Zircon Macrocrysts from the Drybones Bay Kimberlite Pipe ( Northwest Territories , Canada ) : A High-Resolution Trace Element and Geochronological Study

Zircon macrocrysts in (sub)volcanic silica-undersaturated rocks are an important source of information about mantle processes and their relative timing with respect to magmatism. The present work describes variations in trace element (Sc, Ti, Y, Nb, lanthanides, Hf, Ta, Pb, Th, and U) and isotopic (U-Pb) composition of zircon from the Drybones Bay kimberlite, Northwest Territories, Canada. These data were acquired at a spatial resolution of ≤100 μm and correlated to the internal characteristics of macrocrysts (imaged using cathodoluminescence, CL). Six types of zircon were distinguished on the basis of its luminescence characteristics, with the majority of grains exhibiting more than one type of CL response. The oscillatory-zoned core and growth sectors of Drybones Bay zircon show consistent variations in rare-earth elements (REE), Hf, Th, and U. Their chondrite-normalized REE patterns are typical of macrocrystic zircon and exhibit extreme enrichment in heavy lanthanides and a positive Ce anomaly. Their Ti content decreases slightly from the core into growth sectors, but the Ti-in-zircon thermometry gives overlapping average crystallization temperatures (820 ± 26 ◦C to 781 ± 19 ◦C, respectively). There is no trace element or CL evidence for Pb loss or other forms of chemical re-equilibration. All distinct zircon types are concordant and give a U-Pb age of 445.6 ± 0.8 Ma. We interpret the examined macrocrysts as products of interaction between a shallow (<100 km) mantle source and transient kimberlitic melt.


Introduction
The Slave craton in northern Canada is a small (~200,000 km 2 ) but well-exposed fragment of Archean (4.05-2.55Ga) continental crust that hosts a variety of gold, silver, base metal, critical metal, and gemstone deposits.Notably, the latter include >300 kimberlite bodies (vents, dikes, and diatremes) [1].The discovery of first diamondiferous kimberlites in the Lac de Gras area in 1991 sparked the largest claim-staking rush in Canadian mining history [2], which ultimately paved the way to the development of five new mines to date (Ekati since 1998, Diavik since 2003, Jericho in 2006-2008, Snap Lake in 2008-2015, and Gahcho Kué since 2016).Nineteen individual bodies have been, or are presently, mined and about a dozen other kimberlite clusters are under active exploration (Figure 1).An increase in diamond production in recent years has moved Canada up to the second place in the global market in volume terms (23.2 million carats valued at $2.06 billion in 2017).
Minerals 2018, 8, x FOR PEER REVIEW 2 of 16 (Figure 1).An increase in diamond production in recent years has moved Canada up to the second place in the global market in volume terms (23.2 million carats valued at $2.06 billion in 2017).Only a small number of the 300-plus Slave kimberlites have been studied in significant detail.As can be expected, the bulk of published work is focused on the producing areas, especially the Ekati and Diavik mines at Lac de Gras.One of the least-studied areas with a poorly understood potential is the southwestern section of the craton hosting Early-to Mid-Paleozoic kimberlites, which are aligned in a roughly longitudinal fashion from Drybones Bay and Mud Lake along the northern shore Great Slave Lake to the Aquila and Cygnus pipes some 250 km to the north (Figure 1).Very little data is available on these rocks in the literature, which is practically limited to a study of mantle xenoliths from the Drybones Bay pipe [3] and U-Pb geochronological determinations for several of these bodies [4].This lack of information is regrettable because (1) the majority of tested kimberlites from this part of the craton were found to be diamondiferous; (2) geologically, these localities are separated from the other Slave fields by a major crustal-scale fault (see Regional Geology), and (2) similar-aged economic pipes are known in the North China and Siberian cratons, where they have been studied extensively [5,6].The present project was initiated to examine the petrology, mineralogy, and geochemistry of the SW Slave kimberlites using extensive exploration data and drill- Only a small number of the 300-plus Slave kimberlites have been studied in significant detail.As can be expected, the bulk of published work is focused on the producing areas, especially the Ekati and Diavik mines at Lac de Gras.One of the least-studied areas with a poorly understood potential is the southwestern section of the craton hosting Early-to Mid-Paleozoic kimberlites, which are aligned in a roughly longitudinal fashion from Drybones Bay and Mud Lake along the northern shore Great Slave Lake to the Aquila and Cygnus pipes some 250 km to the north (Figure 1).Very little data is available on these rocks in the literature, which is practically limited to a study of mantle xenoliths from the Drybones Bay pipe [3] and U-Pb geochronological determinations for several of these bodies [4].This lack of information is regrettable because (1) the majority of tested kimberlites from this part of the craton were found to be diamondiferous; (2) geologically, these localities are separated from the other Slave fields by a major crustal-scale fault (see Regional Geology), and (2) similar-aged economic pipes are known in the North China and Siberian cratons, where they have been studied extensively [5,6].The present project was initiated to examine the petrology, mineralogy, and geochemistry of the SW Slave kimberlites using extensive exploration data and drill-core material available at the Geological Materials Storage facility operated by the Northwest Territories Geological Survey.This contribution is concerned with the trace element and isotopic characteristics of zircon from the Drybones Bay pipe.This mineral was chosen because of its recognized value as a geochronometer, geothermometer and process indicator in the study of kimberlites and other mantle-derived rocks.

Regional Geology
About one-third of the presently exposed Archean bedrock in the Slave craton is represented by metasedimentary and metavolcanic supracrustal rocks of the Yellowknife Supergroup (2.61-2.71Ga), whereas some 65% is underlain by 2.58-2.64Ga granitoid intrusions [7].The Slave craton has been subdivided into eastern and western domains, based primarily on the presence of 2.85-4.05Ga Mesoarchean rocks and Pb-Nd isotopic variations [8,9].The Mesoarchean rocks comprise tonalitic to dioritic gneisses that so far have been reported only in the western domain [10].Lower crustal xenoliths from kimberlites in the Lac de Gras area (Figure 1) indicate the presence of Mesoarchean rocks also east of the domain boundary [11].The gneisses are overlain by a post-2.93Ga autochthonous cover sequence, which forms the base of the Yellowknife Supergroup [10].Supracrustal packages that can be correlated across the entire craton include 2.66-2.69Ga calc-alkaline subvolcanic-to-extrusive units and younger (2.63-2.66Ga) turbidites [12].Syn-to late-kinematic granitoid plutons are voluminous and widespread [7], but intrusions older than 2.61 Ga are limited to the southwestern part of the craton, including the tonalite-granodioritic Defeat suite (ca.2.62-2.63Ga) at Drybones Bay.Widespread felsic magmatism spanning the entire craton occurred from ca. 2.61 to 2.58 Ga, and produced both I-and S-type granitoids [13].Kimberlites are the most widespread form of post-Archean small-volume mantle-derived magmatism in the Slave craton.It spans a period of ~560 Ma from the emplacement of the 613-Ma Anuri pipe to the eruption of Eocene kimberlites in the Lac de Gras area [14,15].There is a distinct correlation between the spatial distribution of kimberlites and the timing of magmatism, which allows the craton to be subdivided into four domains (Figure 1): (1) [4,15].The reason for this distribution of kimberlite ages is presently unknown.The sublongitudinal boundary of the southwestern domain roughly coincides with the exposed margin of the Mesoarchean rocks, and the kimberlites of this domain (including Drybones Bay) define a linear N-S trend, which parallels the crustal-scale Beniah Lake fault system [4].

Local Geology and Exploration History
The Drybones Bay kimberlite "pipe" is situated in a small bay on the northern shore of Great Slave Lake 45 km SE of Yellowknife.The body is submerged under 38 m of water and blanketed by 67 m of overburden, which consists of lacustrine clay, sand, and a basal layer of boulder conglomerate [16].It intruded the 2.62-2.63Ga Defeat pluton comprising predominantly biotite tonalite and granodiorite with minor metasedimentary xenoliths derived from the Yellowknife Supergroup [17].The kimberlite body is interpreted to be a vent-like structure from drilling and geophysical surveys [18,19], but it is unknown how its morphology or size change with depth.The deepest drill hole (302 m) was stopped in volcaniclastic kimberlite.The structure is lobate in shape and measures ~900 × 400 m, covering an estimated area of 22 ha at the present erosion level [16].The sidewall contacts dip at a lower angle in the western lobe relative to the eastern lobe.The large size and shallow dipping contacts are characteristic of Canadian Class-2 kimberlites, rather than Class-3 pipes characteristic of the central Slave domain [19,20].The Drybones Bay structure comprises a core intrusion (termination of a feeder dike?), several crater-facies units and an epiclastic unit [16,17].Unambiguous tuffisitic facies are notably absent.
During the staking rush of the 1990s, diamond exploration on the northern shore of Great Slave Lake led to the discovery of a geophysical anomaly submerged beneath Drybones Bay (Figure 1).
The anomaly was staked and drilled in the winter of 1994; 24 m of kimberlite was intersected under a blanket of lacustrine clay.Low-level airborne and ground-based magnetic surveys were conducted over the claim block over the following year to delineate the kimberlite body with more precision.This was followed by two extensive drill programs (1995 and 1996) that included 20 holes ranging from 131 to 302 m in length and produced over 5.5 km of drill-core.Most of the recovered kimberlite core was split and processed by caustic fusion.Three samples weighing 0.64, 1.87, and 7.58 tonnes yielded 16, 26, and 48 macrodiamonds, respectively, measuring >0.5 mm in at least one dimension [21].Diamonds have been recovered from each of the lithostratigraphic units, but the highest exploration grades (24-39 carats per hundred tonnes) were measured for the core [16].In the past 24 years, the Drybones Bay project has changed hands several times and been the subject of litigation between developers and Aboriginal groups [22].

Materials and Methods
Cathodoluminescence (CL) images were obtained using a Reliotron VII instrument (Relion Ind., Bedford, MA, USA) operated at 8.5 eV and 400-450 mA, and an electronically cooled Nikon DS-Ri1 camera (Nikon Canada Inc., Mississauga, ON, Canada).The concentrations of major elements were determined by wavelength-dispersive spectrometry (WDS) using a Cameca SX100 (Cameca SAS, Gennevilliers, France) automated electron microprobe operated at 15 kV and 20 nA.The CL and BSE images were used for the selection of areas suitable for further trace element and isotopic analyses.
The abundances of selected trace elements and isotopes were measured by laser-ablation inductively coupled-plasma mass-spectrometry (LA-ICP-MS) using a 213-nm Nd-YAG laser (Elemental Scientific Lasers, Bozeman, MT, USA) connected to an Element 2 sector-field mass-spectrometer (Thermo Fisher Scientific Inc., Bremen, Germany).The zircon grains were analyzed in situ using a spot ablation mode and the following laser parameters: 30 µm beam size, ~0.03 mJ incident pulse energy, 3.1-4.0J/cm 2 energy density on a sample.Areas suitable for analysis were selected on the basis of reflected-light, BSE, and CL images.The ablation was performed in Ar and He atmospheres.The rate of oxide production was monitored during instrument tuning by measuring the ThO/Th ratio and kept below 0.2%.For LA-ICP-MS measurements, synthetic glass standard NIST SRM 610 [23] was employed for calibration and quality control.After considering potential spectral overlaps and molecular interferences, the following isotopes were chosen for analysis: 25 Mg, 29 Si, 45 Sc, 49 Ti, 51   232 Th, and 238 U. Silicon, measured by WDS, was employed as an internal standard.All analyses were performed in a low-resolution mode (~300) using Pt skimmer and sample cones.
The U-Pb geochronology of major zircon varieties was investigated using 100 µm spot analyses obtained at a repetition rate of 10 Hz with an incident laser beam pulse energy of 0.605-0.620mJ and energy density on sample in the range 7.6-8.0J/cm 2 .The detector was operated in counting mode ( 235 U, 232 Th, 206 Pb, 207 Pb, 202 Pb, 202 Hg) and analogue mode ( 238 U).Natural zircon standard GJ-1 (609 Ma; [24] was employed for calibration, and was run at the beginning, in the middle, and at the end of each analytical session.Natural zircon FC-1 (1099 ± 0.6 Ma; [25]) was used as a secondary standard, yielding accuracy and precision levels near 2%.The data reduction was done using the VisualAge DRS procedure [26] in the Iolite software [27].A Concordia diagram showing 2σ error ellipses was produced using Isoplot/Ex 3.75 [28].

CL Imaging
The examined zircon grains are macroscopically brown in color, homogeneous in transmitted light and BSE images, and vary from small angular fragments to large subhedral crystals reaching several mm in size.Cathodoluminescence revealed significant variations in the distribution of CL-active Minerals 2018, 8, 481 5 of 16 centers in the Drybones Bay zircons and their internal complexity (Figures 1 and 2).Based on their luminescence characteristics, several types of zircon were distinguished: • Type 1: oscillatory-zoned yellow to brown in CL images (Figure 2a); • Type 2: broad brown zones confined to blue-luminescing areas (Figure 2b); • Type 3: light blue, homogeneous or subtly zoned (Figure 2c); • Type 4: patchy yellow to greenish-yellowish (Figure 2d); • Type 5: yellowish-grey (Figure 2e); • Type 6: zoned dark-blue with brownish overtones (Figure 2f).active centers in the Drybones Bay zircons and their internal complexity (Figures 1 and 2).Based on their luminescence characteristics, several types of zircon were distinguished: • Type 1: oscillatory-zoned yellow to brown in CL images (Figure 2a); • Type 2: broad brown zones confined to blue-luminescing areas (Figure 2b); • Type 3: light blue, homogeneous or subtly zoned (Figure 2c); • Type 4: patchy yellow to greenish-yellowish (Figure 2d); • Type 5: yellowish-grey (Figure 2e); • Type 6: zoned dark-blue with brownish overtones (Figure 2f).In many cases, more than one type of luminescence is present within a single zoned grain, commonly as concentric growth zones (e.g., Figure 2a,b,f and Figure 3a).In addition, several large zircon grains exhibit a sector pattern consisting of bright yellow and light blue sectors (types 3 and 4, respectively) separated by a well-defined boundary (Figure 3b).The CL features revealed in the present work, including oscillatory zoning (Type 1), broad bands (Type 2), and irregularly shaped areas (Type 5), have been previously observed in zircon from other kimberlite localities, often in the same grain [29][30][31].
In many cases, more than one type of luminescence is present within a single zoned grain, commonly as concentric growth zones (e.g., Figures 2a,b,f and 3a).In addition, several large zircon grains exhibit a sector pattern consisting of bright yellow and light blue sectors (types 3 and 4, respectively) separated by a well-defined boundary (Figure 3b).The CL features revealed in the present work, including oscillatory zoning (Type 1), broad bands (Type 2), and irregularly shaped areas (Type 5), have been previously observed in zircon from other kimberlite localities, often in the same grain [29][30][31].

Trace Element Composition and Variations
In this work, we studied zircon grains with all types of CL response.In total, 90 trace-element analyses were done in samples imaged using CL.Our data show that the Drybones Bay zircon macrocrysts exhibit appreciable variations in trace element abundances and ratios, particularly with regard to rare-earth elements (REE = La-Lu).All samples are relatively enriched in Ta (1.8-12.3ppm), but contain moderate levels of Hf (8200-10,900 ppm) at low levels of REE (<1.5 ppm Lu, <60 ppm Y).None of the seven types exhibit Eu anomalies.The average δEu ratio, which reflects deviation of the Eu concentration from a value interpolated from the chondrite-normalized Sm and Gd abundances, and is measured as Eucn/[0.5 × (Smcn + Gdcn)], is 1.0 ± 0.2.These compositional characteristics are consistent with kimberlitic zircon [32] and rule out crustal rocks or the Defeat granitoids as a potential source of these macrocrysts (cf.[33]).Another notable characteristic of the studied material is its enrichment in Sc (570-930 ppm), which is far in excess of what has been reported for felsic rocks [34].
The most significant compositional differences are observed between grains showing blue and yellow luminescence (Table 1), in accord with the observations made previously by Belousova et al. [30].Areas luminescing blue show the lowest levels of REE and Y (5-10 ppm and 6-11 ppm, respectively, in Type 3; 12-18 and 14-22 ppm, respectively, in Type 6; Figure 4a).In contrast, those showing yellow or yellowish-brown response contain the highest levels of these elements among the examined samples (23-53 ppm REE and 26-59 ppm Y in Type 1; 19-33 and 21-36 ppm, respectively, in Type 4).These differences are particularly striking in sector-zoned grains (Figure 3b), where REE+Y levels increase two-to fourfold across the sector boundary (Figure 4a).These variations are not accompanied by any consistent changes in Sc content.Grains showing yellow luminescence are also enriched in Th and U (3-9 and 8-20 ppm, respectively, in Type 1) relative to blue areas (≤2 ppm Th and ≤5 ppm U in Types 3 and 6; Figure 4b).The Th/U ratio (Figure 4c) tends to be somewhat higher in Type-1, -4, and -5 zircon (0.31-0.47) relative to the blue-luminescing material (0.22-0.40), but overall shows little variation (average Th/U = 0.34 ± 0.07).The abundances of Nb and Ta are comparable in sector-zoned crystals (Types 3 and 4 in Figure 4d) and reach their maximum values in brown growth zones, such as those shown in Figure 2a,b (up to 11 ppm Nb and 8 ppm Ta in Type 1, and up to 12 ppm of both in Type 2).The Nb/Ta ratio is strongly subchondritic and only slightly

Trace Element Composition and Variations
In this work, we studied zircon grains with all types of CL response.In total, 90 trace-element analyses were done in samples imaged using CL.Our data show that the Drybones Bay zircon macrocrysts exhibit appreciable variations in trace element abundances and ratios, particularly with regard to rare-earth elements (REE = La-Lu).All samples are relatively enriched in Ta (1.8-12.3ppm), but contain moderate levels of Hf (8200-10,900 ppm) at low levels of REE (<1.5 ppm Lu, <60 ppm Y).None of the seven types exhibit Eu anomalies.The average δEu ratio, which reflects deviation of the Eu concentration from a value interpolated from the chondrite-normalized Sm and Gd abundances, and is measured as Eu cn /[0.5 × (Sm cn + Gd cn )], is 1.0 ± 0.2.These compositional characteristics are consistent with kimberlitic zircon [32] and rule out crustal rocks or the Defeat granitoids as a potential source of these macrocrysts (cf.[33]).Another notable characteristic of the studied material is its enrichment in Sc (570-930 ppm), which is far in excess of what has been reported for felsic rocks [34].
The most significant compositional differences are observed between grains showing blue and yellow luminescence (Table 1), in accord with the observations made previously by Belousova et al. [30].Areas luminescing blue show the lowest levels of REE and Y (5-10 ppm and 6-11 ppm, respectively, in Type 3; 12-18 and 14-22 ppm, respectively, in Type 6; Figure 4a).In contrast, those showing yellow or yellowish-brown response contain the highest levels of these elements among the examined samples (23-53 ppm REE and 26-59 ppm Y in Type 1; 19-33 and 21-36 ppm, respectively, in Type 4).These differences are particularly striking in sector-zoned grains (Figure 3b), where REE+Y levels increase two-to fourfold across the sector boundary (Figure 4a).These variations are not accompanied by any consistent changes in Sc content.Grains showing yellow luminescence are also enriched in Th and U (3-9 and 8-20 ppm, respectively, in Type 1) relative to blue areas (≤2 ppm Th and ≤5 ppm U in Types 3 and 6; Figure 4b).The Th/U ratio (Figure 4c) tends to be somewhat higher in Type-1, -4, and -5 zircon (0.31-0.47) relative to the blue-luminescing material (0.22-0.40), but overall shows little variation (average Th/U = 0.34 ± 0.07).The abundances of Nb and Ta are comparable in sector-zoned crystals (Types 3 and 4 in Figure 4d) and reach their maximum values in brown growth zones, such as those shown in Figure 2a,b (up to 11 ppm Nb and 8 ppm Ta in Type 1, and up to 12 ppm of both in Type 2).The Nb/Ta ratio is strongly subchondritic and only slightly higher in the oscillatory-zoned core (1.1 ± 0.1) relative to the rest of the data (0.98 ± 0.08; Figure 4e).The highest Ti contents (21-34 ppm) are observed in the core (Figure 4c).Titanium levels are lower (10-25 ppm) and comparable in sector-zoned (Type-3 and -4) and blue-luminescing zircon.The latter is richer in Hf (9500-10,900 ppm) than the oscillatory core and yellow sectors (8200-9400 ppm), which is reflected in Zr/Hf variations depicted in Figure 4e.As can be expected, the Pb content reaches maximum values in U-and Th-rich zones (0.09-0.26 ppm), and ranges from levels below the LA-ICP-MS detection limit to 0.11 ppm in U-Th-poor blue-luminescing zircon (Figure 4f).Type 5 is compositionally transitional between Types 3 and 4.
higher in the oscillatory-zoned core (1.1 ± 0.1) relative to the rest of the data (0.98 ± 0.08; Figure 4e).The highest Ti contents (21-34 ppm) are observed in the core (Figure 4c).Titanium levels are lower (10-25 ppm) and comparable in sector-zoned (Type-3 and -4) and blue-luminescing zircon.The latter is richer in Hf (9500-10,900 ppm) than the oscillatory core and yellow sectors (8200-9400 ppm), which is reflected in Zr/Hf variations depicted in Figure 4e.As can be expected, the Pb content reaches maximum values in U-and Th-rich zones (0.09-0.26 ppm), and ranges from levels below the LA-ICP-MS detection limit to 0.11 ppm in U-Th-poor blue-luminescing zircon (Figure 4f).Type 5 is compositionally transitional between Types 3 and 4.   The REE budget of all six zircon varieties is very similar.Chondrite-normalized patterns show a steep slope rising towards heavy lanthanides (Nd cn /Yb cn = 0.036 ± 0.015), with a conspicuous positive Ce anomaly (Figure 5).The average δCe value, measured as Ce cn /[0.5 × (La cn + Pr cn )], is 8.2 ± 3.9, and none of the individual varieties give δCe below 3.5.Neither Eu nor Y anomalies are observed, and the average Y/Ho ratio is essentially chondritic (25 ± 2).The REE characteristics of zircon (i.e., the shape of chondrite-normalized profiles, Ce, Eu, and Y anomalies) are sensitive to its interaction with fluids and melts [33,35,36].Because the Y/Ho ratio shows the least variation across the dataset, we used it as a "common denominator" to gauge the extent of Pb loss (or gain).In Figure 4f, we compare variations in Pb to the measured U values normalized to the Y/Ho ratio in the same analysis.There is a clear positive correlation (R 2 = 0.79) between the two parameters, and no obvious outliers that would indicate variations in Pb content beyond those caused by heterogeneities in U distribution.The REE budget of all six zircon varieties is very similar.Chondrite-normalized patterns show a steep slope rising towards heavy lanthanides (Ndcn/Ybcn = 0.036 ± 0.015), with a conspicuous positive Ce anomaly (Figure 5).The average δCe value, measured as Cecn/[0.5 × (Lacn + Prcn)], is 8.2 ± 3.9, and none of the individual varieties give δCe below 3.5.Neither Eu nor Y anomalies are observed, and the average Y/Ho ratio is essentially chondritic (25 ± 2).The REE characteristics of zircon (i.e., the shape of chondrite-normalized profiles, Ce, Eu, and Y anomalies) are sensitive to its interaction with fluids and melts [33,35,36].Because the Y/Ho ratio shows the least variation across the dataset, we used it as a "common denominator" to gauge the extent of Pb loss (or gain).In Figure 4f, we compare variations in Pb to the measured U values normalized to the Y/Ho ratio in the same analysis.There is a clear positive correlation (R 2 = 0.79) between the two parameters, and no obvious outliers that would indicate variations in Pb content beyond those caused by heterogeneities in U distribution.

Thermometry
The concentration of Ti in the analyzed macrocrysts ranges between 10 and 34 ppm (Figure 4c).The oscillatory-zoned core, which fluoresces from yellow to yellowish-brown, shows the highest levels of this element, suggesting its crystallization at relatively higher temperature.Crystals showing sector zoning (Types 3 and 4) and brown zones in these crystals (Types 2 and 6) have overlapping Ti levels.The Ti-in-zircon thermometer of Watson et al. [37] gives apparent crystallization temperatures spanning from 742 °C (yellow sector) to 862 °C (core; Table 2).The average estimated temperatures are 820 ± 26 °C for the core and 781 ± 19 °C for the rest of the data, i.e., essentially within the estimated standard deviation of each other.

Thermometry
The concentration of Ti in the analyzed macrocrysts ranges between 10 and 34 ppm (Figure 4c).The oscillatory-zoned core, which fluoresces from yellow to yellowish-brown, shows the highest levels of this element, suggesting its crystallization at relatively higher temperature.Crystals showing sector zoning (Types 3 and 4) and brown zones in these crystals (Types 2 and 6) have overlapping Ti levels.The Ti-in-zircon thermometer of Watson et al. [37] gives apparent crystallization temperatures spanning from 742 • C (yellow sector) to 862 • C (core; Table 2).The average estimated temperatures are 820 ± 26 • C for the core and 781 ± 19 • C for the rest of the data, i.e., essentially within the estimated standard deviation of each other.
They crystallized in mantle rocks affected by metasomatism, possibly related to kimberlitic or alkali-basaltic magmatism, in their source [31,45]; III.
They are early cognate phenocrysts that recorded evolution of their parental magma [38,42].
Macrocrysts from the Drybones Bay pipe were previously investigated by Heaman et al. [4], who interpreted them to derive from a Mesoproterozoic, deep (~1.0 Ga, >150 km) mantle source, and to have remained open to Pb diffusion prior to their entrainment in kimberlite.The temperatures of zircon crystallization obtained in the present work cluster between 780 and 820 • C (Table 2).These values fall well within the temperature range (608-927 • C) determined for kimberlitic zircon from the Siberian, Kaapvaal, Amazonas, and São Francisco cratons by Page et al. [31].When projected to the average Slave geotherm (Figure 7), our data plot above the field of mantle garnets from the Drybones Bay pipe analyzed by Carbno and Canil [3].If the studied macrocrysts were indeed derived from a (metasomatized) mantle source, it had to be much shallower than 160 km (estimated from garnet-clinopyroxene equilibria by Carbno and Canil [3].Presently, there is no consensus on the origin of these enigmatic crystals.Three principally different interpretations have been proposed in the previous literature, i.e., that: I.
They crystallized in mantle rocks affected by metasomatism, possibly related to kimberlitic or alkali-basaltic magmatism, in their source [31,45]; III.
They are early cognate phenocrysts that recorded evolution of their parental magma [38,42].
Macrocrysts from the Drybones Bay pipe were previously investigated by Heaman et al. [4], who interpreted them to derive from a Mesoproterozoic, deep (~1.0 Ga, >150 km) mantle source, and to have remained open to Pb diffusion prior to their entrainment in kimberlite.The temperatures of zircon crystallization obtained in the present work cluster between 780 and 820 °C (Table 2).These values fall well within the temperature range (608-927 °C) determined for kimberlitic zircon from the Siberian, Kaapvaal, Amazonas, and São Francisco cratons by Page et al. [31].When projected to the average Slave geotherm (Figure 7), our data plot above the field of mantle garnets from the Drybones Bay pipe analyzed by Carbno and Canil [3].If the studied macrocrysts were indeed derived from a (metasomatized) mantle source, it had to be much shallower than 160 km (estimated from garnetclinopyroxene equilibria by Carbno and Canil [3].Notably, the Zr/Hf ratio of our samples is consistently superchondritic (≥43 and, on average, 52 ± 4), i.e., departs significantly from the mantle values (~36-37: [46,47]).Published data on peridotitehosted zircon are scarce, but suggest near-chondritic Zr/Hf values for this mineral [45,48,49].Although it is possible that the trace element budget of the hypothetical mantle source was modified by its interaction with a metasomatizing agent, how likely is the ascending kimberlite to have caused the documented Zr-Hf decoupling?Kimberlites have near-chondritic Zr/Hf ratios [50], which are also observed at Drybones Bay: the average Zr/Hf value, calculated for 21 whole-rock analyses of uncontaminated samples, is 37.5 ± 1.6 at 390 ppm Zr and 10.4 ppm Hf (authors' unpublished data).Assuming Rayleigh-style partitioning between a kimberlitic melt and zircon, Zr must partition into the latter ~1.4 times more effectively than Hf to cause the observed decoupling.Note that at very Notably, the Zr/Hf ratio of our samples is consistently superchondritic (≥43 and, on average, 52 ± 4), i.e., departs significantly from the mantle values (~36-37: [46,47]).Published data on peridotite-hosted zircon are scarce, but suggest near-chondritic Zr/Hf values for this mineral [45,48,49].Although it is possible that the trace element budget of the hypothetical mantle source was modified by its interaction with a metasomatizing agent, how likely is the ascending kimberlite to have caused the documented Zr-Hf decoupling?Kimberlites have near-chondritic Zr/Hf ratios [50], which are also observed at Drybones Bay: the average Zr/Hf value, calculated for 21 whole-rock analyses of uncontaminated samples, is 37.5 ± 1.6 at 390 ppm Zr and 10.4 ppm Hf (authors' unpublished data).

Figure 3 .
Figure 3. Cathodoluminescence images of zoned zircon from the Drybones Bay kimberlite (width of the field of view is 1.25 mm for both images): (a) oscillatory-zoned core (Type 1) rimmed by Type 3 zircon; (b) sector-zoned grain comprising Type 3 and Type 4 domains.

Figure 3 .
Figure 3. Cathodoluminescence images of zoned zircon from the Drybones Bay kimberlite (width of the field of view is 1.25 mm for both images): (a) oscillatory-zoned core (Type 1) rimmed by Type 3 zircon; (b) sector-zoned grain comprising Type 3 and Type 4 domains.

Figure 4 .
Figure 4. Variations in trace element composition among different types of zircon (distinguished on the basis of their cathodoluminescence (CL) characteristics).

Figure 4 .
Figure 4. Variations in trace element composition among different types of zircon (distinguished on the basis of their cathodoluminescence (CL) characteristics).

Figure 5 .
Figure 5. Chondrite-normalized patterns of the different types of zircon from the Drybones Bay kimberlite (lower limits of detection are plotted for La and Pr where these elements were not detectable by LA-ICP-MS).

Figure 5 .
Figure 5. Chondrite-normalized patterns of the different types of zircon from the Drybones Bay kimberlite (lower limits of detection are plotted for La and Pr where these elements were not detectable by LA-ICP-MS).

Figure 7 .
Figure 7.Estimated range of crystallization temperatures and depths for zircons (this work) and garnets [3] from Drybones Bay (DBB) projected on the average Slave Province geotherm.

Figure 7 .
Figure 7.Estimated range of crystallization temperatures and depths for zircons (this work) and garnets [3] from Drybones Bay (DBB) projected on the average Slave Province geotherm.

Table 1 .
Trace element composition of zircon from Drybones Bay kimberlite.

Table 2 .
Variations in Ti content and estimated crystallization temperatures of zircon from the Drybones Bay kimberlite.

Table 2 .
Variations in Ti content and estimated crystallization temperatures of zircon from the Drybones Bay kimberlite.

Table 3 .
U-Pb isotopic data for zircon from the Drybones Bay kimberlite.