical modeling. a Solute com- and monazite, which were determined by electron microprobe. The
isotopic composition of minerals at 1,900 Ma has been calculated
position of the percolating¯uid .assumed to be equal to a nepheline
pegmatite of Gremiakha-Vyrmes), and the mode and composition from their current Rb/Sr and Sm/Nd ratios and the isotopic
of minerals of a TTG gneiss used as protolith. The composition of composition of the gneiss SUB-19 .Table 2), 87Sr/86Sr2,900
minerals has been determined by LA-ICP-MS, except for allanite Ma=0.704972 and 143Nd/144Nd2,900=0.509135
Fluid
Gneiss SUB-19
Pl
K-Fsp
Bio
0.14
450
19
±
±
±
±
±
Hnb
0.10
25
189
Apa
0.003
±
±
All
Mon
Modal fraction
Rb
Sr
La
0.49
10
1450
12.6
8.3
0.04
150
690
0.00009
±
±
53400
105000
17900
1890
100
0.00005
±
±
132000
295200
13000
3600
600
35
965
13.0
32.6
15.7
4.0
1.65
0.705000
0.510110
1.2
19.9
65.0
Ce
Nd
Sm
Eu
0.95
0.33
0.10
1.5
34.0
29.3
10.7
1.9
239
532
311
89
2.0
0.34
0.75
0.705254
0.509787
87Sr/86Sr
143Nd/144Nd
0.713838
0.510329
2.540199
0.710339
0.510572
±
0.511434
0.509464
0.510227
b Modal fraction of minerals that dissolve and precipitate
Dissolve
pl
K-fsp
0.00
bio
hnb
apa
0.003
tit
all
0.00009
mon
0.00005
0.15
0.25
0.14
Na-cpx
0.1
0.1
Na-amp
0.1
Precipitate
alb
K-fsp
0.08
0.01

563
clase; the assumed modal fraction of minerals that dis-
solve and precipitate is shown in Table 4.
The equilibrium distribution coecients between
alkaline ¯uids and minerals are virtually unknown.
Therefore, we approached them empirically by analysing
the albite/Na-clinopyroxene, albite/Na-amphibole, and
albite/titanite concentration ratios in Soustov syenites,
and the albite/whole-rock and K-feldspar/whole-rock
concentration ratios in a nepheline syenite pegmatite
from Miass. If we assume that mineral/whole-rock
concentration ratios in the pegmatite are close to real
mineral/alkaline ¯uid distribution coecients, it is a
simple matter to calculate a data set of distribution
coecients for all minerals involved .Table 5). The
strongest point of this approach is that the mineral/
mineral concentration ratios were determined in Soustov
syenites; therefore, the error derived from assumingthat
mineral/whole-rock concentration ratios in the pegma-
tite represent mineral/alkaline ¯uid partitioningwould
Fig. 8 Rb vs. 87Sr/86Sr paths calculated from Eq. 10 for various age
a€ect all the calculated distribution coecients in nearly di€erences between the metasomatic event and the percolated
materials. For the assumed proportions of phases that precipitate
and dissolve, the best ®t is obtained for an age di€erence of
1,000 Ma, which corresponds to an age of 2,900 Ma for the gneiss
the same way and so have little in¯uence on the model
.see below).
The Sr and Nd isotope composition of the protolith
minerals at a given time has been calculated from the
of feldspars and alkali amphibole. Figure 8 shows the
feasibility of in®ltration metasomatism beingresponsible
for changing the Sr isotope composition of the metaso-
matizingagent in a way that nearly exactly matches that
displayed by Soustov syenites, if, as stated before, we
assume that the metasomatic event is 1,000 Ma younger
than the in®ltrated rocks .Fig. 8). The calculation is
quite robust, so that moderate variations in the partition
coecients and the modal fraction of interveningphas-
es, either dissolvingor precipitatin,g do not exert a
strongin¯uence. It is worth mentioningthat, according
to this model, an age di€erence as low as 200 Ma still
can cause a perceptible enrichment in radiogenic Sr
.Fig. 8). The destruction of biotite would release Rb, Cs
and Tl into the ¯uid, thus producingthe good correla-
measured whole-rock isotope data and the LA-ICP-MS
measured Rb/Sr and Sm/Nd ratios of minerals .Table 4);
we assumed that all minerals had the same isotope
composition at the moment of their formation, equal to
that of the whole rock, roughly estimated at 2,900 Ma.
This age implies that the crustal rocks had an age of
ꢁ1,000 Ma when they were metasomatised. For the sake
of simplicity, we kept the same initial isotope composi-
tion in all models even when the age of the protolith
changed.
Sr isotopes
As percolation progresses, the ¯uid becomes enriched in
Rb and 87Sr relative to 86Sr, but depleted in total Sr, tion between these elements and the 87Sr/86Sr. This idea
owingto the destruction of biotite and the precipitation gains additional support from the ®nding that fenites
that developed from a granodiorite were progressively
depleted in 87Sr relative to 86Sr as the intensity of the
Table 5 Set of mineral/¯uid distribution coecients used for
modelingthe changes in the ¯uid composition duringthe fenitisa-
tion of the SUB-19 gneiss. Values have been calculated from albite/
whole-rock and K-feldspar/whole-rock concentration ratios mea-
sured in a nepheline syenite pegmatite from Miass, and from albite/
Na-clinopyroxene, albite/Na-amphibole, and albite/titanite con-
centration ratios measured in Soustov syenites .see text). Whole-
rock analyses by solution-nebulization ICP-MS; mineral analyses
by LA-ICP-MS
fenitisation increased .Sindern and Kramm 2000).
Rare earth elements
In crustal protoliths, the REE mainly reside in amphi-
bole and, especially, in accessory minerals .Bea 1996),
the behaviour of which duringfenitisation processes is
still poorly known. To appraise how the REE might
behave in a process of percolation, at least qualitatively,
we have built a model that, besides the destruction of
amphibole, involves the dissolution of the LREE-rich
accessories of the protolith, monazite, allanite and
apatite .Table 4) and the simultaneous precipitation of
REE-rich titanite, which seems compatible with the ac-
cessory mineralogy of fenitised granite rocks .Montero
Albite
K-feldspar
Aegirine
Riebeckite Titanite
Rb
Sr
0.02
2.5
0.3
3.0
0.02
1.25
0.3
0.35
0.37
0.58
0.29
0.59
0.15
3.0
±
±
La
Ce
Nd
Sm
Eu
Gd
0.07
0.06
0.08
0.15
1.47
0.10
0.05
0.05
0.07
0.14
1.62
0.10
0.94
1.07
1.09
1.63
1.47
1.6
14.1
13.3
16.08
21.7
19.3
25.2

564
et al. 1998). To avoid additional uncertainties arising colation process, the enrichment in 87Sr with respect to
from the behaviour of HREE-rich accessories, such as 86Sr beingcaused by the destruction of biotite, as dis-
zircon, we have limited the calculations to the LREE cussed in the above sections, and the enrichment in
and MREE. The results of applyingEq. 10 with the data 143Nd with respect to 144Nd beingcaused by the di€er-
from Tables 4 and 5 show that this model, however ential leachinginto the ¯uid of a mineral with high Sm/
crude, is able to simulate a REE composition for the Nd that is at the same time signi®cantly older than the
percolating¯uids that nearly exactly matches the com- metasomatic event. Our data on the tonalitic gneiss
position of Soustov syenites .Fig. 9), characterised by SUB-19 .Sm/Nd=0.25) reveal that the range of Sm/Nd
elevated LREE contents that decrease rapidly with in co-existingminerals varies widely, from 0.16 in
increasingatomic number and a small neagtive Eu
plagioclase to 0.36 in amphibole and 0.59 in apatite
anomaly. The main contributor of LREE, Th and U to .Table 4). If, after a signi®cant period of time from its
the ¯uid is monazite; since in the percolated gneiss a formation, the rock is percolated by a metasomatic
signi®cant part of this mineral is included within biotite, agent able to di€erentially leach the high-Sm/Nd min-
both minerals are most likely leached into the ¯uid in the erals, the ¯uid would be progressively enriched in 143Nd
vs 144Nd .and in LREE and 87Sr/86Sr, as mentioned
above). To model this mechanism, we have used Eq. 10
with the data from Tables 4 and 5, with the following
results.
same way, thus causinga good positive correlation be-
tween 87Sr/86Sr, Rb, Cs and Tl on one hand, and LREE,
U, Th, and the LREE, U and Th on the other hand .see
Fig. 7).
Amongthe minerals currently found in the SUB-19
gneiss, amphibole and apatite have the greatest in¯uence
on the enrichment in 143Nd/144Nd of the percolating
¯uid. When present, xenotime may cause even more
dramatic e€ects. Monazite does not in¯uence the Nd
isotope ratio, since it has nearly the same Sm/Nd ratio as
the whole rock. Accordingto our calculations, the ¯uids
are enriched in 143Nd/144Nd only when the age di€erence
between protolith and metasomatism is greater than
ꢁ850 Ma .Fig. 9). Small di€erences in the age of
the protolith can cause elevated di€erences in the
143Nd/144Nd of the ¯uid. Soustov syenites do not follow
a single path of Sm vs 143Nd/144Nd enrichment, but plot
between the paths correspondingto age di€erences of
900 Ma and 1000 Ma which is in concordance with
the model for Sr isotopes .Fig. 10). The idea that the
increased 143Nd/144Nd of the ¯uid may come from the
Nd isotopes
One of the most puzzlingfeatures of Soustov syenites is
the coexistence of high initial 143Nd/144Nd and 87Sr/86Sr
showinga weak positive correlation. If we accept that
high initial 143Nd/144Nd represents a mantle feature,
whereas high initial 87Sr/86Sr represents a crustal feature,
then we are presented with the paradox that Soustov
syenites are simultaneously enriched in both crustal and
mantle components. An alternative explanation, how-
ever, might be that both features are linked to the per-
Fig. 9 Chondrite-normalized REE patterns of the metasomatic
¯uid at the beginning of the process and after percolating a gneiss Fig. 10 Sm vs 143Nd/144Nd paths calculated from Eq. 10 for the
mass .1) twice the mass of the solute and .2) four times the mass of various age di€erences between the metasomatic event and
the solute, as calculated from Eq. 10. The shaded area represents the percolated materials. Note the strongdependence on the age.
Soustov syenites. Note how the solute LREE patterns and Soustov syenites do not ®t a single path, re¯ecting the heterogeneity
abundances approach those of Soustov
of the protolith, which must have an age between 900 and 1,000 Ma

565
destruction of amphibole and apatite ®nds additional changed the ¯uid composition. The destruction of bio-
support from the excellent positive correlation between tite, rich in radiogenic 87Sr, coupled with the precipita-
that parameter and such elements as MREE, Mg, Cr, Ni tion of minerals in which Sr is highly compatible, such as
and P .Fig. 7).
the feldspars, caused the increase in 87Sr/86Sr and the
decrease in elemental Sr in the ¯uid. The destruction of
biotite also caused the elevated Rb, Cs, Nb, low K/Rb,
etc., contents of the ¯uid. The dissolution of monazite
and allanite produced the elevated Th and LREE
contents, negative Eu anomalies and low NdTh char-
A hypothesis for the origin of Soustov syenites
Numerical modelingof the in®ltration metasomatism
has revealed that a mantle-derived metasomatic agent acteristic of Soustov melts.
derived from melts similar to Gremiakha-Vyrmes per-
The dissolution of minerals with elevated Sm/Nd
colatingthrough a ꢁ1,000 Ma older gneissic protolith such as hornblende and apatite caused the high
may acquire a chemical and isotopic composition similar 143Nd/144Nd in the ¯uid. The e€ect of biotite on Sr
to that of Soustov syenites. The metasomatic agent can, isotopes is perceptible when the age di€erence between
in principle, be an H2O-CO2 supercritical ¯uid or a melt. the metasomatic event and the protolith is as low as
The elevated F, Cl, CO2 and S contents of Soustov sye- 200 Ma, but in the case of Nd isotopes it requires at least
nites .F and Cl in Table 1; CO2 and S inferred from the 850 Ma. Numerical modellingwith the expression
abundance of accessory carbonates and sul®des) is a derived for in®ltration metasomatism, usingthe com-
strongpoint in favour of the ®rst alternative, since these position of a Gremiakha-Vyrmes pegmatite as the solute
elements, together with the alkalis and H2O, are the most composition of the ¯uid, and the mode and the mineral
important components of the volatile phase of neph- trace-element and isotope composition of a TTG gneiss
elinitic magmas .Bailey and Hampton 1990). Therefore, as representative of the protolith, indicates that the ¯uid
and takinginto account that the dynamic properties of would acquire a signature closely matching Soustov's if
supercritical ¯uids are more favourable for in®ltrating the gneiss age is ꢁ2.9 Ga, which is probably close to its
solid rocks that those of melts .Spera 1987), we suppose real age.
that the metasomatic agent was a supercritical ¯uid.
As the hydrous supercritical ¯uid percolated through
When released, the percentage of solute in the ¯uid the crust, its solute composition changed from alkali
must have been not less than 30% .Adam et al. 1997). salts to a silica-undersaturated alkaline melt. When the
As percolation progressed, the ¯uid would have dis- ¯uid cooled to a temperature of ꢁ600±550 °C, it reached
solved silica and other components from the materials the point at which vapour and melt are no longer mis-
that fenitised, so that the composition of the solute cible and it therefore split into two components, a va-
would have changed from alkali halides, carbonates and pour phase and a Cl- and F-rich silica-undersaturated
sul®des to a silica-undersaturated alkaline melt. Both silicate melt that crystallised to produce the sodalite-
phases, hydrous ¯uid and silica-undersaturated alkaline bearingsyenites of Soustov. It should be emphasised
melt, are completely miscible at moderate pressures if that the fate of the percolating¯uid does not have to be
the temperature is higher than a critical value that, in the the unmixingat relatively low temperatures of a silica-
case of nepheline, has been estimated at 550 °C at 15 kb undersaturated melt, but, dependingon the PT regime
.Bureau and Keppler 1999). It therefore seems that, once and composition of the percolated crust, it may also
the temperature dropped below the critical value, the cause local anatexis or mix in any way with pre-existing
¯uid would have split into two components, a vapour crustal melts, thus producingsilica-oversaturated mag-
phase with a small proportion of solute and a silicate mas. We tentatively suggest that the variability of
melt that crystallised into the ¯uorite- and sodalite- 87Sr/86Srinitial and the simultaneous occurrence of high
bearingleucocratic syenites with a strongcrustal imprint
that form the Soustov intrusion.
87Sr/86Srt and ꢀNdt occasionally found in peralkaline
granitoids .Bennet et al. 1984; Eby 1990; Montero et al.
1998) might thereby be explained.
At the present level of knowledge, the crustal perco-
lation model ®ts well with our observations, but it
should be considered as a mere workinghypothesis. To
re®ne it, we need to understand in detail the behaviour
of major and accessory minerals, as well as the chemical
and isotopic exchanges involved during fenitisation. We
hope that our current, ongoing research on some key
examples from Kola might contribute to this objective.
Conclusions
The trace-element and isotope signatures of Soustov
syenites re¯ect a strongcrustal in¯uence that cannot be
accounted for by the bulk assimilation of host gneisses.
We propose that Soustov melts derived from a metaso-
matic hydrous supercritical ¯uid, rich in alkali elements,
F and Cl, that percolated through Archean tonalitic
gneisses, fenitising them. The source of the ¯uid was a
mantle-derived ultrabasic alkaline magma similar to
those that formed Gremiakha-Vyrmes.
Acknowledgements The authors are indebted to A. Vinogradov for
his help with samples from Gremiakha-Vyrmes and fruitful
discussions about the geology of Kola, and to K. Mengel and
U. Kramm for their revisions of the manuscript. The help of
Christine Laurin in improving the English is gratefully acknowl-
The mineralogical changes caused by reaction among
the ¯uid and the minerals of the gneiss profoundly edged. This work bene®tted from the ®nancial support of NATO

566
.grant 974904), the Spanish DGICYT .grant PB98-1345) and the
Russian Foundation for Basic Research .grant 00-05-64229).
and four international laboratories. Geostand Newsletters
XVIII.2): 211±300
Green TH .1995) Signi®cance of Nb/Ta as an indicator of geo-
chemical processes in the crust-mantle system. Chem Geology
120: 347±359
Kober B .1986) Whole-grain evaporation for 207Pb/206Pb-age
investigations on single zircons using a double ®lament thermal
ionization source. Contrib Mineral Petrol 93: 482±490
Kresten P, Morogan V .1986) Fenitization at the Fen complex,
southern Norway. Lithos 19: 27±42
References
Adam J, Green TH, Sie SH, Ryan CG .1997) Trace element par-
titioningbetween aqueous ¯uids, silicate melts and minerals.
Eur J Mineral 9: 569±584
Bailey DK, Hampton CM .1990) Volatiles in alkaline magmatism.
Lithos 26: 157±165
Balashov YA .1996) The upper-mantle di€erentiation and mixing
Kukharenko AA, Bulakh AG, Ilyinsky GA, Shinkarev NF, Orlova
M .1971) Metallogeny of alkaline rock series of the Eastern
Baltic Shield .in Russian). Nedra, Leningrad
processes. New model on Sm.Nd isotopic data basis .in McKie D .1966) Fenitization. In: Tuttle OF, Gittins J .eds) Car-
Russian). Dok Akad Nauk 347: 81±85
Batieva ID, Belkov IV, Kravchenko MP, Latyshev LN, Pushkariev McLemore VT, Modreski PJ .1990) Mineralogy and geochemistry
bonatites. Wiley, New York, pp 261±295
YuD, Ulianenko NA .1983) The age of the Soustov alkaline
massif, Kola Peninsula .in Russian). Dok Akad Nauk 270: 931±
933
of altered rocks associated with lemitar carbonatites, central
New Mexico, USA. Lithos 26: 99±113
Mitrofanov FP, Pozhilenko VI, Smolkin VF, Arzamastsev AA,
Yevzerov VYu, Lyubtsov VV, Nikolaeva SB, Fedotov ZhA
.1995) Geology of the Kola Peninsula .Baltic Shield). Kola
Science Center, Apatity
Bea F .1996) Residence of REE, Y, Th and U in granites and
crustal protoliths; Implications for the chemistry of crustal
melts. J Petrol 37: 521±552
Bea F, Montero P .1999) Behavior of accessory phases and redis-
tribution of Zr, REE, Y, Th, and U duringmetamorphism and
partial meltingof metapelites in the lower crust: an example
from the Kinzigite Formation of Ivrea-Verbano, NW Italy.
Geochim Cosmochim Acta 63: 1133±1153
Montero P, Bea F .1998) Accurate determination of 87Rb/86Sr and
143Sm/144Nd ratios by inductively-coupled-plasma mass spect-
rometry in isotope geoscience: an alternative to isotope dilution
analysis. Anal Chim Acta 358: 227±233
Montero P, Floor P, Corretge G .1998) The accumulation of rare-
earth and high-®eld-strength elements in peralkaline granitic
rocks: the Galineiro orthogneissic complex, northwestern
Spain. Can Mineral 36: 683±700
Bea F, Montero P, Stroh A, Baasner J .1996) Microanalysis of
minerals by an Excimer UV-LA-ICP-MS system. Chem Geol
133: 145±156
Bea F, Montero P, Molina JF .1999) Ma®c precursors, peralumi-
nous granitoids, and late lamprophyres in the Avila batholith:
a model for the generation of Variscan batholiths in Iberia.
J Geol 107: 399±419
Bennet JN, Turner DC, Ike E, Bowden P .1984) The geology of
some northern Nigerian anorogenic ring complexes. Overseas
Geology and Mineral Resources n1/4 61, British Geological
Survey, London
Puskharev YD .1990) Megacycles in the evolution of the system
crust ± mantle .in Russian). Nauka, Leningrad
Savatenkov VM, Sulimov RB, Sergeev AV, Goncharov GN,
Pushkarev YD .1998) Sm-Nd, Rb-Sr, and Pb-Pb isotope syst-
ematics of basic and ultrabasic rocks of Gremyakha-Vyrmes:
the role of crust-mantle interaction in magma generation and
ore-formingprocess .in Russian). Zap Vseross Mineral Ova 5:
15±25
Bureau H, Keppler H .1999) Complete miscibility between silicate
melts and hydrous ¯uids in the upper mantle: experimental
evidence and geochemical implications. Earth Planet Sci Lett
165: 187±196
Eby N .1990) The A-type granitoids: a review of their occurrence
and chemical characteristics and speculations on their petro-
genesis. Lithos 26: 115±134
Sindern S, Kramm U .2000) Volume characteristics and element
transfer of fenite aureoles: a case study from the Iivaara alkaline
complex, Finland. Lithos 51: 75±93
Spera J .1987) Dynamics of translithospheric migration of meta-
somatic ¯uid and alkaline magma. In: Menzies MA, Hawkes-
worth CJ .eds) Mantle metasomatism. Academic, London,
pp 1±20
Govindaraju K, Potts PJ, Webb PC, Watson JS .1994) 1994 Report Stacey JS, Kramers JD .1975) Approximation of terrestrial lead
on Whin Sill Dolerite WS-E from England and Pitscurrie
Micrograbbro PM-S from Scotland: assessment by one hundred
isotope evolution by a two-stage model. Earth Planet Sci
Letters 26: 207±221