On the Welding of Pyroclasts from Very Low-Viscosity Magmas: Examples from Kimberlite Volcanoes R. J. Brown,1 B. Buse, R. S. J. Sparks, and M. Field Department of Earth Sciences, Wills Memorial Building, University of Bristol, Queen’s Road, Bristol BS8 1RJ, United Kingdom (e-mail: [email protected])

ABSTRACT Lithofacies in kimberlite pipes in southern Africa exhibit features consistent with welded pyroclastic rocks. These include flared conduit-filling geometries, abundant lithic clasts, lithic clast layers, subhorizontal clast fabrics, gradational contacts with volcaniclastic rocks and sintered and coalesced globular ash, and lapilli and melt-coated particles. The welding dynamics of kimberlite pyroclasts differ from those of glassy, vesiculated pyroclasts in silicic volcanic systems. Low melt viscosity (∼0:1 Pa s) results in the efficient separation of volatiles and melt and the breakup of magma into nonvesicular spherical droplets. The glassy state is difficult to form in silica-deficient magmas because the crystallization kinetics are fast in low-viscosity melts. Three pyroclastic lithofacies are recognized that primarily relate to the state of the initial melt phase in the pyroclasts on deposition: (1) densely welded kimberlite forms where the initial melt phase in pyroclasts remains as pure melt, coalescing almost immediately to form a degassed homogeneous melt that subsequently crystallizes and may be texturally indistinguishable from igneous kimberlite; (2) incipiently welded kimberlite forms where the melt phase in pyroclasts is mostly crystalline with some residual melt (pyroclasts are resistant to deformation but are sticky and can sinter together); and (3) nonwelded kimberlite forms where the melt phase in pyroclasts is fully crystalline on deposition. Transitions between these pyroclast conditions may be abrupt. Gradations to nonwelded deposits and overlaps in their textural features in several kimberlite pipes suggest that welded rocks may be deposited from large-scale fluidized systems. The presence of crystals and lithic clasts may inhibit compaction deformation-welding textures by the formation of hard-particle percolation networks. Similar rocks in other kimberlite pipes may turn out to be welded rocks. The processes outlined here may be generally applicable to other low-viscosity magmas.

Introduction Welding of pyroclastic material comprises two different and independent processes, namely, sintering together of hot clasts at point contacts (Guest and Rogers 1967) and deformation of hot clasts accompanied by densification involving reduction of pore space (Smith 1960). It commonly results in increases in rock density and the progressive development of rock fabrics (Smith 1960; Peterson 1979; Quane and Russell 2005). Welded rocks are common in proximal volcanic environments because of the small heat loss experienced by the pyroclasts before deposition. Examples of welded pyroclastic rocks include ignimbrites (Smith 1960; Ross and

Smith 1961; Milner et al. 1992; Bachmann et al. 2000; Sumner and Branney 2002), welded fallout deposits (Sparks and Wright 1979; Soriano et al. 2002), and tuffs and lapilli tuffs infilling dykes and conduits (Almond 1971; Ekren and Byers 1976; Wolff 1986; Reedman et al. 1987; Sparks 1988; Kano et al. 1997; Soriano et al. 2006). Welded rocks can be difficult to interpret, and in some cases, it can be hard to discriminate between high-grade welded ignimbrites and coherent effusion-fed lavas (e.g., Branney et al. 1992; Henry and Wolff 1992; Le Pennec and Fernandez 1992; Green and Fitz 1993; Sumner 1998; Yasui and Koyaguchi 2004) and between welded fall deposits and welded ignimbrites (see differing interpretations of the Green Tuff, Pantelleria, in Borsi et al. 1963; Schmincke 1974; Wright 1980; Branney et al. 2004). Within a volcanic conduit, it may be difficult to distinguish between a welded tuff and an igneous intrusion (Sparks 1988;

Manuscript received September 6, 2007; accepted March 7, 2008. 1 Present address: Volcano Dynamics Group, Department of Earth and Environmental Sciences, Open University, Walton Hall, Milton Keynes MK7 6AA, United Kingdom.

[The Journal of Geology, 2008, volume 116, p. 354–374] © 2008 by The University of Chicago. All rights reserved. 0022-1376/0000/11604-0003$15.00 DOI: 10.1086/588832

354

Journal of Geology

WELDING OF PYROCLASTS

Sparks et al. 1999). There may also be a continuum between densely welded pyroclastic rocks and clastogenic fountain-fed lavas in terms of both processes and geological features. The distinction is important because the eruption dynamics that create welded rocks can be significantly different from those that produce lavas and intrusive rocks without an intervening pyroclastic stage. The processes may be much more hazardous: lavalike ignimbrites are deposited by widespread, fastmoving, very high-temperature pyroclastic density currents. Welded rocks are common products of explosive volcanism across a wide range of magma compositions, particularly low-viscosity peralkaline magmas (e.g., Schmincke 1974; Mahood 1984; Sumner 1998), but they have not been widely documented in the conduit-filling pyroclastic products of kimberlite magmas. Sparks et al. (2006) suggested that kimberlite rock types previously interpreted as hypabyssal intrusions in some kimberlite pipes might be welded pyroclastic rocks. These rock types, hereafter referred to as coherent kimberlite (CK; following Cas et al., forthcoming), can form large-volume bodies within kimberlite pipes (Nowicki et al. 2004; Field et al., forthcoming). Descriptions are given by Mitchell (1975, 1986), Clement (1982), Clement and Skinner (1985), Clement and Reid (1989), Field and Scott Smith (1999), Hetman et al. (2004), Skinner and Marsh (2004), and Downes et al. (2007). These rocks typically comprise fresh and pseudomorphed olivine macrocrysts and phenocrysts in a crystalline groundmass composed of primary minerals. The groundmass can contain irregular submillimetric domains of typically serpentine and/or calcite (“segregations”). Lithic inclusions can reach in excess of 20 vol%, many of which are highly altered and may be completely replaced mainly by serpentine and calcite. Bodies of CK can have horizontal, inclined, or vertical gradational contacts with clastic kimberlite rocks (e.g., Venetia K1 kimberlite pipe, South Africa [Kurszlaukis and Barnett 2003]; Koffiefontein, South Africa [Naidoo et al. 2004]; Leslie and Pigeon pipes, Canada [Nowicki et al. 2004]; and Victor kimberlite [Webb et al. 2004]). CK is commonly interpreted as an intrusive rock type and as being characteristic of, but not confined to, the “root zones” of kimberlite pipes (Hawthorne 1975; Mitchell 1986; Field and Scott Smith 1999). In this work, we detail compositional, structural, and petrographic evidence from two kimberlite fields in South Africa and Botswana that indicates that some CK rocks are welded deposits, as first raised by Sparks et al. (2006). We highlight key

355

features that are consistent with known processes that occur during the welding of hot pyroclasts, and we draw attention to problems associated with interpreting these rocks as hypabyssal intrusions. The rocks in this study exhibit a range of textures that we interpret as reflecting varying degrees of welding intensity. We explore possible mechanisms of welding of kimberlite pyroclasts and discuss how the welding processes and, importantly, the resulting rock textures differ from those seen in welded rocks of silicic volcanic systems. While unambiguous intrusive kimberlite bodies are present in many kimberlite pipes and the crust (dykes and sills), we conclude that many occurrences of CK found within kimberlite pipes may be better interpreted as welded pyroclastic rocks. We adopt a modified version of the welding terminology summarized by Sumner et al. (2005). We define compactional welding as postdepositional sticking together at point contacts (sintering) and compaction of particles under the influence of a deposit’s in situ cooling rate and load pressure. We define agglutination as the sintering of particles on deposition, which may or may not be accompanied by deformation of the particles (Guest and Rogers 1967). Coalescence is used to describe a process whereby very low-viscosity pyroclasts rapidly form a homogeneous liquid on deposition and in which original particle outlines are lost.

Geological Observations Venetia K2 Kimberlite Pipe, South Africa. K2 is a steep-sided volcanic pipe, ∼250 × 300 m in diameter, with a known vertical extent of >900 m. It is the second-largest pipe in the 512-Ma Venetia kimberlite cluster, South Africa (fig. 1A; Phillips et al. 1999; Kurszlaukis and Barnett 2003). It is crudely pear shaped in plan view at the present surface and becomes elongated N-S at depth. Thin kimberlite dykes occur discontinuously around the margins (fig. 1B). K2 has a complex internal geology comprising layered and nested breccia units of varying grain size and composition and is divided into two lithofacies associations, K2 east and K2 west (fig. 1B; Brown et al., forthcoming). K2 east consists entirely of massive volcaniclastic kimberlite (MVK), a volcaniclastic rock that typically comprises ∼8– 20 vol% angular lithic lapilli supported in a matrix of poorly sorted crystal-lithic lapilli tuff. K2 west comprises a sequence of layered breccias with variable proportions of accidental lithic clasts. These breccias form inward-dipping lenses, layers, and pods defined by changes in lithic clast abundance

356

R. J. BROWN ET AL.

Figure 1. Map and cross section of the K2 kimberlite pipe, Venetia Mine, South Africa. A, Location map of the Venetia and B/K9 kimberlite pipes. B, Simplified geological map of K2 showing the main geological units in the pipe. C, Schematic cross section as inferred from drill core. The upper half of the pipe comprises a series of gently dipping layered volcaniclastic breccias (VKBr) and country rock breccias (Br). The lower half comprises a complex layered sequence of VKBr, Br, and coherent kimberlite (CK). Drill core data indicate the continuation of the layered volcaniclastic sequence below 350 m below surface level, but data are insufficient to delineate geological units. Field of view for boreholes is ±25 m north and south of cross section.

and grain size (fig. 1C). Alteration is pervasive in rocks in K2, and all olivine crystals have been replaced by serpentine, talc, and carbonate. Layered with the breccias in deeper parts of K2 west (>350 m below surface; fig. 1C) are numerous horizons of competent CK (fig. 2A). These horizons can reach many tens of meters thick. CK is characterized by 2–20 vol% of accidental lithic lapilli and blocks, although it can contain up to 45 vol% lithic clasts. Lithic clast abundances commonly vary and can define sharp to diffuse layering (fig. 3). The exact geometry of units (whether sheetlike or podlike) cannot be fully constrained from the drill core, but comparisons with similar sequences in

outcrop at higher levels of the mine suggest that it comprises discontinuous sheets and lenses meters to tens of meters in lateral and vertical dimensions (fig. 1C). Lithic clasts are commonly completely replaced by serpentine, calcite, and phlogopite and may have highly irregular alteration halos (fig. 2B). Average maximum lithic clast size in most CK units is 20–30 cm (fig. 4). Typical lithic clast abundances and grain sizes are much lower than those of the adjacent units of volcaniclastic kimberlite breccia (see fig. 4). Most CK units have an inequigranular texture composed of 10–30 vol% pseudomorphed anhedral olivine mantle xenocrysts (fig. 2C) and euhedral and

Journal of Geology

WELDING OF PYROCLASTS

357

Figure 2. Coherent kimberlite (CK) in drill core in K2 west (500 m below surface; fig. 1). A, Characteristic dark gray homogeneous CK. Note uniform distribution of lithic clasts (light gray). B, Highly altered lithic clast that has been almost entirely replaced by calcite and serpentine. C, Polished slab showing abundant coarse olivine macrocrysts with a preferred orientation (30°–45° from horizontal).

subhedral pseudomorphed olivine phenocrysts in a variably altered crystalline groundmass composed of olivine, phlogopite, calcite, and spinel microcrysts (<100 μm) in a clay-serpentine base (fig. 5). Olivine crystals have aspect ratios of up to 1∶5 (fig. 2C). A preferred alignment (30°–45° to horizontal) of elongate olivine crystals is present in some units (figs. 2C, 5). Olivine crystal abundance varies on a centimeter to meter scale, and in several units, olivine crystals form diffuse clusters up to several centimeters in diameter (fig. 5). Alteration is variably strong, and most olivine crystals have been partly or wholly replaced by serpentine, calcite, or both.

Three main types of groundmass texture are apparent in thin section. The first is a patchy, globular texture consisting of framework-supported rounded and globular masses of groundmass (<0:5 mm) and subspherical groundmass-coated grains (olivine crystals or lithic clasts) in interstitial serpentine and calcite (fig. 6A). The interstices form irregular, pinched domains. Sparry calcite crystals line the exteriors of some groundmass blebs. The second type exhibits a partly holocrystalline inequigranular texture composed of olivine macrocrysts and phenocrysts (10%–20%) in a crystalline groundmass of primary kimberlite minerals. Locally present within the groundmass are fine-grained (∼50–

358

R. J. BROWN ET AL.

100 μm) irregular, pinched calcite and serpentine domains, similar to the interstices in the first textural type (fig. 6B–6D) that partially surround small blebs of groundmass or groundmass-coated grains (fig. 6C). Some blebs of groundmass appear to have been deformed between crystals and pelletal grains (fig. 6C, 6D). The third textural type is essentially holocrystalline and uniformly textured and lacks the serpentine-calcite domains (fig. 6E). Groundmass spinel microcrysts are ubiquitous and can be poikilitically enclosed by phlogopite. Weak textural heterogeneity is defined by small zones (<500 μm) that are devoid of olivine phenocrysts and have elevated abundances of spinel microphenocrysts. B/K9 Pipes, Damtshaa Mine, Botswana. The B/K9 kimberlite body of the Orapa kimberlite cluster, Botswana, comprises three coalesced pipes (North, Central, and South pipes) that diverge at a depth of 250 m below the surface (fig. 7). The kimberlites intruded through Archaean granite gneisses into Karoo Supergroup strata (Field et al. 1997). At present, the mine pit has been excavated to 50 m, and Stormberg Formation Karoo basalts form the adjacent country rock. The Central Pipe crosscuts both the South and North pipes. The kimberlite rocks in the Central Pipe are overlain by a basinfilling deposit of mass movement–derived basalt breccia (fig. 7). The B/K9 pipes contain large pipe-shaped bodies of CK macroscopically similar to those seen in K2. The CK is a competent black rock with 2–7 vol% lithic lapilli (fig. 8) and 20–25 vol% olivine macrocrysts and phenocrysts in a partly crystalline groundmass. The polymict lithic population comprises fresh and oxidized basalt, highly altered basalt, and fresh and oxidized peridotite. The composition of the lithic population varies across the pipe from basalt dominated (e.g., localities D1, D4, and D5; fig. 10) to peridotite dominated (e.g., localities D2 and D3; fig. 10). The ratio of fresh basalt clasts to altered basalt clasts also varies considerably within CK units (fig. 10). Lithic-enriched and lithic-poor lenses, layers, and pods occur on a decimeter to meter scale, as do variations in lithic clast

Figure 3. Graphic log of coherent kimberlite (CK) units in drill core in K2. CK occurs layered with volcaniclastic breccias and exhibits variations in the abundances of lithic clasts. Contacts between different lithofacies can be sharp or gradational. Depth plotted in meters below present surface. Borehole dipped 60° along an azimuth of 150°.

Journal of Geology

WELDING OF PYROCLASTS

359

Figure 4. Maximum lithic clast diameter plotted against lithic abundance for the major lithofacies in K2 from drill core data. The lithofacies (VKBr, CK, and MVK) are petrographically distinct. CK has a low maximum lithic clast size (<400 mm) and typically <20 vol% lithic clasts (left of dashed line). Kimberlite dykes typically contain <1 vol% lithic clasts and cannot be plotted on this graph. See figure 1 for explanation of abbreviations.

size (fig. 8B). Lithic clasts are typically <10 cm in diameter, although large, outsized megablocks and boulders of basalt are present near the pipe margins (fig. 9). Layers of clast-supported basalt breccia occur conformably interbedded with CK (fig. 9). These can reach 10 m thick and pinch out toward the center of the pipe. Platy lithic clasts (aspect ratios >3∶1) in CK exhibit strong subhorizontal fabrics along faces that parallel the circumference of the pit (fig. 11), while moderate conjugate dips (35°W and 55°E) are present along exposures radial to the pipe. Radial exposures (locality D6; figs. 10, 11) commonly show a gradual transition from shallow inward dips to horizontal alignments toward the center (fig. 11). In thin section, CK consists of serpentinized anhedral olivine macrocrysts (20%–25%) and serpentinized subhedral to euhedral olivine phenocrysts (<0:75 mm) in a patchy groundmass composed of globular oxide-rich domains and oxide-free serpentine-calcite domains (fig. 12A), similar to that shown in figure 6A for CK in K2. The oxiderich groundmass domains form an interconnected network in which the oxide-free domains form the interstices. Globular oxide-rich groundmass domains are 20–500 μm in diameter and are composed mainly of microcrysts of perovskite and spinel

(fig. 12B). The margins of these domains are crenulate and irregular. Oxide-rich groundmass commonly forms thin (tens of microns) rims around olivine phenocrysts (fig. 12C, 12D). Oxide-free domains (5–10 vol%) are highly irregular in shape, with pinched terminations between globular masses of oxide-rich groundmass. They comprise serpentine with marginal isopachous apatite or calcite rimmed by secondary serpentine. Some serpentine domains display zoning that is similar to that seen in serpentinized olivine phenocrysts and macrocrysts of lizardite-saponite rims with lizardite cores. Interpretation Macroscopic Features. The CK documented in K2 and B/K9 exhibits macroscopic features that together are consistent with a pyroclastic origin. These include (1) polymict crustal lithic clasts intimately mixed with juvenile kimberlite material; (2) variations in the abundance, composition, and grain size of lithic clasts, including the occurrence of internal lithic layers and fabrics within CK; (3) occurrence within layered sequences of volcaniclastic rocks and breccias; and (4) gradational contacts with adjacent volcaniclastic rocks. CK

360

R. J. BROWN ET AL.

Figure 5. Centimeter-scale variations in olivine macrocryst abundance and grain-size distribution in coherent kimberlite in drill core (K2). Contacts between different zones are gradational. Note the macrocryst-free zone in the upper third.

shares these characteristics with other types of pyroclastic rocks in kimberlite pipes. MVK (see Sparks et al. 2006) commonly exhibits cryptic lithic clast layering and subtle variations in grain size, fabric, and lithic clast abundances (see Gernon et al., forthcoming b); a paucity of large lithic clasts (>10 cm), in contrast to adjacent breccias (fig. 4); and layering and clast fabrics that dip into the

interior of the pipe (fig. 9). In MVK, these features have been attributed to a pyroclastic mixture by gas fluidization (Walters et al. 2006; Gernon et al., forthcoming a). Gradational contacts between CK and MVK have been observed in other pipes (e.g., Clement and Skinner 1985; Clement and Reid 1989; Hetman et al. 2004; Skinner and Marsh 2004). Previous workers have suggested that the lithic clasts in CK-type rocks may have been incorporated during the percolation of low-viscosity magma into pore space in breccias (“kimberlite breccias” in Dawson and Hawthorne 1973; Clement and Skinner 1985). However, lithic clast abundances are generally too low for CK to have formed in this manner, and the grain size distribution of the lithic clasts should be similar to that of the putative host breccia (i.e., a mass movement–derived or explosion-derived breccia). This is not the case, and lithic clasts in CK typically have small maximum clast sizes (<20 cm), unlike adjacent breccias in the pipe (fig. 4). Marginal quench textures, which might be expected when hot kimberlite magma percolates through a cold lithic breccia, have not been observed. CK units within layered sequences in either pipe do not exhibit sharp, intrusive contacts typical of dykes, and they appear to have been emplaced in sequence with the volcaniclastic rocks and breccias. We propose that the most efficient way of mixing lithic clasts with juvenile material is within an explosive pyroclastic jet. Lithic clasts can become entrained in the jet by a number of mechanisms, including erosion from the pipe walls during eruption, spalling of the conduit walls, entrainment of wall rock by explosive magma-water interaction, and mass wasting (e.g., Macedonio et al. 1994; Valentine and Groves 1996). Brown et al. (forthcoming) interpreted the layered sequence of pyroclastic rocks and breccias in the K2 pipe as the deposits of pyroclastic jets and of mass-flow events (collapse of the pipe walls to form breccias). Variations in the abundance of lithic clasts within CK in K2 are similar to those seen in kimberlite breccias (Brown et al., forthcoming). Many bodies of CK have been interpreted as hypabyssal intrusions (e.g., Hawthorne 1975; Mitchell 1986; Field and Scott Smith 1999; Downes et al. 2007). However, the occurrence of abundant lithic clasts and of the layering of lithic clasts within CK is problematic for an intrusive origin. It is dynamically difficult to mix lithic clasts into laminar intrusions, and they are typically rare in igneous bodies. Where they are present, they tend to be concentrated at the margins (e.g., Holness and Humphreys 2003; Brown et al. 2007). An intrusive emplacement mechanism for CK seems incompat-

Journal of Geology

WELDING OF PYROCLASTS

361

Figure 6. Photomicrographs of thin sections of coherent kimberlite (CK) in K2. A, CK comprising rounded ash-sized juvenile clasts of kimberlite groundmass and groundmass-coated olivine grains and lithic grains with interstices of serpentine and calcite. Individual juvenile clasts and coated clasts appear to be sintered at contacts. This rock is interpreted as an incipiently welded pyroclastic rock. B, Generally uniformly textured kimberlite with localized irregularly shaped calcite domains. This rock is interpreted as a densely welded rock. In C and D, calcite domains form the interstices between small plastically deformed juvenile ash grains (dashed line) and groundmass-coated pseudomorphs of olivine phenocrysts. E, Uniformly textured CK comprising pseudomorphs of olivine macrocrysts and phenocrysts in a finely crystalline groundmass of olivine, spinel, phlogopite, serpentine, and calcite. This is interpreted as a densely welded rock in which the original pyroclast outlines have been lost due to coalescence. O ¼ olivine macrocrysts and phenocrysts, s þ c ¼ serpentine and calcite, p ¼ phlogopite, c ¼ calcite domains.

ible with the absence of sharp, intrusive chilled contacts (e.g., Field and Scott Smith 1999; Hetman et al. 2004; Skinner and Marsh 2004). Most undis-

puted kimberlite intrusions in pipes are thin (centimeters to meters thick), sharp-sided dykes and sills (Dawson and Hawthorne 1973; Andrews and

362

R. J. BROWN ET AL.

Figure 7. B/K9 kimberlite body in Orapa, Botswana. A, Simplified geological map of the B/K9 body. B, Cross section of the B/K9 body constructed from drill core data showing major units of coherent kimberlite.

Emeleus 1975; Clement 1982; Mitchell 1986). However, CK in B/K9 locally fills the pipe from wall to wall (fig. 7; see examples given in Nowicki et al. 2004), and it seems uncontroversial to interpret it as a pipe-filling pyroclastic deposit rather than a magmatic intrusion. The upward-flaring geometry of CK bodies in the B/K9 North Pipe (fig. 7) is consistent with the geometry of a filled conduit. Similar flared bodies of CK-like rocks are present in some Russian kimberlite pipes (e.g., Yubileinaya kimberlite, Yakutia; Mitchell 1995 and references therein). The geometries are similar to those created during analog sandbox experiments into fluidization in conduits (Walters et al. 2006). An additional concern is that the inferred low viscosities for kimberlite magma (0:1–1 Pa s; Sparks et al. 2006) would inhibit its emplacement as thick, pipe-filling intrusions (e.g., Bruce and Huppert 1989; Wada 1994). Microscopic Features. We interpret the globular and rounded masses of groundmass and groundmass-coated particles in some CK in K2 and B/K9 as juvenile ash grains (figs. 6A, 12), similar to those seen in volcaniclastic kimberlite rocks (“pelletal lapilli” in Mitchell 1986). The textural relationship between these juvenile grains and the serpentinecalcite domains, which terminate in pinched asymptotic spaces between groundmass regions, is consistent with the latter, representing filled porosity (e.g., fig. 6C, 6D). Similar textures occur in other kimberlite pipes and have been termed “segregationary” and “globular segregationary” textures (Dawson and Hawthorne 1973; Clement 1975, 1982; Mitchell

1975, 1984; Donaldson and Reid 1982; Armstrong et al. 2004; Hetman et al. 2004; Skinner and Marsh 2004). The calcite-serpentine domains (“segregations”) have been interpreted as (1) liquid immiscibility features of an early-crystallizing silicateoxide melt and a late-stage calcite phase (Mitchell 1975); (2) magmatic precipitates following the vesiculation of magma in closed systems at shallow levels (Clement 1975, 1982; Mitchell 1986; Field and Scott Smith 1999; Hetman et al. 2004; Skinner and Marsh 2004; Skinner, forthcoming); (3) amygdales (Dawson and Hawthorne 1970); and (4) pore space in welded rocks (Sparks et al. 2006). The interpretation of the calcite-serpentine domains in CK in K2 and B/K9 as liquid immiscibility features, magmatic precipitates, or filled vesicles is problematic, given the macroscopic evidence for a pyroclastic origin. The expected gas-melt textural relationship for vesicles within a melt is the opposite of that seen in the CK rocks (fig. 13). Similar particle-void textural relationships develop during the sintering of metal-ceramic powders (e.g., Liu et al. 1999; Lame et al. 2003; Zuo et al. 2003) and ceramic beads (Quane and Russell 2006). The rock textures outlined for CK in K2 and B/K9 form a continuum from a clastic framework of rounded and globular groundmass grains and groundmasscoated crystals with interstitial serpentine and calcite (fig. 6A) to uniform holocrystalline textures (fig. 6E). These textures are interpreted as the result of varying degrees of welding (e.g., Walker 1983; Quane and Russell 2005). CK units with abundant serpentine-calcite interstices represent incipiently welded rocks (fig. 6A) in which the pyroclasts have

Journal of Geology

WELDING OF PYROCLASTS

Figure 8. Coherent kimberlite (CK) in the North Pipe, B/ K9. A, CK showing near-homogeneous distribution of lithic clasts (light gray). B, Heterogeneous distribution of lithic clasts in CK with preferred alignment of platy clasts (l). Scale in centimeters; 25 cm on rule.

sintered at point contacts (e.g., Smith 1960), while uniformly textured CK units lacking calciteserpentine interstices are densely welded rocks (fig. 6E) in which fluidal pyroclasts have coalesced and lost their outlines. This is most clearly illustrated in figure 6B, where the original pyroclast outlines remain visible only in small zones (fig. 6C, 6D) within an otherwise uniformly textured crystalline rock. Discussion Welding Processes. Welding of pyroclastic rocks is complex and involves a number of different interacting, overlapping, and competing processes (Smith 1960; Guest and Rogers 1967; Riehle et al. 1995; Sparks et al. 1999). Sintering occurs by diffu-

363

sion and viscoplastic flow at point contacts, while densification involves collapse of pores (both vesicles and interclast spaces), influence of load pressure, capillary-driven flow due to surface tension effects that strive to minimize surface energy, volatile escape by permeable flow, and possible dissolution of volatiles. Welding processes are counteracted by cooling that increases glass viscosity and decreases diffusion coefficients. They can also be stopped by crystallization of the glass (devitrification) and precipitation of vapor phase crystals in pore spaces. The presence of crystals (phenocrysts and xenocrysts) and lithic clasts in hot pyroclastic deposits also influences welding processes. Detailed studies of welded rocks have focused largely on glassy and vesicular silicic pyroclastic rocks (e.g., Smith 1960; Guest and Rogers 1967; Peterson 1979; Branney and Kokelaar 1992; Henry and Wolff 1992; Soriano et al. 2002; Grunder et al. 2005; Quane and Russell 2005; Russell and Quane 2005). Less attention has been given to more fluid magma types, such as basalt, although there are some valuable studies (e.g., Sumner et al. 2005). The problem here is that some features widely held to be diagnostic for welded pyroclastic rocks are the consequence of the very high viscosity of silicic glasses, the extreme sluggishness of the crystallization kinetics, and the vesicular character of the pyroclasts. It cannot be assumed that the same dominant characteristics of welded silicic pyroclasts will also be manifest in putative welded kimberlite pyroclasts. Many of the same generic processes might be expected to be very different because kimberlite magmas are likely to have differing physical properties and behaviors. In particular, it may be extremely difficult to form glasses from very silicapoor and probably carbonate-rich melts because the melt viscosities are very low. The crystallization kinetics are likely to be very fast, a corollary of the difficulty of forming glass in such melt compositions, even in laboratory experiments with quenching rates faster than typical in nature. A further difficulty is that the exact compositions, emplacement temperatures, and volatile contents are not well established in kimberlites. Another contrast with silicic pyroclasts is that unequivocal juvenile pyroclasts in kimberlites are typically nonvesicular to poorly vesicular (Mitchell 1986). Our discussion therefore will contrast the features of the proposed welded kimberlites with much better-known silicic welded rocks and will interpret the differences in terms of contrasting magma properties and consequent eruptive conditions of eruption and welding dynamics.

Figure 9. Layered coherent kimberlite (CK) and collapse basalt breccias at the margins of the South Pipe, B/K9. Layering within the upper unit of CK is parallel to the contact with the basalt breccia. Note the large, outsized boulders and megablocks of basalt in CK. This layered sequence is truncated by basalt breccias of the Center Pipe.

Journal of Geology

WELDING OF PYROCLASTS

365

Figure 10. Compositional variations in the lithic clast population within coherent kimberlite (CK) in the North Pipe, B/K9. The clast population is dominated by variously altered Karoo basalt and peridotite.

Here we discuss some of the properties of kimberlite magmas during eruptions that are pertinent to the occurrence of welded pyroclastic rocks. Establishing typical compositions and physical properties of kimberlite magmas is difficult because these rocks are almost always pervasively altered and because they commonly contain abundant xenocrysts. There is general agreement that they are Properties of Erupting Kimberlite Magmas.

silica-poor magnesian magmas with highly volatile contents (Mitchell 1986). Attempts to reconstruct their compositions have focused on aphanitic sills and dykes (e.g., Price et al. 2000; Le Roex et al. 2003), the freshest of which commonly contain significant amounts of igneous carbonate. The interest here is the compositions of the melt represented by the groundmass in CK rocks and as rims in groundmass-coated clasts. This remnant

366

R. J. BROWN ET AL.

Figure 11. Clast fabrics in coherent kimberlite (CK) at the margins of the North Pipe, B/K9. A, Pronounced subhorizontal fabrics occur in circumferential sections of the pipe. B, Slightly steeper fabrics are present in radial sections. The fabrics are consistent with an inward-dipping layering similar to that seen in the South Pipe (fig. 9).

igneous mineralogy includes various combinations of spinel, perovskite, phologopite, apatite, monticellite, and sometimes igneous calcite. The pervasive serpentine in the groundmass of dykes, sills, and CK rocks limits the reconstruction of melt compositions because assumptions have to be made about whether the serpentine was formed by openor closed-system processes and about whether the serpentinization alteration involves isovolumetric or isochemical replacement of olivine, which is the main source of groundmass serpentine (Stripp et al. 2006). Kimberlite melts in eruption are assessed to have low SiO2 (estimated at 15%–25%), quite high CaO, MgO, and FeO and carbonate components, and some water to stabilize phlogopite. Emplacement temperatures are also poorly constrained,

except for those reported by Fedortchouk and Canil (2004), who estimated temperatures of 1070°–1130° C from olivine-ilmenite pairs. While alteration prevents exact calculation of melt compositions, the likely properties of erupting kimberlite melts can be evaluated. Their silicadeficient character allows high CO2 solubility, predominantly as CO2
3 (Brooker et al. 2001). The melt viscosity will be very low and has been estimated as ≪1 Pa s by Sparks et al. (2006). Very low melt viscosity is also indicated by the very thin rims (<100 μm) around pelletal clasts. Furthermore, the rounded, droplike character to these rims indicates that they were formed by surface tension effects in pure melt (Clement 1973; Dawson 1980; Sparks et al. 2006; Wilson and Head 2007). Similar inter-

Journal of Geology

WELDING OF PYROCLASTS

367

Figure 12. Petrography of coherent kimberlite (CK) in the North Pipe, B/K9. These rocks are interpreted as welded pyroclastic rocks. A, Globular-textured CK comprising olivine macrocrysts and phenocrysts in a patchy groundmass of oxide-rich and oxide-free zones. B, Close-up of groundmass. Irregularly shaped oxide-free calcite domains form the interstices between oxide-rich domains and groundmass-coated grains, both of which are interpreted as originally being pyroclasts. C, SEM backscatter image of CK showing the globular oxide-rich domains (light, speckled) and the irregularly shaped oxide-poor serpentine domains in the interstices (dark, uniform). D, SEM backscatter image showing detail of rock in C. Oxide-rich domains, which surround serpentinized olivine grains, are interpreted as pyroclasts that have sintered at point contacts. Irregular oxide-free serpentine domains are visible and represent the original porosity in the pyroclastic rock. O ¼ olivine macrocrysts and phenocrysts, C ¼ calcite domains, S ¼ serpentine domains.

pretations have been given for spherical juvenile pyroclasts (“globules”) in peralkaline pyroclastic deposits (Schmincke 1974; Hay et al. 1979), for basaltic achneliths (Walker and Croasdale 1971), and for lunar glassy spherical pyroclasts (Heiken et al. 1974). Another key property is that glasses are virtually impossible to form from low-silica, carbonate-rich melts in laboratory experiments (R. Brooker, pers. comm.), even under extreme quenching due to the very fast crystallization kinetics in very low-viscosity melt compositions. Thus, the welding processes in silicic pyroclasts that involve the long-lived existence of supercooled melt or glass are unlikely to occur in kimberlites.

Further, if kimberlite magma is broken up into droplets during explosive eruption, these droplets can be expected to flash-crystallize if cooling by degassing shifts the melt below the liquidus, and all melt will be eliminated if conditions shift below the solidus. We infer that the initial melt in erupting kimberlite magma can have three different forms in depositing pyroclasts: as pure melt (with suspended olivine phenocrysts and xenocrysts), as partially crystallized clasts with some interstitial melt, and in a completely crystalline state. However, the highly viscous glassy state will not occur. A final property of juvenile kimberlite pyroclasts is that vesicles are typically absent to low in

368

R. J. BROWN ET AL.

Figure 13. Graphic illustrating possible melt-gas textural relationships. Sintered melt particles on the left. The gas phase occurs in void space between melt particles and forms interstices with thinned asymptotic terminations. This is the relationship seen in coherent kimberlite (CK) in figures 6A–6D and 12 and is interpreted to result from the sintering of melt particles. This is contrasted with the textural relationships of vesiculation, where the melt forms asymptotic bubble walls.

abundance: primary pyroclasts occur as pelletal clasts and poorly vesicular amoeboid lapilli (Mitchell 1986). Their poorly vesiculated character has been attributed to phreatomagmatic processes, with the magma being quenched before vesiculation takes place (Lorenz 1975, 1986; Lorenz and Kurzslaukis 2007). An alternative explanation is that erupting kimberlite melts have such low viscosities that the separation of gas and melt is very efficient due to very rapid coalescence of both bubbles and the melt phase into liquid drops. The timescale, τ , for such processes is approximately given by the ratio of surface tension to viscosity: τ¼

μr ; 2σ

(1)

where μ is the melt viscosity, r is the typical vesicle or drop dimension, and σ is the surface tension. For σ ¼ 0:05–0:17 Nm (surface tension of dacite and basalt magmas; Khitarov et al. 1979; Mangan and Sisson 2005), μ ¼ 0:1–1 Pa s (Sparks et al. 2006),

and r ¼ 1 mm, the timescale is 10
1 –10
4 s. This efficient separation process can be observed in the opening of a pressurized lemonade bottle, where the dissolved gas propels an exsolving gas-liquid mixture into the atmosphere and the liquid drops fall back to the ground and coalesce to form thoroughly degassed liquid. Emplacement of Welded Kimberlite Rocks. It is increasingly recognized that large-volume bodies of CK pass gradationally into volcaniclastic lithofacies (e.g., Clement and Skinner 1985; Clement and Reid 1989; Field and Scott Smith 1999; Hetman et al. 2004; Skinner and Marsh 2004). MVK comprises abundant fragmental olivine macrocrysts and phenocrysts, pelletal lapilli (consisting of a thin selvage of crystallized kimberlite groundmass around an olivine crystal), and variable abundances of lithic clasts. The interstices are typically filled with a serpentine-diopside cement (see Clement and Skinner 1985; Mitchell 1986; Skinner and Marsh 2004; Sparks et al. 2006). Lithofacies that share characteristics of both CK and MVK have been recognized at several pipes at the contact between these two rock types. Transitions from CK to MVK typically show (1) an increase in the abundance of olivine crystals and of interstitial microlitic diopside, (2) an increase in the abundance of discrete juvenile clasts, and (3) a decrease in the abundance of matrix calcite (e.g., Field and Scott Smith 1999; Hetman et al. 2004; Skinner and Marsh 2004). The gradation from CK into MVK suggests that the two rock types may have a similar genesis. Recent experimental and theoretical studies, along with particle size distribution studies, suggest that fluidization plays a major role in the formation of MVK (Sparks et al. 2006; Walters et al. 2006; Gernon et al., forthcoming a). It is envisaged that as the eruption starts to wane, pyroclasts start to accumulate in the pipe (Sparks et al. 2006). Proximal fallout of material from eruption jets is chaotic, and high rates of deposition are common (Sable et al. 2006). The deposits may not show the characteristic features of more distal fall deposits (e.g., well sorted, mantle bedded) and may be poorly sorted, chaotic, or massive to very poorly bedded. Pyroclasts accumulating in a pipe may become fluidized by streaming gases and explosive jets erupting into the base of the accumulating pile of material, which could result in the loss of fine ash by elutriation and the diminution of large lithic clasts. Transitions from MVK into CK (e.g., Field and Scott Smith 1999; Hetman et al. 2004; Skinner and Marsh 2004) suggest that CK is also deposited by fluidized systems. Experimental work indicates that particles fluidized at temperatures above their

Journal of Geology

WELDING OF PYROCLASTS

minimum sintering temperature (T s ), defined as the temperature at which the particle surface starts to soften, can rapidly defluidize by the sintering of particles (Gluckman et al. 1975). This occurs by viscoplastic flow between adjacent particles, and the efficacy of this is a balance between the time needed for growth of sinter necks and the characteristic times of quiescent motion of particles in a bed (Seville et al. 1998). Gas flow rates in excess of the minimum fluidization velocity (U mf ) are required to maintain fluidization at T > T s (Gluckman et al. 1975; Siegell 1984). Defluidization proceeds as particles aggregate along the base and margins of the bed. We suggest that hot pyroclasts in a kimberlite pipe, erupted at >T s , are capable of forming welded deposits from fluidized systems. Sintering of hot particles is aided by high temperatures and very low viscosities, which allow necks to form between adjacent particles by viscoplastic flow (Gluckman et al. 1975; Seville et al. 2000). In cases of extreme low viscosity, particles can almost instantaneously form a homogeneous liquid (“coalescence” in Branney and Kokelaar 1992). Coalescence generates spatter-fed lava flows and lava lakes at the base of Hawaiian fire fountains and can occur when accumulating particles have viscosities of <105 Pa s (Sumner et al. 2005). Coalescence, subsequent flowage, and postdepositional crystallization can obliterate original particle outlines in the liquid (e.g., Schmincke 1974). Foliations may not be preserved. Substituting values of σ ¼ 0:3, μ ¼ 1000 (from Sumner et al. 2005), and r ¼ 50 mm into equation (1) gives timescales for coalescence of a bed of basaltic particles in the region of 10–102 s. It is possible that texturally uniform bodies of CK in kimberlite pipes may have formed from the coalescence of low-viscosity pyroclasts at the base of a fire fountain. Comparable processes are well known at modern basaltic and carbonatitic volcanoes (e.g., Kilaeau, Hawaii; Erta Ale, Ethiopia; and Oldoinyo Lengai, Tanzania; Tazieff 1994). Formation of CK by passive rising of degassed kimberlite magma, a process well known for forming lava lakes, is discounted because quiet effusion cannot account for the presence of lithic clasts within CK. Further evidence for a pyroclastic origin of CK is provided by Nowicki et al. (forthcoming), who demonstrate that large bodies of CK in Canadian kimberlite pipes are depleted in incompatible elements relative to kimberlite dykes but enriched in incompatible elements relative to nonwelded kimberlite facies (“pyroclastic kimberlite”). They relate this to the loss of kimberlite groundmass by the elutriation of fine ash during explosive fire fountain–type eruptions. A similar phenomenon

369

is seen in pyroclastic density currents, where the elutriation of fine vitric ash results in the bulk rock chemistry of the deposited ignimbrite being depleted in incompatible elements relative to the bulk rock chemistry of the parent magma (Horwell et al. 2001). Co-ignimbrite ash fall deposits are consequently enriched in incompatible elements. We suggest that rapid transitions between welded and nonwelded facies are probable during kimberlite eruptions. Three contrasted states of the initial melt phase in the magma are envisaged on deposition following magma fragmentation. Each results in distinct deposit types. The first is where the initial melt in the magma remains pure melt (above the liquidus temperature) through fragmentation. The depositing pyroclasts coalesce almost immediately to form a degassed homogeneous liquid. Ephemeral lava lakes and lava flows could develop within the conduit from this process. This results in the formation of uniformly textured CK (fig. 6E). The second state is where initial melt has largely but not completely crystallized during fragmentation, retaining some residual melt. The resulting pyroclasts are difficult to deform but are still sticky and are able to sinter at point contacts by viscoplastic flow. Pyroclasts are deposited at temperatures between the liquidus and the solidus, i.e., the sintering temperature of Gluckman et al. (1975). This results in incipiently welded, porous pyroclastic deposits (e.g., figs. 6A, 12). The final state is where the initial melt has fully crystallized during fragmentation (subsolidus). The pyroclasts form a porous, nonwelded granular deposit on deposition (e.g., MVK). Transitions between these states could result in rapid changes in deposit characteristics. The effects of these changes in a fluidized system remain to be explored: clasts start to sinter and aggregate above the solidus within fluidized systems if small volumes of melt are present (Gluckman et al. 1975; Siegell 1984; Seville et al. 1998). However, if clasts are erupted as liquid drops, fluidized systems cannot develop. Densification during Welding of Kimberlite Pyroclasts. The welded rocks in K2 and B/K9 do not

show the compactional welding textures and foliations of silicic welded rocks (e.g., eutaxitic textures, fiamme, and flattened spatter clasts in agglutinates). During postdeposition compactional welding, silicic pumice and scoria clasts deform principally by losing their porosity (e.g., Ragan and Sheridan 1972; Riehle et al. 1995). Typically, the aspect ratios of fiamme can be explained only by collapse of vesicles, and it is unnecessary to invoke additional pure shear compaction, although this cannot be excluded. Additional pure or simple shear

370

R. J. BROWN ET AL.

can occur to form parataxitic rocks and lineations in the plane of the compactional foliation (Smith 1960). The strong foliations observed in silicic welded rocks are formed as a consequence of two main factors, namely, the porosity collapse of vesicular clasts and the existence of the glassy state. Neither of these factors occurs in kimberlite pyroclasts. Kimberlite pyroclasts generally comprise subspherical to amoeboid-dense fragments of magma and melt-coated grains (pelletal clasts) that are not capable of deforming by loss of internal porosity. Instead, any compaction during postdepositional welding must be accommodated by loss of interparticle (deposit) pore space and spreading. A situation may occur where the pores within a hot welded kimberlite pyroclastic deposit retain low pore pressures, perhaps by being connected to the atmosphere. The pressure difference between the load of the juvenile clasts (lithostatic pressure) and the pore pressure may induce rapid closing of the pores by the migration of remaining liquid melt under strong capillary and surface tension forces, and the rock becomes densely welded. This may be partly analogous to processes observed during the liquid phase sintering of metal alloys and ceramics (e.g., Park et al. 1984; German 1985) and is enhanced when the particle size is small (Hwang et al. 1987). Further investigation is required for volcanic systems. The high abundances of macrocrysts and phenocrysts in CK could inhibit compaction of CK because of the formation of percolation networks of solid particles. Experimental work on composite metal-ceramic powders indicates that their densification potential is strongly influenced by the characteristics of the particulate mixture (e.g., the nature of the phase fractions and the size, shape, and strength of the particles; Bouvard and Lange 1991; Bouvard 2000). Percolation networks of hard particles that can resist, wholly or partially, compactional loading can start to develop at relatively low-volume abundances of hard phases (e.g., 20%– 30%), depending on the size ratio of hard to soft particles (Bouvard and Lange 1991). CK in the studied rocks typically comprises 15–25 vol% of olivine phenocrysts and macrocrysts along with 2– 20 vol% lithic clasts. The total volume of hard particles in most CK units may well exceed the percolation threshold. The deformation potential of CK is therefore probably much less than, for example, a silicic pyroclastic deposit that comprises vesicular pumice clasts and ductile glass shards and that lacks abundant solid particles (lithic clasts and crystals). Strong welding-compaction folia-

tions, which are common in silicic welded rocks, seem unlikely to develop in CK. Implications for Models of Kimberlite Volcanism.

The features outlined here for CK in K2 and B/K9 that indicate a pyroclastic origin are shared by many other CK-type rocks in other kimberlite pipes. We would expect some of these other rocks to be welded pyroclastic rocks on further investigation. This has implications for models of kimberlite volcanism. Hetman et al. (2004) and Skinner and Marsh (2004) interpreted transitions between CK and MVK as the point of volatile exsolution within a frozen degassing front. Given the inherent dynamical difficulties of preserving such a feature as a deposit (Sparks et al. 2006), our alternative model in which these transitions represent changes between welding states seems appealing and has the benefit of satisfying both the microscopic and the macroscopic features of the two rock types. Such transitions are common in almost every other volcanic system, modern and ancient. What causes the transition from one state to the other remains to be constrained in kimberlite eruptions, but as in other volcanic systems, changes in eruptive temperature, mass flux, deposition rate, and volatile content of the erupting gas-particulate flows are likely to have profound effects on the nature of the deposits (e.g., Head and Wilson 1989). Conclusions Lithofacies occurring in two kimberlite pipes in South Africa (Venetia K2) and Botswana (Orapa B/K9) are here interpreted as welded pyroclastic rocks on the basis of geometry and composition, structural considerations, and macroscopic and microscopic textures. The rocks show a range of microscopic textures interpreted as densely welded to incipiently welded. We infer that the dynamics and mechanics of the welding of kimberlite pyroclasts are different from those that occur during the welding of glassy, vesiculated pyroclasts in silicic volcanic systems because of the low melt viscosity and the efficient separation of volatiles and melt, such that vesicular clasts are rarely preserved. Critically, the glassy state is difficult to form in silicadeficient magmas, and the crystallization kinetics is rapid. Three major types of deposits are recognized (densely welded, incipiently welded, and nonwelded), and we relate these to the state of the initial melt phase in the pyroclasts (molten, partially molten, and solid) on deposition in the pipe. Because of the absence of a glassy state, transitions between these pyroclast conditions and, consequently, between welding intensity may be

Journal of Geology

WELDING OF PYROCLASTS

rapid. Gradations between nonwelded deposits and welded deposits in several kimberlite pipes, together with overlapping in their macroscopic textural features, indicate that welded rocks can be deposited from large-scale fluidized systems within kimberlite pipes. The relatively high percentage of nondeformable particles (crystals and lithic clasts) in kimberlite pyroclastic deposits may inhibit the development of compaction deformation foliations through the formation of percolation networks of hard particles. CK-type rocks are common in many kimberlite pipes (hypabyssal kimberlites; e.g., Nowicki et al. 2004; Field et al., forthcoming), and we speculate that they may prove to be welded pyroclastic rocks

371

on further investigation. The processes outlined here may be generally applicable to other lowviscosity magmas (e.g., basaltic, peralkaline, carbonatite, and melilitite magmas).

ACKNOWLEDGMENTS

We acknowledge funding from the De Beers Group and Debswana. B. Buse acknowledges a Natural Environment Research Council Cooperative Award in Science and Engineering (De Beers) studentship awarded at the University of Bristol. K. Russell and an anonymous reviewer greatly improved the manuscript.

REFERENCES CITED

Almond, D. C. 1971. Ignimbrite vents in the Sabaloka cauldron, Sudan. Geol. Mag. 108:161–176. Andrews, J. R., and Emeleus, C. H. 1975. Structural aspects of kimberlite dyke and sheet intrusion in southwest Greenland. Phys. Chem. Earth 9:43–50. Armstrong, J. P.; Wilson, M.; Barnett, R. L.; Nowicki, T.; and Kjarsgaard, B. A. 2004. Mineralogy of primary carbonate-bearing hypabyssal kimberlite, Lac de Gras, Slave Province, Northwest Territories, Canada. Lithos 76:415–433. Bachmann, O.; Dungan, M. A.; and Lipman, P. W. 2000. Voluminous lava-like precursor to a major ash-flow tuff: low-column pyroclastic eruption of the Pagosa Peak Dacite, San Juan volcanic field, Colorado. J. Volcanol. Geotherm. Res. 98:153–171. Borsi, S.; Marnelli, G.; Mazzoncini, F.; Mittenpergher, M.; and Tedesco, C. 1963. Reconnaissance of some ignimbrites of Pantelleria and Eolian Islands. Bull. Volcanol. 25:359–367. Bouvard, D. 2000. Densification behaviour of mixtures of hard and soft powders under pressure. Powder Technol. 111:231–239. Bouvard, D., and Lange, F. F. 1991. Relation between percolation and particle coordination in binary powder mixtures. Acta Metall. Mater. 39:3083–3090. Branney, M. J.; Barry, T. L.; and Godchaux, M. 2004. Sheathfolds in rheomorphic ignimbrites. Bull. Volcanol. 66:485–491. Branney, M. J., and Kokelaar, B. P. 1992. A reappraisal of ignimbrite emplacement: progressive aggradation and changes from particulate to non-particulate flow during emplacement of high-grade ignimbrite. Bull. Volcanol. 54:504–520. Branney, M. J.; Kokelaar, B. P.; and McConnell, B. J. 1992. The Bad Step Tuff: a lava-like rheomorphic ignimbrite in a calc-alkaline piecemeal caldera, English Lake District. Bull. Volcanol. 54:187–199. Brooker, R. A.; Kohn, S.; Holloway, J. R.; and McMillan,

P. F. 2001. Structural controls on the solubility of CO2 in silicate melts. I. Bulk solubility data. Chem. Geol. 174:225–239. Brown, R. J.; Kavanagh, J.; Tait, M.; Sparks, R. S. J.; and Field, M. 2007. Mechanically disrupted and chemically weakened zones in segmented dike systems cause vent localization: evidence from kimberlite volcanic systems. Geology 35:815–818. Brown, R. J.; Tait, M.; Field, M.; and Sparks, R. S. J. Forthcoming. Geology of a complex kimberlite pipe (K2 pipe, South Africa): insights into conduit processes during explosive ultrabasic eruptions. Bull. Volcanol., doi:10.1007/s00445-008-0211-4. Bruce, P. M., and Huppert, H. E. 1989. Thermal control of basaltic fissure eruptions. Nature 342:665–667. Cas, R. A. F.; Hayman, P. C.; Pittari, A.; and Porritt, L. A. Forthcoming. A new approach to kimberlite facies terminology using a revised general approach to the nomenclature of all volcanic rocks and deposits: descriptive to genetic. J. Volcanol. Geotherm. Res., doi: 10.1016/j.jvolgeores.2007.12.018. Clement, C. R. 1973. Kimberlites from the Kao pipe, Lesotho. In Nixon, P. H., ed. Lesotho kimberlites. Maseru, Lesotho National Development Program, p. 110–121. ———. 1975. The emplacement of some diatreme facies kimberlites. Phys. Chem. Earth 9:51–59. ———. 1982. A comparative geological study of some major kimberlite pipes in the Northern Cape and Orange Free State. PhD thesis, University of Cape Town. Clement, C. R., and Reid, A. M. 1989. The origin of kimberlite pipes: an interpretation based on the synthesis of geological features displayed by southern African occurrences. In Ross, J.; Jaques, A. L.; Fergusan, J.; Green, D. H.; O’Reilly, S. Y.; Danchin, R. V.; and Janse, A. J. A., eds. Kimberlites and related rocks. Geol. Soc. Aust. Spec. Publ. 14:632–646.

372

R. J. BROWN ET AL.

Clement, C. R., and Skinner, E. M. W. 1985. A texturalgenetic classification scheme for kimberlites. Trans. Geol. Soc. S. Afr. 88:403–409. Dawson, J. B. 1980. Kimberlites and their xenoliths. New York, Springer, 252 p. Dawson, J. B., and Hawthorne, J. B. 1970. Intrusion features of some hypabyssal South African kimberlites. Bull. Volcanol. 34:740–757. ———. 1973. Magmatic sedimentation and carbonatitic differentiation in kimberlite sills in Benfontein, South Africa. J. Geol. Soc. Lond. 129:61–85. Donaldson, C. H., and Reid, A. M. 1982. Multiple intrusion of a kimberlite dyke. Trans. Geol. Soc. S. Afr. 85: 1–12. Downes, P. J.; Ferguson, D.; and Griffin, B. J. 2007. Volcanology of the Aries micaceous kimberlite, central Kimberley basin, Western Australia. J. Volcanol. Geotherm. Res. 159:85–107. Ekren, E. B., and Byers, F. M., Jr. 1976. Ash-flow fissure vent in west-central Nevada. Geology 4:247–251. Fedortchouk, Y., and Canil, D. 2004. Intensive variables in kimberlite magmas, Lac de Gras, Canada, and implications for diamond survival. J. Petrol. 45:1725– 1745. Field, M.; Gibson, J. G.; Wilkes, T. A.; Gababotse, J.; and Khutjwe, P. 1997. The geology of the Orapa A/K1 Kimberlite, Botswana: further insight into the emplacement of kimberlite pipes. In Dobretsov, N. L.; Gol’din, S. V.; Kontorovich, A. E.; Polyakov, G. V.; and Sobolev, N. V., eds. Proceedings of the Sixth International Kimberlite Conference. Russ. Geol. Geophys. 38:24–39. Field, M.; Robey, J. V.; Stiefenhofer, J.; and Kurszlaukis, S. Forthcoming. Kimberlite-hosted diamond deposits of southern Africa: a review. Ore Geol. Rev. Field, M., and Scott Smith, B. H. 1999. Contrasting geology and near-surface emplacement of kimberlite pipes in southern Africa and Canada. In International Kimberlite Conference, 7th (Cape Town, South Africa, 1999), Proc., p. 217–237. German, R. M. 1985. Liquid phase sintering. New York, Plenum, 240 p. Gernon, T. M.; Gilbertson, M. A.; Sparks, R. S. J.; and Field, M. Forthcoming a. Gas-fluidisation in an experimental tapered bed: insights into processes in diverging volcanic conduits. J. Volcanol. Geotherm. Res., doi: 10.1016/j.jvolgeores.2007.12.034. Gernon, T. M.; Sparks, R. S. J.; and Field, M. Forthcoming b. Degassing structures in volcaniclastic kimberlite: examples from southern African kimberlite pipes. J. Volcanol. Geotherm. Res., doi:10.1016/j.jvolgeores .2007.12.035. Gluckman, M. J.; Yerushalmi, J.; and Squires, A. M. 1975. Defluidization characteristics of sticky or agglomerating beds. In Keairns, D. L., ed. Fluidization technology. New York, McGraw-Hill, p. 395–422. Green, J. C., and Fitz, T. J. 1993. Extensive felsic lavas and rheoignimbrites in the Keweenawan midcontinent rift plateau volcanics, Minnesota: petrographic and field recognition. J. Volcanol. Geotherm. Res. 54:177–196.

Grunder, A. L.; Laporte, D.; and Druitt, T. H. 2005. Experimental and textural investigation of welding: effects of compaction, sintering, and vapor-phase crystallization in the rhyolitic Rattlesnake Tuff. J. Volcanol. Geotherm. Res. 142:89–104. Guest, J. E., and Rogers, P. S. 1967. The sintering of glass and its relationship to welding in ignimbrites. Proc. Geol. Soc. Lond. 1641:174–177. Hawthorne, J. B. 1975. Model of a kimberlite pipe. Phys. Chem. Earth 9:1–15. Hay, R. L.; Hildreth, W.; and Lambe, R. N. 1979. Globule ignimbrite of Mount Suswa, Kenya. In Chapin, C. E., and Elston, W. E., eds. Ash-flow tuffs. Geol. Soc. Am. Spec. Publ. 180:167–175. Head, J. W., and Wilson, L. 1989. Basaltic pyroclastic eruptions: influence of gas-release patterns and volume fluxes on fountain structure, and the formation of cinder cones, spatter cones, rootless flows, lava ponds and lava flows. J. Volcanol. Geotherm. Res. 37:261– 271. Heiken, G.; McKay, D. S.; and Brown, R. W. 1974. Lunar deposits of possible pyroclastic origin. Geochim. Cosmochim. Acta 38:1708–1718. Henry, C. D., and Wolff, J. A. 1992. Distinguishing strongly rheomorphic tuffs from extensive silicic lavas. Bull. Volcanol. 54:171–186. Hetman, C. M.; Scott Smith, B. H.; Paul, J. L.; and Winter, F. W. 2004. Geology of the Gahcho Kué kimberlite pipes, NWT, Canada: root to diatreme transition zones. Lithos 76:51–74. Holness, M. B., and Humprheys, M. C. S. 2003. The Traigh Bhàn na Sgùrra sill, Isle of Mull: flow localization in a major magma conduit. J. Petrol. 44:1961– 1976. Horwell, C. J.; Braña, L. P.; Sparks, R. S. J.; Murphy, M. D.; and Hards, V. L. 2001. A geochemical investigation of fragmentation and physical fractionation in pyroclastic flows in the Soufrière Hills volcano, Montserrat. J. Volcanol. Geotherm. Res. 109:247–262. Hwang, K. S.; German, R. M.; and Lenel, F. V. 1987. Capillary forces between spheres during agglomeration and liquid-phase sintering. Metall. Trans. A 18: 11–17. Kano, K.; Matsuura, H.; and Yamauchi, S. 1997. Miocene welded tuff infilling a funnel-shaped eruption conduit, Shiotani, southeast of Matsue, SW Japan. Bull. Volcanol. 59:125–135. Khitarov, N. I.; Lebedev, Y. B.; Dorfman, A. M.; and Bagdasarov, N. S. 1979. Effects of temperature, pressure and volatiles on the surface tension of molten basalt. Geochem. Int. 16:78–86. Kurszlaukis, S., and Barnett, W. P. 2003. Volcanological and structural aspects of the Venetia kimberlite cluster: a case study of South African kimberlite maardiatreme volcanoes. S. Afr. J. Earth Sci. 106:165–192. Lame, O.; Bellet, D.; Di Michiel, M.; and Bouvard, D. 2003. In situ microtomography investigation of metal powder compacts during sintering. Nucl. Instrum. Methods Phys. Res. B 200:287–294. Le Pennec, J.-L., and Fernandez, A. 1992. Fragmental lava

Journal of Geology

WELDING OF PYROCLASTS

versus welded ignimbrite on Mount Etna: arguments inferred from crystal preferred orientation. J. Volcanol. Geotherm. Res. 51:323–337. Le Roex, A. P.; Bell, D. R.; and Davis, P. 2003. Petrogenesis of group I kimberlites from Kimberley, South Africa: evidence from bulk rock geochemistry. J. Petrol. 44: 2261–2286. Liu, J.; Lal, A.; and German, R. M. 1999. Densification and shape retention in supersolidus liquid phase sintering. Acta Material. 18:4615–4626. Lorenz, V. 1975. Formation of phreatomagmatic maardiatreme volcanoes and its relevance to kimberlite diatremes. Phys. Chem. Earth 9:17–29. ———. 1986. On the growth of maars and diatremes and its relevance to the formation of tuff rings. Bull. Volcanol. 48:265–274. Lorenz, V., and Kurszlaukis, S. 2007. Root zone processes in the phreatomagmatic emplacement model and consequences for the evolution of maar-diatreme volcanoes. J. Volcanol. Geotherm. Res. 159:4–32. Macedonio, G.; Dobran, F.; and Neri, A. 1994. Erosion processes in volcanic conduits and application to the AD 79 eruption of Vesuvius. Earth Planet. Sci. Lett. 121:137–152. Mahood, G. A. 1984. Pyroclastic rocks and calderas associated with strongly peralkaline magmatism. J. Geophys. Res. 89:8540–8552. Mangan, M., and Sisson, T. 2005. Evolution of melt-vapor surface tension in silicic volcanic systems: experiments with hydrous melts. J. Geophys. Res. vol. 110, doi:10.1029/2004JB003215. Milner, S. C.; Duncan, A. R.; and Ewart, A. 1992. Quartz latite rheoignimbrite flows of the Etendeka Formation, northwestern Namibia. Bull. Volcanol. 54:200–219. Mitchell, R. H. 1975. Geology, magnetic expression, and structural control of the central Somerset Island kimberlites. Can. J. Earth Sci. 12:757–764. ———. 1984. Mineralogy and origin of carbonate-rich segregations in a composite kimberlite sill. N. Jahrb. Mineral. Abh. 150:185–197. ———. 1986. Kimberlites: mineralogy, geochemistry and petrology. New York, Plenum, 442 p. ———. 1995. Kimberlites, orangeites and related rocks. New York, Plenum, 410 p. Naidoo, P.; Stiefenhofer, J.; Field, M.; and Dobbe, R. 2004. Recent advances in the geology of Koffiefontein mine, Free State Province, South Africa. Lithos 76:161–182. Nowicki, T.; Crawford, B.; Dyck, D.; Carlson, J.; McElroy, R.; Oshust, P.; and Helmstaedt, H. 2004. The geology of kimberlite pipes of the Ekati property, Northwest Territories, Canada. Lithos 76:1–27. Nowicki, T.; Porritt, L.; Crawford, B.; and Kjarsgaard, B. Forthcoming. Geochemical trends in kimberlites of the Ekati property, Northwest Territories, Canada: insights on volcanic and resedimentation processes. J. Volcanol. Geotherm. Res., doi:0.1016/j.jvolgeores .2007.12.030. Park, H. H.; Cho, S. J.; and Yoon, D. N. 1984. Pore filling processes in liquid phase sintering. Metall. Trans. A 15:1075–1080.

373

Peterson, D. W. 1979. Significance of the flattening of pumice fragments in ash-flow tuffs. In Chapin, C. E., and Elston, W. E., eds. Ash-flow tuffs. Geol. Soc. Am. Spec. Publ. 180:195–204. Phillips, D.; Kiviets, G. B.; Barton, E. S.; Smith, C. B.; Viljoen, K. S.; and Fourie, L. F. 1999. 40 Ar=39 Ar dating of kimberlites and related rocks: problems and solution. In International Kimberlite Conference, 7th (Cape Town, South Africa, 1999), Proc., p. 677–688. Price, S. E.; Russell, J. K.; and Kopylova, M. G. 2000. Primitive magma from the Jericho Pipe, NWT, Canada: constraints on primary kimberlite melt chemistry. J. Petrol. 41:789–808. Quane, S. L., and Russell, J. K. 2005. Welding: insights from high-temperature analog experiments. J. Volcanol. Geotherm. Res. 142:67–87. ———. 2006. Bulk and particle strain analysis in hightemperature deformation experiments. J. Volcanol. Geotherm. Res. 154:63–73. Ragan, D. M., and Sheridan, M. F. 1972. Compaction of the Bishop Tuff, California. Geol. Soc. Am. Bull. 83: 95–106. Reedman, A. J.; Park, K. H.; Merriman, R. J.; and Kim, S. E. 1987. Welded Tuff infilling a volcanic vent at Weolsong, Republic of Korea. Bull. Volcanol. 49: 541–546. Riehle, J. R.; Miller, T. F.; and Bailey, R. A. 1995. Cooling, degassing and compaction of rhyolitic ash-flow tuffs: a computational model. Bull. Volcanol. 57:319–336. Ross, C. S., and Smith, R. L. 1961. Ash-flow tuffs: their origin, geological relations and identification. U.S. Geol. Surv. Prof. Pap. 366:1–77. Russell, J. K., and Quane, S. L. 2005. Rheology of welding: inversion of field constraints. J. Volcanol. Geotherm. Res. 142:173–191. Sable, J. E.; Houghton, B. F.; Wilson, C. J. N.; and Carey, R. J. 2006. Complex proximal sedimentation from Plinian plumes: the example of Tarawera 1886. Bull. Volcanol. 69:89–103. Schmincke, H. U. 1974. Volcanological aspects of peralkaline silicic welded ash-flow tuffs. Bull. Volcanol. 38: 1–43. Seville, J. P. K.; Silomon-Pflug, H.; and Knight, P. C. 1998. Modelling of sintering in high temperature gas fluidisation. Powder Technol. 97:160–169. Seville, J. P. K.; Willett, C. D.; and Knight, P. C. 2000. Interparticle forces in fluidisation: a review. Powder Technol. 113:261–268. Siegell, J. H. 1984. High-temperature defluidization. Powder Technol. 38:13–22. Skinner, E. M. W. Forthcoming. The emplacement of class 1 kimberlites. J. Volcanol. Geotherm. Res., doi: 10.1016/j.jvolgeores.2007.12.022. Skinner, E. M. W., and Marsh, J. S. 2004. Distinct kimberlite classes with contrasting eruption processes. Lithos 76:183–200. Smith, R. L. 1960. Zones and zonal variations in welded ash-flows. U.S. Geol. Surv. Prof. Pap. 354-F:149–159. Soriano, C.; Galindo, I.; Marti, J.; and Wolff, J. 2006.

374

R. J. BROWN ET AL.

Conduit-vent structures and related proximal deposits in the Las Cañadas caldera, Tenerife, Canary Islands. Bull. Volcanol. 69:217–231. Soriano, C.; Zafrilla, S.; Martí, J.; Bryan, S.; Cas, R.; and Ablay, G. 2002. Welding and rheomorphism of phonolitic fallout deposits from the Las Cañadas caldera, Tenerife, Canary Islands. Geol. Soc. Am. Bull. 114: 883–895. Sparks, R. S. J. 1988. Petrology and geochemistry of the Loch Ba ring-dyke, Mull (NW Scotland): an example of the extreme differentiation of tholeiitic magmas. Contrib. Mineral. Petrol. 100:446–461. Sparks, R. S. J.; Baker, L.; Brown, R. J.; Field, M.; Schumacher, J.; Stripp, G.; and Walters, A. 2006. Dynamical constraints on kimberlite volcanism. J. Volcanol. Geotherm. Res. 155:119–137. Sparks, R. S. J.; Tait, S. R.; and Yanev, Y. 1999. Dense welding caused by volatile resorption. J. Geol. Soc. Lond. 156:217–225. Sparks, R. S. J., and Wright, J. V. 1979. Welded air fall tuffs. In Chapin, C. E., and Elston, W. E., eds. Ash-flow tuffs. Geol. Soc. Am. Spec. Publ. 180:155–166. Stripp, G. R.; Field, M.; Schumacher, J. C.; and Sparks, R. S. J. 2006. Post-emplacement serpentinization and related hydrothermal metamorphism in a kimberlite from Venetia, South Africa. J. Metamorph. Geol. 24: 515–534. Sumner, J. M. 1998. Formation of clastogenic lava flows during fissure eruption and scoria cone collapse: the 1986 eruption of Izu-Oshima Volcano, eastern Japan. Bull. Volcanol. 60:195–212. Sumner, J. M.; Blake, S.; Matela, R. J.; and Wolff, J. A. 2005. Spatter. J. Volcanol. Geotherm. Res. 142:49–65. Sumner, J. M., and Branney, M. J. 2002. The emplacement history of a remarkable heterogeneous, chemically zoned, rheomorphic and locally lava-like ignimbrite: “TL” on Gran Canaria. J. Volcanol. Geotherm. Res. 115:109–138. Tazieff, H. 1994. Permanent lava lakes: observed facts and

induced mechanisms. J. Volcanol. Geotherm. Res. 63: 3–11. Valentine, G. A., and Groves, K. R. 1996. Entrainment of country rock during basaltic eruptions of the Lucero volcanic field, New Mexico. J. Geol. 104:71–90. Wada, Y. 1994. On the relationship between dike width and magma viscosity. J. Geophys. Res. 99:17,743– 17,755. Walker, G. P. L. 1983. Ignimbrite types and ignimbrite problems. J. Volcanol. Geotherm. Res. 17:65–88. Walker, G. P. L., and Croasdale, R. 1971. Characteristics of some basaltic pyroclastics. Bull. Volcanol. 35:303– 317. Walters, A.; Phillips, J.; Brown, R. J.; Field, M.; Gernon, T.; Stripp, G.; and Sparks, R. S. J. 2006. The role of fluidisation processes in the formation of volcaniclastic kimberlite: grain size observations and experimental investigation. J. Volcanol. Geotherm. Res. 155:119– 137. Webb, K. J.; Scott Smith, B. H.; Paul, J. L.; and Hetman, C. M. 2004. Geology of the Victor kimberlite, Attawapiskat, northern Ontario, Canada: cross-cutting and nested craters. Lithos 76:29–50. Wilson, L., and Head, J. W. 2007. An integrated model of kimberlite ascent and eruption. Nature 447:53–57. Wolff, J. A. 1986. Welded-tuff dykes, conduit closure, and lava dome growth at the end of explosive eruptions. J. Volcanol. Geotherm. Res. 28:379–384. Wright, J. V. 1980. Stratigraphy and geology of the welded air-fall tuffs of Pantelleria, Italy. Geol. Rundsch. 69: 263–291. Yasui, M., and Koyaguchi, T. 2004. Sequence and eruptive style of the 1783 eruption of Asama Volcano, central Japan: a case study of an andesitic explosive eruption generating fountain-fed lava flow, pumice fall, scoria flow and forming a cone. Bull. Volcanol. 66:243–262. Zuo, R.; Aulbach, E.; and Rodel, J. 2003. Experimental determination of sintering stresses and sintering viscosities. Acta Material. 51:4563–4574.