To understand relationships between the crystallisation consequence, the mineral structures, the morphology and density of defects in crystalline lattices, and the trapped component abundances, mineral separates were investigated. Ultrabasic rocks showing the highest abundances of trapped He appears to be the most interesting for this study. Table 5 comprises He isotope abundances in olivine, pyroxene and magnetite concentrates from olivinite SB-3 (Tables 1, 2). It should be noted that the whole-rock U, Th, and He data indicate a good retention, 50%, of radiogenic 4He* by this sample. As it is shown below in this Section, even better 3He retention is expected. Therefore the present-day measured 3He concentrations should approach those trapped initially. 3He concentrations are almost constant in the clinopyroxene and Ti-magnetite separates
but vary mildly in the olivine slightly increasing with its size. 3He concentrations in the pyroxene and in bigger fractions of the olivine are similar, 1810-9 cc/g. The Ti-magnetite shows much higher abundance of trapped mantle He. This enrichment could originate from the consequence of mineral crystallisation governing magnetite segregation after olivine and clinopyroxene. Rare gases are incompatible elements and a rare gas enrichment (~10 times) adequate to the mass balance (~10% of magnetite in the rocks) is expected in the last portion of basic melt producing magnetite. Also the crystalline lattice of magnetite shows channel-like interstitials between elementary crystal domains with cross size 2 A which is similar to diameter of He atom. In contrast to that, crystalline lattice of olivine is rather dense and defects/inclusions are needed to accommodate trapped species.
Generally olivine is used as a proper carrier of trapped He [e.g., Marty and Tolstikhin, 1998, and references therein]. Tacking into account high concentrations of trapped He and its rather good retention in magnetite (see Figure 22), this mineral can be recommended as a promising natural sampler of trapped rare gases. High 3He concentrations and low 4He/3He ratios in several other magnetite separates (e.g., samples KV-13, 14, 21; SB-7-9, LV-4, 5, Table 2) support this suggestion.
4He concentrations in the mineral separates varies more widely then 3He implying inhomogeneity of U (Th) concentrates, or different retention of radiogenic 4He*, or a contribution of U-bearing mineral, like perovscite (SB-10, Table 2) having the highest concentration of 4He among UAC samples. The last column in Table 5 shows hypothetical contributions of perovscite which are able to reconcile 3He and 4He inventories in all separates.
Step-wise heating experiments with Ti-magnetites and olivine (Figure 21) clearly indicate different settings for 4He *, always related to radiation damage tracks, and 3He. At low temperatures 800oC fluid inclusions are decripitated liberating a dominant portion of major fluid components, CO 2 and H 2 O, and variable portions of 4He* and 3He. In Ti-magnetite BEG-1 (Figure 21b) portion of 3He released within these low-temperature steps coincides with that liberated by crushing, ~10%. Because of the mean size of the post-crushing powder is about 0.001-0.01 cm, the inclusions in this mineral should be of a similar size or somewhat less.
4He* is mainly released under moderate temperatures, from 600 to 1100o C. Radiation tracks especially those crossing grain boundaries and/or fluid-bearing vesicles clearly play an important role stimulating 4He* loss under low and moderate temperatures.
In contrast to 4He *, trapped 3He is dominantly released under rather high temperatures, > 1100oC, almost together with destruction of the crystalline lattices. An adequate amount of 40Ar is also released under same high temperature so that measured 4He/40Ar* ratios are close to this ratio in the magmatic fluid 3 (Section 4.3.2). Vacuum conditions during step-wise heating experiments shows that large high-temperature portions of 3He (and relevant portion of 40Ar* ) are not accompanied by adequate amounts of the major volatile
components. The separation calls for extremely small volumes of trapped-He-bearing sites, somewhat between size of He atom and CO2 molecule.
More work is needed to understand and quantify observations discussed above. However the conclusion on a good trapping capacity and rare gas rateability of magnetites is fortified by the step-wise-heating data.
Figure 22 reveals two general tendencies: increasing of 4He/3He ratios in trapped helium (liberated by milling) from basic through alkaline to carbonatitic rocks and elevated ratios in samples from the Ozernaya Varaka, Khibiny, and Dike Complexes. These tendencies could result from varying post-crystallisation abundances of trapped helium governed by different degassing rates or trapping capacities of host minerals together with subsequent variable contributions of in-situ produced helium. 4He/3 He mill versus (U+0.24Th)/ 3Hemill plot allows to check this explanation (Figure 23). At first glance low and similar parent/daughter ratios for Kovdor and Seblyavr ultrabasic rocks, and slightly enhanced 4He/3Hemill in samples from the Kovdor (Figure 23) would have implied an isotopic heterogeneity of trapped helium, in contrast to previous conclusions (Section 4.1.4). However, a more detail look at the data shows, that Kovdor ultrabasic rocks (minerals) had lost a larger portion of radiogenic 4He* ( 35%) than relevant samples from Seblyavr ( 10%). Some portion of 4He* releasing from related a -tracks penetrates into vesicles already containing trapped He. A small contribution of this 4He* ~0.1 ( 4He *cal - 4He*meas ), could ensure the observed enhanced 4He/3He mill in all but one Kovdor samples. Moreover, this contribution is also recorded by 4Hemill / 4Hemelt ratios which are higher in the Kovdor samples (the Average 4Hemill / 4Hemelt = 0.22) relative to the Seblyavr (0.037). Enhanced 4He/3He mill ratios in carbonatites could also result from the particularly low retention of radiogenic 4He* in these rocks (see Figure 15) and its partial migration into vesicles.
The data presented in Figure 23 indicate that migration of in-situ produced He and its admixture to trapped He retained in vesicles do increase 4He/3He mill ratios. While Figure 23 comprises a limited data-set, this mechanism could be responsible for 4He/3He mill variations within a factor of 1000 seen in Figure 14 and 22 which include all available data. It should be emphasised that a great range of (U+Th)/ 3Hemill ratios is dominantly due to 3Hemill variations: U and Th concentrations cluster within a relatively narrow diapason (Section 4.1.4). For example, the average values for ultrabasic rocks from the Seblyavr, U = 6.2 ppm, Th/U = 3.4, are quite similar to these for the relevant samples from Dykes Complex, U = 4.1 ppm and Th/U = 3.8 (Table 2), whereas 3He concentrations in these Complexes differ by ~2 orders of magnitude.
Therefore melt degassing and trapping capacities of host minerals appears to be the principal parameters controlling present-day helium isotope composition both in mineral lattices and in vesicles. The compositional array basic alkaline carbonatitic rocks corresponds to the intrusion - crystallisation stages, and also to decreasing densities and viscosities of parent melts. These patterns predict a better degassing of the later intrusive stages, and the increasing 4He/3He mill ratios along the array, e.g., the Kovdor, Turiy, Vuoriyarvy and Ozernaya Varaka Complexes in Figure 22, are in full accord with this prediction. The degassing conditions during the last carbonatitic intrusion stage appears to be especially variable causing adequate variations of 4He/3Hemill and 3He (Figure 22). For example, 3He concentration in Kovdor Vuoriyarvy carbonatites varies within 2 orders of magnitude. A highly variable trapped He concentrations in carbonatites reflect multi-pulse intrusions of these rocks together with different degassing conditions at each pulse.
As it is seen in Figures 11, 14 and 16, the ultrabasic rocks preserve initial 4He/3He ratios better than the others. Figure 22 presents the Complexes in order of increasing average 4He/3Hemill in these rocks. Again the different degassing rate appears to be responsible for increasing of 4He/3Hemill from one Complex to another.
Degassing history is controlled, among other factors, by conditions of crystallisation differentiation in magma chambers, including depths of the chambers. These conditions could be partially restored using geological and petrographical data: position of intrusions within the cross section of supracrustal rocks, quenching phases near contacts and eruptive breccias in internal parts, fingerprints of explosive processes, structural and textural peculiarities of rocks especially in contact zones.
The above features imply that the Khininy alkaline massif and the Dykes Complex belong to the uppermost subsurface formations [Polkanov and U-Li-Zhen', 1961; Galakhov, 1975]; the dykes are considered as eruptive channels to volcanic explosion fields removed by erosion [Bulakh and Ivannikov, 1984]. This is in full agreement with the low 3He abundances and the high 4He/3He ratios indicating a substantial pre-crystallisation degassing [Figure 22, see also Figure 14).
In contrast, the Kovdor, Seblyavr, Vuoriyarvy, and Lesnaya Varaka Complexes composed by ultrabasic, alkaline and carbonatitic rocks and showing signatures of hypabyssal intrusions were formed at greater depths [Kukharenko et al., 1965], in agreement with He isotope data (Figures 23, 14). Mineral - fluid inclusions in early apatites (the Kovdor) recorded the minimal fluid pressure ~1500 atm Corresponding to 5 km depth [Sokolov, 1981].
Intermediate depths were suggested to other Complexes are considered to be formed at [Kukharenko et al., 1965]. This agrees with He isotope data for all but one Complex: according to Figure 23, the Tury Peninsula belongs to the least degassed group.
The degassing rate of a Complex could be recorded by the related development of host rocks, i.e., fenitization [Le Bas, 1989]. Fore example, the Lesnaya Varaka mainly composed by basic rocks with low 4He/3He ratios is surrounded by a thin rim of fenitised rocks which is in contrast to the mainly alkaline Ozernaya Varaka Complex having higher 4He/3He ratio and a thicker rim [Ikorsky et al., 1998].
Resuming, the simplest explanation of the available data envisages two processes: (i) trapping of initially isotopically homogeneous helium ( 4He/ 3He initial ~30,000) by growing crystals; concentrations of trapped He were controlled by progressive crystallisation, variable degassing and trapping capacities and (ii) subsequent migration of trapped and in-situ produced radiogenic He. Conclusion on a homogeneous pre-crystallisation fluid also followed from study of primary inclusions in silicate rock of the Kovdor Complex [Sokolov, 1981]. Several deep sources of helium having different isotopic compositions are not required to satisfy the data contrary to the inferences from Rb-Sr and Sm-Nd systematics [e.g., Kramm, 1993; Zaitsev and Bell., 1995].
Variable mixing of mantle and crustal materials during the magmatic and early post-magmatic stages followed by trapping of He with variable 4He/3He ratios have not been clearly recorded by the UAC rocks in contrast to the Monche layered intrusion [Tolstikhin et al., 1992]. However these processes can not been completely ruled out either. Ne isotope mixing array [see Figure 17) and low 40Ar/36Ar ratios in the mantle end-member (Figure 19 and 20) imply a contribution of air-related component.
The production 4He */21Ne* ratio is almost constant independently on natural environments, e.g., U-bearing minerals or ordinary rocks of various composition [Kyser and Rison, 1982; Verkhovsky and Shukolyukov, 1991; Yatsevich and Honda, 1997]. Because the production ratio is known reasonably well, He-Ne isotope relationships are able to quantify the rate of noble gas elemental fractionation and shed light on related processes [Verkhovsky et al., 1983]. 3He/22Neprim versus 4He*/ 21Ne* plot presents He-Ne relationships for Kola UACC samples along with MORB and OIB data. All 3He measured in the samples is considered as the primordial component (subscript prim), and 22Neprim, 4He* and 21Ne* are calculated from equations:
where subscripts m, atm and ini define measured, atmospheric and initial values, respectively. The relevant primordial isotope compositions are considered to be solar [Anders and Grevesse, 1989] and the calculated initial values depend on proportion of mixing of solar and atmospheric species in each individual sample. In contrast to almost constant production 4He*/ 21Ne* ratio, the measured ratios of both radiogenic and primordial species vary within 4 orders of magnitude and correlate. The primordial nuclides, 3He and 22Neprim, were stored in a less degassed mantle reservoir since the earth accretion, 4.5 Ga [O'Nions and Oxburgh, 1983; Kellogg and Wasserburg, 1990; O'Nions and Tolstikhin, 1994, 1996].
The radiogenic nuclides, 4He* and 21Ne*, were manly produced within the upper mantle reservoir having much higher (U+0.24Th)/ 3He ratio than the less degassed mantle [O'Nions and Tolstikhin, 1994]. Estimates of the mean residence time of a highly incompatible elements in the upper mantle results in 1 Ga [Galer and O'Nions, 1985; Kellogg and Wasserburg, 1990; O'Nions and Tolstikhin, 1994, 1996]. These time constraints imply that the direct correlation in Figure 24 should result from fractionation process(es) occurred only slightly before, during or even after the gases were trapped by solids: otherwise the ratios between ancient ( 3He, 22Neprim ) and young ( 4He* and 21Ne* ) species would not have correlated.
The straight line shown in Figure 24 reasonably well fit the correlation; the product of the slope of this line, 3.510-7, and the mean 4He*/21Ne* production ratio, 1.510-7, allows the initial 3He/22Neprim ratio (preceded the fractionation) to be recovered, 3He/22Neprim 5, which is similar to the solar system primordial ratio 3 [Anders and Grevesse, 1989]. The similarity envisages the unfractionated solar-like primordial gases in the less degassed reservoir, in accord with inferences from steady-state models of layered mantle [e.g., O'Nions and Tolstikhin, 1994]. The samples from Kola UACC do not deviate far from the production/primordial ratio (Figure 24) in contrast to those from MORB and OIB. A substantial decrease of He/Ne ratios (relative to the primordial/production values) could results from preferential migration of He isotopes from fluid inclusions in OIB and MORB samples. He shows much higher penetrability through silicates than Ne and Ar [Morozova and Ashkinadze, 1971] and specifically through silicate glasses, major noble gas hosts in ocean ridge and seamount environments. While He is migrating from the inclusions through basalt glass into free fluids, the residual He/Ne ratios are decreasing.
A 50 fold increase of He/Ne ratios requires a special explanation. Such a trend could originate from partial melts degassing owing to the better solubility of He in silicate melts than the solubilities of heavier gases [Jambon et al., 1986; Lux, 1987]. Noble gas partitioning among solid, gas and silicate melt in a magma chamber is described as [Spasennykh and Tolstikhin, 1993]:
where Ln and Lm are final over initial concentration ratios for species m and n, and S, b, R, T, F and K in the power are the solubility, the volume gas/melt ratio, the Boltzman gas constant, the temperature, the volume solid/melt ratio, and the solid/melt partition coefficient, respectively. The maximal fractionation would be expected if b 0, K 0, and for the conventional solubility [cc/(g atm)] S(He) = 0.0006 and S(Ne) = 0.00025 [Jambon et al., 1986, Lux, 1987] the required 50-fold increase of He/Ne ratio could originate if 95% of He had been lost from degassing melts and only 5% retained. At first glance such rate of degassing seems reasonable. However analysis of the data presented in Figure 24 shows that among 250 data-points, 50 samples having highest 3He concentrations, >110-10 cc STP g -1, all deviate to the right-top off the production/primordial ratios. The ratio of this concentration and the helium retention coefficient predicted by fractional degassing ( 0.05) gives the expected 3He concentration in undegassed melts, 210-9 cc STP g-1 or even 210-8 cc STP g -1 for the most 3He-rich MORB glasses. These values are by 10 to 100 times exceed the initial 3He abundance in basaltic melts estimated from steady-state and evolutionary degassing models (5.6 3)10-10 cc STP g-1 [Tolstikhin and Marty, 1998]. Also the present day production of the oceanic crust, 61016 g a-1 [Crisp, 1984], and 3He flux into the oceans, 2.24107 cc STP a-1 [Craig et al., 1975; Farley et al., 1995], give 3He concentration in MORB melts at 3.5610-10 cc STP g-1, similar to the above estimate but at least an order of magnitude less than that required to satisfy the fractional degassing hypothesis. The preferential helium diffusion from vesicles through silicate glasses proposed above produces high 4He/21Ne* ratios in a complementary pore fluid phase. Helium concentrations in pore fluids could by an orders of magnitude higher than those in the fliud-bearing rocks [e.g., Tolstikhin et al., 1996]. If the fluid would have been trapped into "secondary" inclusions of host rocks/minerals, 4He concentrations and 4He/21Ne* ratios could both be high in a qualitative accord with the data. Also a non-equilibrium degassing process, when He migrates into ascending bubbles faster than Ne, could be responsible for the enhanced He/Ne ratios. More work is needed to understand and quantify this alternative mechanism.
Similar isotopic and chemical characteristics of small-volume continental magmas, including alkaline melts, and alkali basalts from small oceanic islands or seamounts calls for related source(s) and processes involved. While both astenospheric [Nelson et al., 1988; Kwon et al., 1989] and lithospheric [McKenzie and O'Nions, 1995] source regions were suggested using isotopic arguments, models of carbon- and alkali-rich melt generation and development generally envisaged a metasomatically enriched lithospheric source [Wyllie et al., 1990]. The following discussion agrees with the model [McKenzie and O'Nions, 1995] which reconciled isotopic, geochemical, geochronological and geophysical data. The model envisages: (i) the subcontinental lithosphere including a MORB-source-like bottom layer and a depleted (relative to the MORB source) layer above as the most plausible environments, and (ii) processing of these layers by addition of 10 to 30 % of a metasomatic melt originated by extraction of ~0.3-0.5 % melt from the MORB-source astenospheric mantle. Because the subcontinetal lithosphere is a long-life conservative reservoir [e.g., Kramers, 1979, 1991; Richardson et al., 1984], time interval between the metasomatic processing and the mobilisation of parent alkaline magmas (intruded into the continental crust) or detachment of the processed domains (entrained into the thermal convection within the astenospheric mantle) could be long and variable which allows enriched (relative to the MORB) isotopic signatures to be generated.
This time interval is crucial to constraint the 3He-bearing source for parent melts of the Kola UAC Complexes. It should be emphasised that both trace-element- [McKenzie and O'Nions, 1995] and major-element-related [Wyllie et al., 1990] models do not envisage a deep-mantle plume-like source for the metasomatic low-partial-melting melts. In the past the upper mantle could also show lower 4He/3He ratios due to e.g. a higher flux of 3He rich material from the lower mantle. To understand whether the ancient upper mantle could be a source of the metasomatic low 4He/3He -melts, the age when this reservoir had have 4He/3He ratio similar to that in parent melts of Kola Devonian UACC should be compared with the age of metasomatism inferring from other isotopic systematics, e.g., Rb-Sr or Sm-Nd.
Figure 25 comprises 4He/3He upper mantle evolutionary trends compiled from several recent degassing models. While segments of the trends related to the early earth are different depending on assumptions involved, all post-3-Ga segments show higher 4He/3He ratios than that obtained above for UACC (Section 4.1.1, see Figure 16). The model-derived upper mantle 4He/3He ratios were similar to those initial for Kola UAC rocks approximately 3 Ga ago.
In contrast to the above quite ancient age, a shorter metasomatism-extraction interval is inferred from Rb-Sr systematics. Sr isotope composition of UAC Complexes is well constraint by Rb-Sr isochrone dating and low Rb/Sr rocks/minerals, such as apatites or carbonatites: initial 87Sr/86Sr varies from within 0.7030 - 0.7040 exceeding the present day average normal MORB ratio 0.7024 (Ito et al., 1987]. The average Rb/Sr of highly differentiated UAC Complexes appears to be less reliable. Two values suggested by Gerasimovsky  for the Lovozero massif, Rb/Sr 0.377 [see also Kramm et al., 1993], and by Kukharenko et al.  for the Khibiny 0.15 are available. These estimates together with the model Sr isotope evolution trend for the upper mantle [Azbel and Tolstikhin, 1988] give the model age of the upper mantle metasomatism within 420 - 700 Ma, substantially less than that predicted by U-He systematics. This difference rules out the upper mantle as a source of He-bearing 700-Ma-old material and suggests the lower mantle (or its stagnant less-degassed domain) as a host reservoir for primordial rare gases in UACC.
Available model estimates of noble gas abundances in the principal terrestrial reservoirs [Tolstikhin and Marty, 1998] allow contribution of the three reservoirs, the lower mantle, the upper mantle, and the crust (represented by groundwater containing presumably atmospheric gases) to be quantified (Table 6). A minor contribution from the less degassed reservoir imply that the plume itself only stimulated metasomatism of the subcontinental lithosphere and the major role in this processes belongs to melts from the upper mantle, in accordance with recent geochemical and petrological models [Wyllie et al., 1990; McKenzie and O'Nions, 1995].