Tuffisitic Kimberlites and Their Emplacement Processes: A Review of Some Current Hypotheses
Kimberlite diatremes are filled by the variety of kimberlite known variously as tuffisitic kimberlite (TK), volcaniclastic kimberlite (VK), subvolcanic magmaclastic kimberlite (SMK), or autolithic tuff breccia (AKB). The distinctive characteristics of these rock type are reviewed and compared with other volcaniclastic diatreme- zone rocks occurring in melilitite vents. It is shown that TKs are petrologically unique in consisting of massive unsorted matrix-to-clast supported rocks containing anhedral olivine macrocrysts and subhedral phenocrysts; a specific assemblage of sub-spherical to elliptical magmaclasts (formerly pelletal lapilli), xenolithic and autolithic clasts, thin cryptocrystalline mantles of microlitic diopside, phlogopite and apatite on the preceding constituents, and an interclast material consisting mainly of mixed layer phyllosilicates (chlorite-smectite) with lesser diopside. Magmaclasts are defined as solidified former melt-bearing clasts formed during, or prior to, emplacement by any process of magma disruption. Magmaclasts in TKs consist of chlorite-pseudomorphed olivine, fresh phlogopite, spinel, perovskite, apatite and chlorite. Petrologically they are similar to the groundmass of hypabyssal kimberlites but lack monticellite and carbonates. Microlitic diopside mantles decorating magmaclasts grade continuously into the interclast matrix and represent a continuum of crystallization. Clastic or carbonate matrices, accretionary and/or deformed welded or vesicular clasts are absent from TKs. Mechanisms proposed for the formation of TKs include fluidization, phreatomagmatism , embryonic pipe modification, and in-vent column collapse; however it very possible that diatremes are unlikely to form by any single process. TK formation has been considered to represent either downward or upwards migration of the foci of volcanic activity. Apart from their terminology, outstanding problems in TK genesis include: the formation of the magmaclasts and their relationships to either sub-aerial pyroclasts and/or root zone holocrystalline kimberlites; the formation of the microlitic diopside fringing mantles; and the origins of the interclast matrix. The latter is variously interpreted to be post-emplacement pore space filling, fine ash replacement by externally-derived fluids or deuteric hydrothermal mineralization. Data are presented for the oxygen isotopic composition of chlorite-smectite in TK and hypabyssal kimberlite which suggests that post- emplacement interactions of TK with meteoric water cannot be extensive. There is no mineralogical evidence for pervasive replacement of all components of TKs of diverse ages and location by material deposited from externally-derived groundwater.
Some Peculiarities of Kimberlite Petrogenesis
Kimberlites are igneous rocks and the attempts to unravel their mechanisms of genesis and evolution have drawn heavily on classical methods of igneous petrology. Here I review some important differences between kimberlites and other mafic or silicic magmas and examine their impact on our understanding of kimberlite. Four factors are considered. At the source they are (1) the hybrid nature of kimberlite and its impact on the concept of a "primary kimberlite magma", and (2) assumptions surrounding source processes, notably melting and metasomatism. Near the surface they are (3) degassing behavior of CO2 and its effects on eruption, and (4) water and the influence of phase stability in deuteric-meteoric-hydrothermal processes on kimberlite composition. The concepts of partial melting and of a primary or primitive melt composition are central in experimental and geochemical research into kimberlite petrogenesis. Petrography presents strong evidence for the hybrid character of kimberlite. It can be argued (1) that there is no such thing a primary kimberlite liquid composition, because solid and liquid components of kimberlites may never have been separated and interacted continuously, or (2) that there is an impractically wide range of primary kimberlite liquid compositions caused by mixing processes between end-members. Geochemical models require source enrichment of kimberlite, but isotopes limit the time available for this event. It is probable that enrichment and "melting" are facets of the same process. Because the timing of kimberlite eruption likely depends as much on lithospheric conditions such as state of stress as it does on mantle melting, it is not clear whether kimberlite is a product of mantle heating or mantle cooling. Instead of enrichment, melting and differentiation from a parent liquid, kimberlite petrogenesis should be considered in terms of chemical and thermal hybridization between end-member components, such as: carbonatite melt of peridotite or eclogite, silicate melt of peridotite or eclogite, and solid mantle. The volatile content of ascending kimberlite magma is poorly known, but is thought to play a role in explosive eruption. Models of explosive eruption based on the degassing of silicic to basaltic magmas may be inappropriate to kimberlite because kimberlites frequently differentiate to carbonatite rather than (or in addition to) exsolving a vapor phase. The catastrophic "degassing" of explosive kimberlite eruptions may be magma decomposition rather than vapor exsolution, having its origins in factors other than pressure change, and calling for a different thermodynamic approach to that employed in the case of silicate melts. The water content of kimberlite is a more tractable problem and can be estimated from basalt trace element systematics and also from magmatic phenocrysts with trace structural OH. The composition of kimberlite is dictated by minerals that remain after deuteric and meteoric-hydrothermal interactions. Volatile components and alkalis (notably Na and K), of which a fraction may be housed in water soluble fractions, are likely to be lost, leading to erroneous inferences about alkali-bearing minerals in the source.
A thermodynamic Model to Describe Temperature Changes During Kimberlite Ascent
We present a thermodynamical model of kimberlite magma ascent from 200 and 400 km depth. We model four different magma compositions to bound the behaviour, namely a basalt, a carbonatite and two intermediate cases that represent kimberlite magma. Kimberlite 1 has properties intermediate between carbonatite and basalt, and kimberlite 2 has properties that are closer to basalt. The results show that adiabatic expansion of the melt phase can be a large cooling effect during kimberlite melt ascent, accounting for 140 °C cooling of kimberlite 1 and ∼90 °C cooling of kimberlite 2. Melts with high volatile contents are more corrosive during magma ascent, enriching the melts in MgO as olivine xenocrysts are assimilated. Kimberlite magma temperatures decrease during ascent up to the onset of rapid pressure- induced olivine crystallization. The models show that little olivine assimilation occurs during kimberlite ascent (<1%), and this implies the magma composition is set at depth and is not acquired via olivine dissolution. Latent heat release counteracts contemporaneous cooling mechanisms such as gas exsolution and lithospheric entrainment, and in this regime the magma temperature increases as the pressure decreases. At shallow levels gas exsolution and expansion become dominant processes and the magma temperature cools during the final stages of ascent. Our models suggest that shallow magma temperatures consistent with estimates from geothermetric studies (1030 - 1170 °C) occur when the volatile content of the ascending kimberlite magma is less than 10 wt% H2O. Models with 5 wt% H2O + 5 wt% CO2 are consistent with observations of approximately 25% phenocrysts and 25% xenocrysts in many kimberlites.
Role of Volatiles in Kimberlite Ascent and Eruption
The unique aspect of kimberlite magmas is their potential for having high dissolved contents of primary
volatiles (e.g., H2O and CO2) coupled to a high ascent rate. The high ascent rates help couple the exsolved
fluid to the magma as it rises to the point of eruption. During ascent the system evolves from a system
featuring 30-40% suspended solids in a silicate melt to a system that is volumetrically dominated by the
exsolved fluids (due to exsolution and expansion). The physical-chemical properties of kimberlite melt govern
the transport and eruption behaviour of kimberlite magmas. For example, exsolution of a CO2-H2O fluid phase
provides a logical and efficient means of reducing magma density and promoting the buoyancy critical for rapid
ascent and eruption. The composition of the exsolved fluid depends on the total dissolved fluid content of the
melt as well as the T-P ascent path (e.g., Holloway & Blank 1994). Under conditions of equilibrium degassing
(e.g., closed system), the original dissolved fluid content limits the range of fluid compositions produced during
ascent. Under perfect fractional degassing (open system), increments of equilibrium fluid are released and
"fractionated". Such situations arise when 2-phase flow (melt and gas) develops and the gas phase decouples
from the host magma. Separated two-phase flow is likely to develop in kimberlite and allows for highly
transient fluid compositions beginning with fluids extremely enriched in CO2, and ending with H2O-dominated
fluid. The physical properties and behaviour of the fluids during ascent are, thus, constantly changing in
response to the evolving fluid composition. Here we use computational models calibrated on experimental
data for multicomponent melts (e.g., MELTS; Ghiorso & Sack 1995) saturated with a CO2-H2O fluid (e.g.,
Papale et al. 2006) to explore the physical-chemical properties of volatile-saturated kimberlite during ascent
and eruption. The exsolved magmatic fluid is modelled as mixtures of CO2 and H2O. No speciation
calculations were attempted. The thermodynamic properties of the fluids were retrieved using program
REFPROP (Lemmon et al. 2007) that employs the GERG-2004 equation of state and mixing models (Kunz et
al. 2007). We then compute how the properties (V, H, S) of the expanding fluid change as a function of ascent
path. As the magma decompresses, the fluid phase increases in mass and volume, and the thermal
consequences of adiabatic expansion begin to dominate. We have explored the isentropic and isenthalpic
adiabatic expansion paths (e.g., Spera 1984; Mastin & Ghiorso 2003) for the ascending magma. The paths
are based on "intrinsic" thermodynamic properties (Dodson, 1971) and do not include energy associated with
motion or position in the gravitational field. References Dodson MH 1971. Nature 234, 212.
Ghiorso MS & Sack RO 1995. Contributions Mineralogy & Petrology 119, 197-212.
Holloway JR & Blank JG 1994. MSA 20, 187-230.
Kunz O et al. 2007. GERG Technical Monograph 15. Fortschritt Berichte VDI, Reihe 6, 557.
Lemmon EW et al. 2007. Fluid Thermodynamic & Transport Properties - REFPROP Version 8.0, NIST, Boulder, CO.
Mastin LG & Ghiorso MS 2001. Contrib Mineral Petrol 141: 307-321.
Papale P, Moretti R, Barbato D (2006) Chemical Geology 229, 78-95.
Sparks RSJ et al. 2006. Journal Volcanology & Geothermal Research 155, 18-48.
Spera, F.J. 1984. Contributions Mineralogy & Petrology, 88: 217-232.
Volatiles in Kimberlite Magmas: Experimental Constraints
Kimberlites are volatile-rich magmas, and primary igneous carbonate and phlogopite indicate of the presence of CO2 and H2O. However, there are few constraints on their volatile content and solubility of CO2 and H2O. Studies of volatile solubility and effects on phase equilibria are limited by the incomplete understanding of the exact compositions of kimberlite melts, which is a consequence of the almost ubiquitous groundmass alteration. It is widely accepted that kimberlite melts must be silica deficient with a high proportion of cations, notably Mg and Ca. We carried out experiments at 100 and 200 MPa pressure with excess CO2, excess H2O, and CO2-H2O mixtures, using some recently proposed generic kimberlite magma compositions as well as a composition with an even lower silica to cation ratio (reconstructed to remove the effects of serpentinization). At 1250°C, the generic kimberlite compositions typically have <30% melt and are dominated by Ca-rich (~2-4% CaO) olivine, with lesser amounts of spinel and monticellite. The experimental melt compositions have SiO2 contents of 20 to 25% and are very Ca-rich (25 to 35%). Although these experimental melt compositions are not exact matches to natural kimberlite melts, they provide an opportunity for investigating CO2 and H2O solubilities in silica-poor Ca-rich melts, which are sufficiently close to the silica to cation ratio characteristics of proposed generic kimberlite compositions to place constraints on volatile behaviour. The CO2 and H2O contents of the experimental glasses were measured by ion probe. The pressure-dependent solubility of H2O as a function of pressure is 20 to 30% less than in basalt melts with a best-fit solubility coefficient of 2.2 x 10-6 Pa-1/2, assuming a parabolic solubility law. These results confirm the broad trend of decreasing H2O solubility as SiO2 decreases, and is likely reflects decreasing melt polymerisation. In contrast these melts can dissolve several percent CO2 at 100 MPa and solubility increases approximately linearly with increasing pressure. This is also consistent with the general trend of increasing CO2 solubility with decreasing SiO2 in natural silicate melts, with a continuum through to carbonatites as the cations become completely over SiO2. Recent work on groundmass mineralogy suggests that kimberlite melts were originally lower in silica and magnesium, and higher in calcium and carbonate prior to the ubiquitous alteration processes that produce serpentinization. Our experiments on such a reconstructed composition show up to 60% melt at 1250°C with a significantly higher CO2 solubility, but water solubility remains low, similar to the other more silica-rich melt compositions. These reconstructed low silica compositions can better explain higher CO2 contents (~10 wt%) observed in emplaced rocks, although even lower silica compositions closer to carbonatites are required if the magmas are to reach the root zone without exsolving their CO2 (as suggested by high level carbonate-rich dykes and sills). If kimberlites have high H2O contents ( > 5 wt%) then the low solubility in the experimental kimberlite melts indicates that gas exsolution can initiate at great depth and H2O may thus be the major propellant in their explosive eruption. A conundrum is that pure carbonate melts have very high H2O solubility. Thus, if kimberlites crystallize to generate carbonatitic residual melts, the solubility of H2O might actually increase with decreasing pressure.
Xenoliths as magmatic 'menthos'©
Recent popular experiments using carbonated liquids and rough-surfaced solids (e.g. 'Menthos' © candies) demonstrate the importance of surfaces to facilitate rapid, heterogeneous nucleation of bubbles from a supersaturated liquid. We examine the likelihood that this phenomenon applies to silicate magmas as well, by examining evidence for the presence of xenoliths in mantle-derived magmas enhancing bubble nucleation. We also explore the possible consequences of such a relationship for xenolith transport and eruption dynamics of kimberlite and other mantle-derived mafic magmas. It is generally accepted that crystals in silicate magmas can serve as nucleation sites for vapor bubbles of H2O/CO2 (e.g., Perfitt and Wilson, 2007). We suggest that the presence of xenoliths in magmas might also greatly enhance nucleation rates of bubbles, generating a positive feedback that facilitates xenolith transport in magmas by: 1) lowering the density of the xenoliths if bubbles coat xenolith surfaces for extended periods of time, or 2) increasing the viscosity of the melt via volatile loss and the magma via the presence of bubbles (at low to moderate strain rates). Field evidence from many kimberlite localities and tephra cones is consistent with the hypothesis that for many mantle-derived magmas xenoliths as transported near the top of the magma column. For example, Prindle volcano, a basanitic cone in eastern Alaska comprises a spatter cones comprising dominantly lava-coated, xenolith-cored bombs of peridotite, pyroxenite and granulite facies xenoliths. Xenoliths also occur in the cone-breaching lava flow, but they are in general smaller in size and less abundant. The crater-facies of kimberlites may likewise be enriched in xenoliths and xenocrysts in comparison to the dike facies. Recently Lensky et al (2006) suggested that deep vesiculation events could lead to microfracturing and incorporation of xenoliths into mantle-derived melts. We suggest that, although this certainly is feasible, it may more likely be true that incorporation of the xenoliths would further enhance vesiculation and create a positive feedback that would also enhance xenolith transport to the surface.
Kimberlite Ascent: Insights from Olivine
Olivine is ubiquitous in both extrusive and intrusive kimberlite deposits worldwide. Within kimberlite, it is mainly present as xenocrysts derived from the disaggregation of mantle-derived peridotitic xenoliths. Many textural and chemical features within the mantle-derived olivine xenocrysts result from post entrainment processes. On that basis, these features record physical and chemical changes attending kimberlite ascent and can be used to elucidate the transport and eruption of kimberlite magma. Our textural study of kimberlitic olivine is based on intrusive and pyroclastic kimberlite from the Diavik kimberlite cluster and from the Igwisi Hills kimberlitic lava flows. Based on these observations and the physical and chemical properties of olivine we derive a relative sequence of textural events. Textural features include: sealed cracks, healed cracks, phases trapping in cracks, rounded grains, overgrowths and phase trapping in overgrowths. These features record processes that operate in kimberlite during ascent, and from these features we create a summary model for kimberlite ascent: -- Olivine is incorporated into kimberlitic melts in peridotitic mantle xenoliths continuously during ascent. Xenolith incorporation is focused at the crack tip where the stress regime is highest. -- Shortly after the incorporation of these xenocrysts the tensile strength of the xenoliths is reached at a maximum of 2 km from its source. Disaggregation of mantle xenoliths (producing xenocrysts) is facilitated by expansion of the minerals within the xenoliths causing intra-crystal slip (i.e. along grain boundaries). -- Continued decompression causes olivine to also break in tension approximately 20 km from source. The void space produced by the failure of the crystals (inter-crystal cracks) is filled with melt and crystals consisting of primary carbonate (high-Sr), chromite and spinel crystals. The carbonate later crystallizes to produce sealed fractures. -- Mechanical rounding of the xenocrysts occurs during pressure release (ascent) events characterized by phase separated (fluid at top: liquid at bottom) flow. Turbulent and fluidized flow heads the propagation whereas kimberlite melt state follows. -- Settling of olivine crystals into the melt state occurs when crack propagation halts. At these points, cracks that are not totally sealed begin to heal, and are ultimately present as healed fractures in olivine in the end deposit. -- Saturation of olivine produces rounded overgrowths on large xenocrysts, euhedral overgrowths on smaller xenocrysts, and a volumetrically minor population of olivine phenocrysts. Olivine growth traps fluid, solid and melt inclusions. Calculations based on these relationships suggest that the melt saturates with olivine at a maximum depth of 20 km and a minimum depth of 7 km.