Columbite (designated the columbite-group within) has the general formula AB2
where the “A” site is occupied mainly by Fe, Mn, and Mg and the “B” site by Nb and Ta. Three end-members, columbite-(Fe), columbite-(Mn), and columbite-(Mg), are known to occur in nature. Columbite-(Fe) is the most common, whereas columbite-(Mg) is rare [1
]. The columbite minerals occur in magmatic deposits of W, Sn, and rare earth elements (REE) [2
], alkaline and carbonatitic intrusions [3
], and pegmatites [4
]. Coltan, a dark ore of columbite-group minerals, is an important source for the strategic elements niobium and tantalum and, for this reason, ignited and funded armed conflicts in some African countries [5
Despite relatively low uranium concentrations, columbite-group minerals can be used for high resolution U-Pb geochronology because they have sufficient uranium concentrations and low common Pb content [2
]. Dating columbite-group minerals is an important geochronological tool for rocks that lack zircon, rocks in which zircon is metamict or inherited, and for fingerprinting of conflict minerals [9
]. Isotope dilution-thermal ionization mass spectrometry (ID-TIMS) was the first method used to date columbite [2
], followed LA-MC-ICP-MS [8
], and LA-ICP-MS [13
]. In these studies, monazite and zircon were used as external standards because of the absence of a columbite standard and developing a matrix matched standard is central to further the utility of columbite geochronology and the study of its trace-element geochemistry.
This study characterizes the columbite-group minerals from niobium-yttrium-fluorine (NYF-type) pegmatite occurrences in southern New York (Figure 1
We present major and trace element concentrations by electron microprobe (EMP) and LA-ICP-MS, respectively. In addition, U-Pb ages determined by LA-MC-ICP-MS are presented for columbite-(Fe) from two pegmatites from Westchester County, southern New York (Kinkel and Baylis) and columbite-(Mn) from Fort George and Harlem River Drive in New York City (Figure 1
). Co-existing high-U and Hf zircon (“cyrtolite”) from the Kinkel pegmatite was also analyzed for comparison purposes. This work further validates the use of columbite as a geochronometer and is a crucial step towards development of a matrix- matched columbite standard.
2. Geological Setting
NYF-type pegmatites are known in southern New York and New York City where they intrude rocks metamorphosed to amphibolite facies conditions such as the Precambrian Fordham gneiss, the Paleozoic Bedford gneiss (southern New York), and the Paleozoic Manhattan Schist (New York City) (Figure 1
The ca. 1170 Ma Fordham gneiss is the oldest recognized rock unit underlying New York City [20
]. It is composed of interlayered felsic and mafic units of meta-sedimentary and meta-igneous rocks [21
]. The Ordovician Bedford gneiss is a biotite-quartz-plagioclase gneiss with interlayered amphibolite [22
]. In places, andesine and microcline form “augen”, a characteristic feature of the gneiss [21
]. The pegmatites intruding the Bedford gneiss were mined for feldspar in several quarries from 1878 to 1949, when they closed [24
], and in 1962 and 1963 the sites were developed for residential purposes [19
Geological information on two of the main pegmatite bodies, Kinkel and Baylis, is found in several references [23
]. The 133-m long and 67-m wide pegmatite body mined at the Baylis Quarry displayed a border, wall, outer and inner intermediate, and core zones [19
]. No zonation was reported for the 200 m long and 100 m wide main Kinkel pegmatite body. The complex mineral composition of the Kinkel and Baylis pegmatites are reported in many sources [18
]. The exact location of columbite in the Baylis pegmatite is uncertain. The occurrence of columbite is described as at “the point between the feldspar and the country rock” [29
] (p. 46) or from the inner intermediate zone of the pegmatite body [19
The Manhattan Schist (Cambrian-Ordovician) is a coarse-grained schist and gneiss containing biotite, muscovite, quartz, and plagioclase with rare occurrences of sillimanite, kyanite, tourmaline, and almandine. The columbite-(Mn) crystals used for this study were from two small pegmatite bodies intruding the Manhattan Schist at the intersection of 185th Street with Harlem River Drive and at Fort George (Figure 1
). The mineral associations are simple, quartz, microcline, and beryl (185th Street with Harlem River Drive pegmatite) and quartz, microcline, almandine, and muscovite (Fort George pegmatite).
3. Analytical Methods
The columbite samples analyzed in this paper are from the New York State mineral collection and include columbite-(Fe) NYSM #25232 and #22361 (Kinkel), #BCB-COL (Baylis) and columbite-(Mn) NYSM #525.7 (185th Street and Harlem River Drive) and #525.8 (Fort George). Sample locations are shown on Figure 1
. The identity of each sample was confirmed by X-ray diffraction at St. Lawrence University (Figure 2
3.1. Major Elements
Centimeter- to millimeter-sized fragments of columbite-group minerals were mounted in epoxy, polished, and coated with carbon under vacuum and then analyzed using a CAMECA SX 100 electron microprobe at the Department of Earth and Environmental Sciences, Rensselaer Polytechnic Institute, Troy, NY. The analysis of major elements included Nb, Ta, Fe, Mn, and Ti. Mg and Ca were also included in the analysis, although high concentrations of these elements were not found (Table 1
To avoid potential problems that could influence U-Pb dating, such as the presence of tiny uraninite inclusions or other U-rich minerals, we screened both columbite-(Fe) and columbite-(Mn) in back scattered electron mode and produced X-ray maps (Figure 3
) for U, Nb, Ta, Fe, Ti and Mn.
No inclusions were found and none of the columbites displayed zoning. Only sample NYSM #525.7 displays compositional variation in Nb, Ta, Fe, and Mn (Figure 3
), but is homogeneous in Ti and U (not shown).
All elements were analyzed at 15 keV accelerating voltage and a 60 nA beam current, using integral PHA mode. Due to the high concentration of niobium (~50 wt. %), the Nb 2nd order (Lα2) line was used to suppress the high-count rate (measured on the TAP crystal) and reduce dead time correction. The remaining elements measured were Ta Lα1 (on LLIF), Fe Kα and Mn Kα (on LLIF), Ti Kα and Ca Kα (on LPET), Mg Kα (on TAP), and O Kα (on PC1). All elements were background corrected using traditional two-point off-peak background interpolation. The oxygen Kα peak was free of significant high-order interferences from major elements. Background selection on the high-energy side of the oxygen peak avoided potential interferences such as Mn Ll and Ln (1st order), Ta Mα (3rd order), and Nb Lα (4th order). An interference correction was applied for the tail overlap of Mn Kβ with Fe Kα (0.39 cps/nA). Standard reference materials included niobium metal for Nb (C. M. Taylor), tantalum metal for Ta (C. M. Taylor), synthetic fayalite for Fe (RPI), tephroite for Mn (Smithsonian), rutile for Ti (Harvard collection), diopside for Ca (Geophysical Lab) synthetic forsterite for Mg (RPI) and quartz for O (Smithsonian). The niobium metal was found to contain a significant oxidation layer resulting in apparent Nb excess in unknown materials. Therefore, columbite from the Kinkel Quarry (NYSM #22361) was chosen as an internal reference material for Nb based on the low concentration of trace elements. Ideal Nb concentration was calculated by difference from Ta (assuming these were the only two elements in the B-site).
3.2. Trace Elements
Trace elements in columbite were measured by LA-ICP-MS on a Photon Machines Analyte193 G1 short-pulse eximer laser coupled to Varian 820 quadrupole inductively coupled plasma mass spectrometer (Table 2
and Table 3
) at RPI.
All trace elements were calibrated using NIST standard glasses (#610, primary; and #612, secondary). Laser parameters included 360 shots at 6 Hz repetition rate per ablation, at 58% laser power attenuation, 6.5 J/cm2 fluence using 20-micron square spot. Mass spectrometer acquisitions were 140 s windows each, including 60 s ablation, 20 s washout, and 60 s background acquisition. Masses analyzed by LA-ICP-MS (with millisecond dwell times in parentheses) included Ca-43 (30), Sc-45 (10), Ti-49 (1), Mn-55 (0.2), Fe-57 (1), Y-89 (10), Zr-90 (10), Sn-118 (10), La-139 (30), Ce-140 (30), Pr-141 (30), Nd-146 (30), Sm-149 (20), Eu-153 (20), Gd-157 (20), Tb-159 (20), Dy-163 (10), Ho-165 (10), Er-166 (10), Tm-169 (10), Yb-172 (10), Lu-175 (10), Hf-178 (10), W-182 (10), Pb-206 (10), Th-232 (20), and U-238 (30). The total quadrupole scan time was 465 ms per cycle, resulting in >120 cycles within each 60 s ablation event. Each columbite sample was ablated five times in distinct locations and standards were reanalyzed every five ablations. Data were processed using the Iolite software’s trace element data reduction scheme using Ti as internal standard (as measured by EPMA).
U-Pb columbite geochronology was carried out at the Arizona Laserchron Center, University of Arizona. Analyses by LA-MC-ICP-MS were conducted as described for zircon [32
]. Ablation of columbite was done with a Photon Machines Analyte G2 DUV193 Excimer laser using spot diameters of 30 microns. The ablated material was carried with helium gas into the plasma source of a Nu Instruments HR ICPMS equipped with a flight tube of sufficient width that U, Th, and Pb isotopes can be measured simultaneously. All measurements were made in static mode, using Faraday detectors for 238
Th, and 208-206
Pb, and discrete dynode ion-counters for 204
Pb, and 202
Hg. Ion yields are typically ~1 mv per ppm. Each analysis consists of one 15 s integration on peaks with the laser off (for backgrounds), fifteen 1-s integrations with the laser firing, and a 30 s delay to purge the previous sample and prepare for the next analysis. The ablation pit is ~15 microns in depth.
For each analysis, the errors in determining 206
U and 206
Pb result in a measurement error of ~1% (at 2σ level) in the 206
U age. The errors in measurement of 206
Pb and 206
Pb also result in ~1% (2σ) uncertainty in age for grains that are >1.0 Ga, but they are substantially larger for younger grains due to the low intensity of the 207
Pb signal. For most analyses, the crossover in precision of 206
U and 206
Pb ages occurs at ca. 1.0 Ga. The 206
U ages are reported here. Common Pb correction is accomplished by using the measured 204
Pb and assuming an initial Pb composition [33
], with uncertainties of 1.0 for 206
Pb and 0.3 for 207
Pb. The measurement of 204
Pb is unaffected by the presence of 204
Hg because backgrounds are measured on peaks (thereby subtracting any background 204
Hg and 204
Pb), and because very little Hg is present in the argon gas. Interelement fractionation of Pb/U is generally ~20%, whereas apparent fractionation of Pb isotopes is generally <2%. In-run analysis of fragments of a large zircon crystal (generally every fifth measurement through the course of analyses) with known age of 564 ± 4 Ma (2σ error) is used to correct for this fractionation. The uncertainty resulting from the calibration correction is generally ~1% (2σ) for both 206
Pb and 206
U ages. The analytical data are summarized in Table 4
and reported in full in the Table S1A–F
. Uncertainties shown in these tables are at the 1σ level and include only measurement errors. The reported ages were calculated using Isoplot (Figure 4
and Figure 5
Concordia ages are given for comparison on Tera–Wasserberg diagrams (Figure 5
) and are the same as weighted averages within error limits.
Final weighted average age diagrams (Figure 4
) show two sigma uncertainties. The smaller uncertainty (labeled Mean) is based on the scatter and precision of the set of 206
U or 206
Pb ages, weighted according to their measurement errors (shown at 2σ). The larger uncertainty (final age), which is the reported uncertainty of the age, is determined as the quadratic sum of the weighted mean error plus the total systematic error for the set of analyses. The systematic error, which includes contributions from the standard calibration, age of the calibration standard, composition of common Pb, and U decay constants, is generally ~1–2% (2σ).
Analyses were conducted on large columbite grains or grain fragments mounted in epoxy plugs and polished to yield a significant cross-section through the grain. As noted below, the grains were analyzed in back scattered electron mode and x-ray mapping to find and avoid inclusions and zoning, if present (Figure 3
). The sampling strategy involved analyzing transects across each grain from edge to edge, passing through the center, to check for inconsistencies that might arise from zoning, cores, or Pb-loss from rims. Cracks, fractures, inclusions, or other heterogeneities were avoided. Except where noted below, the grains gave concordant analyses which overlapped within uncertainty regardless of where the ablation pit was located.
Columbite-group minerals have the general formula AB2
where Fe, Mn, Mg, Nb, and Ta are the main constituents and the significant range of substitutions between these components contributes to the occurrence of the three species of the group mentioned above. The substitutions of Mn for Fe and Ta for Nb are used to define the fractionation trend in the rare-element granitic pegmatites. High Mn/(Mn+Fe) and Ta/(Ta+Nb) ratios characterize the more fractionated pegmatites [40
]. The calculated ratios (Table 1
) show that the pegmatites from New York City are somewhat more fractionated compared with those from Bedford.
The trace elements Sc, Zr, Hf, and REEs are incorporated into the columbite structure in different amounts by coupled substitutions. Scandium is incorporated through the coupled euxenite-type substitution, Sc3+
. Scandium resides in the A-site where it substitutes preferentially for Fe because of their closer charge and ionic radius. The other two components of the coupled substitution, Sn and Ti, partition into the B-site [41
]. Zr and Hf occur in significant concentrations in both New York City and Bedford pegmatites and Zr/Hf ratio is medium (9.73 in NYSM #22361) to low (6.95 in NYSM #525.7) which is common for less evolved NYF pegmatites lacking zircon. The enrichment of Zr and Hf in columbite is typical for the highly fractionated pegmatites of the lithium-cesium-tantalum (LCT) family [35
]. The metamict zircon, “cyrtolite”, occurs in the Kinkel pegmatite and displays 54,100 ppm Hf. The incorporation of relatively significant amounts of Zr and Hf in the columbite-group minerals we analyzed could be the result of: (a) little appreciable difference in the partitioning of these elements in zircon or columbite-minerals [35
]; or, possibly, (b) columbite and zircon (“cyrtolite”) formed from different melt batches. Unfortunately, a thorough description of the occurrence of these two minerals is missing from literature and New York State Museum database and remains unresolved.
The REEs substitute in six-fold coordination and are incorporated into the columbite-group minerals by two types of coupled substitutions: (a) euxenite-type A2+
; and (b) samarskite-type 2A2+
= 3(La + Fe)3+
]. The REE/chondrite diagrams (Figure 6
) display three types of patterns for columbite-(Fe) and columbite-(Mn).
The first type represented by the REE pattern for columbite-(Fe) shows an ascending plot line for the LREEN
from La to Sm, strong negative Eu anomaly, a “dome” of the middle REE (MREEN
) and a descending trend for the HREEN
A–C). The MREEN
are higher than the HREEN
. The second type of REE pattern is shown by columbite-(Mn), NYSM 525.7 (Figure 6
E). It is like that of “cyrtolite” (Figure 6
D) displaying an ascending trend for the LRREN
, negative Eu anomaly and MREEN
slightly dominating over the HREEN
. NYSM 525.8 shows an ascending trend from La to Sm, negative Eu anomaly, and another ascending trend for the HREEN
F) that dominate over the MREEN
. These different REE patterns probably describe the degree of evolution of the pegmatites [42
], thought to be relatively low due to the small concentrations of LREEs. However, LREEs may also be preferentially excluded from lattice sites in the columbite-group minerals due to their greater size. The “cyrtolite”/chondrite diagram (Figure 6
D) shows the LREEN
at low concentrations with an almost flat plateau between Ce and Nd. Other characteristics include positive Sm anomaly, negative Eu anomaly suggesting a significant plagioclase crystallization in the parent melt, and a dome for the group Gd to Ho. The Ho- to Lu group is on a flat plateau.
As shown in Table 4
, the U-Pb columbite geochronology displays a narrow range of weighted mean ages: 372.2 ± 8.2 Ma for the Baylis Quarry, 371.3 ± 7.3 and 383.4 ± 8.9 Ma for the Kinkel Quarry, and 383 ± 15 and 372 ± 11 Ma for New York City pegmatites. In addition, “cyrtolite” yielded a U-Pb weighted mean age of 376.9 ± 4.3 Ma. Although two samples of columbite and one of cyrtolite from the Kinkel Quarry differ in age by 12 million years (371.3–383.4 Ma), the data are not precise enough to determine if this age range represents true analytical error or two pegmatite intrusion events. This range of ages indicates that emplacement of the pegmatites occurred during the Neoacadian orogeny (395 to 360 Ma) [43
]. The age of the southern New York pegmatites is consistent with the older group of LCT-type pegmatites along the Appalachians [44
]. The LCT pegmatites display two groups of ages: (a) 371 Ma; and (b) 274–264 Ma [44
]. We consider the pegmatites to be related to the Neoacadian magmatism known in the northern Appalachians. The New York pegmatites appear to represent the southernmost expression of the Neoacadian orogeny in New England.
Several characteristics suggest some columbite crystals may have potentially as a U-Pb isotopic standard for oxide mineral analysis. Columbite-group minerals have a simple formula with limited solid solution and often form large chemically and isotopically homogeneous crystals. In terms of U-Pb geochronology, columbite excludes common-Pb during crystallization (high 206
Pb ratios) and has sufficient uranium concentrations to accumulate radiogenic Pb over geologic time, even within the late Paleozoic samples analyzed here. In fact, the variation in U-content determined by LA_MC-ICP-MS in some samples is quite limited (Table 4
). For example, twelve analyses of sample #525.8 yields a uranium concentration and standard deviation of 500.2 ± 4.5 ppm. Further work defining the extent of chemical and isotopic variation, metamictization, and blocking temperatures will determine the limits of columbite’s use in geochronological and isotopic studies.