ho

la C

GeochemistryMineralogyLine-scan imagingRock coresMulti-sensor core logger

onserspoteeralal alteration, or provide chemo-stratigraphic constraints, for example. But mea-

ay toand on

Ore Geology Reviews 53 (2013) 93111

Contents lists available at SciVerse ScienceDirect

Ore Geology

l seassigned to core drilling. Exploration companies have the cores visuallylogged by a geologist, who also decides which intervals to sample forassays of the sought-after metals (Ag, Au, Cu, Ni, Pb, Zn, etc.). Corerecovery parameters such as the rock quality designation (RQD) arealso routinely recorded. In addition, some exploration companies mea-sure the magnetic susceptibility of the rock cores with handhelddevices, and/or send some samples out to commercial laboratories forwhole-rock geochemistry.

However, a lot more information could be extracted out of these

database for a certain region given the high cost of data acquisition,the time involved, and the destructive nature of several conventionalanalytical techniques.

A newmobile laboratory at Institut national de la recherche scientique(INRS), Quebec, Canada, contains a high-resolution, semi-automatedcore logger that measures several parameters near-simultaneously onrock cores, in a non-destructive manner, at core repositories. Currently,the logger can measure the density and magnetic susceptibility ofrocks, quantify several chemical elements by energy-dispersive X-rayvery costly drill cores, such as the physicaltheir mineralogy, and their geochemistry, at(small measurement spacing). Quantifyinghelp reaching goals such as (1) better plannphysical surveys; (2) modeling the geology

Corresponding author. Tel.: +1 418 654 3773.E-mail address: rossps@ete.inrs.ca (P.-S. Ross).

0169-1368/$ see front matter 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.oregeorev.2013.01.002e of the favorite tools ofstages of an exploratione part of the budget is

ne-grained and difcult to sort apart. Various devices and laboratorytechniques are available for measuring each property at a time, butthis approach makes it unrealistic to build a large multi-parametermineral exploration. Typically, after the initialprogram for base or precious metals, a largMineral exploration

1. Introduction

Diamond drill holes are the only waccess to rocks in the third dimension,due to cost or time considerations and, for some parameters, it would destroy the core (e.g. geochemistry).In this paper we describe a multi-sensor core logger and its use on rock cores from exploration diamond drillholes. This semi-automated system can measure near-simultaneously, non-destructively and at high spatialresolution, the following parameters: (1) volumetric magnetic susceptibility; (2) density using gamma-rayattenuation; (3) several chemical elements through energy-dispersive X-ray uorescence spectrometry;and (4) visible/near infrared spectrometry, which allows numerous minerals to be detected and character-ized. The logger also acquires a continuous image of the core using a line-scan camera, which allows theuser to compare other properties with the visual aspect of the core and creates a complete virtual archive.The aim of this mostly methodological paper is to describe the logger as a whole and then each instrumentor sensor separately, outlining the numerous tests that have been performed to assess and improve dataquality. We also present preliminary results from the Matagami mining camp of Canada, a base metal district.

2013 Elsevier B.V. All rights reserved.

gain direct, high-quality

rocks in three dimensions; (3) documenting hydrothermal alteration;and (4) improving the visual logs of the cores where alteration makesthe protoliths difcult to determine, and/or where the rocks areKeywords:Physical propertiessuring all the parameters one by one at high spatial resolution by traditional methods would be impractical

3D, characterize hydrothermA multi-sensor logger for rock cores: MetMatagami mining camp, Canada

P.-S. Ross , A. Bourke, B. FresiaInstitut national de la recherche scientique, Centre Eau Terre Environnement, 490, rue de

a b s t r a c ta r t i c l e i n f o

Article history:Received 12 October 2012Received in revised form 13 December 2012Accepted 14 January 2013Available online 23 January 2013

Diamond drilling typically cgenerates thousands of metrently utilized to their fullties, geochemistry and min

j ourna l homepage: www.eproperties of the rocks,a high spatial resolutionthese parameters coulding or interpreting geo-or physical properties of

rights reserved.dology and preliminary results from the

ouronne, Qubec (QC), Canada G1K 9A9

titutes a major part of costs in advanced mineral exploration programs. Thisof rock cores during major exploration campaigns, but the cores are not cur-ntial. They could supply three-dimensional information on physical proper-ogy; such data could be used to model the geology or physical properties in

Reviews

v ie r .com/ locate /oregeorevuorescence (XRF), and characterize mineralogical assemblages by visi-ble light andnear infrared (Vis/NIR) spectrometry (Fig. 1). The logger canalso acquire a high-quality continuous image of the core using a line-scancamera, creating a virtual archive of the drill hole. In thismostlymethod-ological paper, we present the characteristics of the logger as a wholeand then of each instrument or sensor separately, outlining the numer-ous tests that have been performed to assess and improve data quality.We also present preliminary results from the Matagami mining camp

94 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111aof Canada, an Archean base metal district, and compare the INRS loggerwith other systems.

2. General description of the logger and preliminary tests

The multi-sensor core logger (MSCL) at INRS has been designed touse primarily on exploration diamond drill cores. It is contained in amobile laboratory housed in a modied cargo trailer, which measures6.1 m by 2.6 m at the base (Fig. 1a). Windows, a heater and anair-conditioning system allow the users to maintain a nearly constanttemperature in the laboratory, which is important for density mea-surements as will be discussed below. The logger takes a few hours

Vis/NIR

Track

Gammsource

b

Fig. 1. (a) Photo of the INRS mobile laboratory. (b) Three dimensional representation of thfrom this drawing are the electronic racks and the MSCL computer.at most to setup at destination, since most of the MSCL's componentsare permanently attached to the trailer.

The MSCL at INRS was designed, built and installed by GeotekLtd. of Daventry, England, using sensors and scientic instrumentsmanufactured by this company and other suppliers. Prior to this,Geotek had produced over 100 loggers for other clients, mostly forthe analysis of water-saturated sediment cores in plastic liners, withmagnetic susceptibility, density and seismic (P-wave) velocity beingthe most common parameters measured (e.g., Rothwell and Rack,2006;Weber et al., 1997). Before INRS's loggerwas built, fewGeotek log-gers had been destined for rock cores (e.g., Vatandoost et al., 2008), onlyone had had an XRF analyzer integrated, and none contained a Vis/NIR

a

Pusher

e MSCL inside the laboratory, modied from an image supplied by Geotek Ltd. Missing

95P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111spectrometer. A comparison between the INRS logger and other systemsis provided in Section 9.

2.1. Principle of operation and logging velocity

OurMSCL contains fourmain elements: (1) a series of sensors and in-struments; (2) the horizontal track onwhich the core is pushed in a coreboat; (3) an electronic interface; and (4) a computer which controls theequipment and records the parameters and images (Geotek Ltd., 2010;Fig. 1b). Typical wooden storage boxes in Canada contain four (BQsize) or three (NQ size) core sections, each measuring about 1.5 m inlength. The core from each box is transferred into custom-madenon-magnetic berglass core boats slightly longer than 1.5 m, and theboats travel in queue on the track.

For line-scan imaging purposes, the movement of the core is contin-uous, whereas for other parameters, the core stopswithin or under eachdevice during the time required tomake ameasurement (~1 s for mag-netic susceptibility, 110 s for Vis/NIR spectrometry, 10 s for gammadensity, and 6090 s for the XRF analyzer). The Vis/NIR contact probeand one XRF analyzer are mounted on an arm that moves up anddown so that the instruments come in rm contactwith the core duringthe measurement, but there is no friction while the core is movinghorizontally.

The pusher is moved by an electrical motor and the precision on thepositioning of the core in the MSCL is 0.5 mm, so the parameters are allmeasured at the same spot on the core (themeasurements are automat-ically co-located). The precision on core positioning within the logger ismuch better than the error on the absolute down-hole depth withinthe core boxes, which is at best about 10 cm.

The current routine logging speed is about 60 m of exploration drillcore per eight hour work shift, including gamma density calibrations(described below). The mentioned logging velocity takes into accounta spatial resolution (measurement spacing) of 2030 cm; runningthe core in the MSCL once for imaging and a second time for density,Vis/NIR spectrometry and magnetic susceptibility; and taking XRF mea-surements separately with two portable XRF devices installed on labora-tory stands, while other core sections are on the MSCL. The logger istechnically capable of acquiring the line-scan images and all otherparameters in a single pass, but proper imaging of rock cores requireswetting the rock surface, whereas Vis/NIR spectrometry requires a fullydry surface as explained in Section 2.4 (the cores have time to dry in be-tween the two passes). The reason for currently operating the XRF de-vices separately from the logger is simply to save time, since the XRFmeasurements are the longest by far; also only one XRF analyzer hasbeen integrated to the MSCL.

2.2. Core preparation

The proper preparation of rock cores prior to logging is critical to ob-tain good data. While still in their wooden boxes, the cores are cleanedvigorously with a brush, to get rid of the dust and small debris from thesurface. Transferring the core into the boats results in a 180 axial rota-tion, exposing the other side for cleaning as well. The cores are imagedwhilewet,with thewater appliedwith a paint brush; this gets rid of anyremaining dust on the upper core surfaces. Water used for the brushingis changed regularly. The cores are left to dry at least ten minutes afterimaging, and then the other parameters can be measured during asecond pass through the MSCL. The reason for this thorough cleaningof the cores is that both Vis/NIR spectrometry and portable XRF are es-sentially surface measurements, having a low depth of penetration inthe rock. The dramatic impact of added water on these two techniques(see Section 2.4 below) justies the long drying period after theimagery.

Before the imaging is performed, pieces of core are brought closelytogether to close the gaps, and replaced in their inferred original po-

sition relative to each other. The beginning of each section is alignedush left with the core boat (core boats travel from right to left in theMSCL), and the down-hole depth in meters for the beginning of thatsection is marked on the core. This depth appears on the line-scan im-ages for reference. A laser beam is interrupted when the left end ofthe core section arrives just before the camera, and all instrumentsand sensors are located along the track relative to this position laser(Fig. 1b).

2.3. Core thickness measurements

Core thickness (diameter) inuences the density calculations (andto a lesser extent, the magnetic susceptibility calculations) so it mustbe measured, not assumed. To evaluate how many thickness determi-nations are necessary to get a representative prole of thickness versusdepth, we measured the core thickness of a 686 m-deep hole, LEM-18,from the Chibougamau (Quebec) area, with a vernier caliper displayinghundredths of mm. Thickness measurements were done at each MSCLanalysis point (n=2153 for whole cores), and also at a different spot,once per core box, i.e. about every 4.5 m (n=158). We obtainedmean core thicknesses of 4.744 and 4.743 cm, respectively, for thetwo series, compared to the nominal NQ thickness of 4.76 cm. Standarddeviations of 0.014 and 0.015 cm were obtained, respectively (relativestandard deviations of 0.3%). We calculated the uncorrected gammadensity for each depth using the two series (thickness measuredevery point vs. once per box), and the average value of the absolute dif-ference between the two densities at each depth was 0.003 g/cm3. Foronly 16 points out of 2153 the uncorrected density difference wasmore than 0.03 g/cm3. For reference, the precision of the method is ap-proximately 0.01 g/cm3, but gamma density proles are quite wigglydue to mineralogical heterogeneities. The uncorrected density vs.depth proles for both series were so similar that the practice to mea-sure the core thickness only once per box was adopted, for future log-ging. For the Matagami data presented here, however, we onlymeasured core thickness every 25 m.

2.4. Inuence of added water

The presence of externalwater on the rock surface is detrimental forboth portable XRF methods and Vis/NIR spectrometry. Ge et al. (2005)reports a deterioration of precision, accuracy and detection limits onXRF measurements in the presence of water. Known effects on infraredspectrometry of rocks and soils include the addition of absorptionpeaks, attenuation of legitimate peaks, and wavelength offsets onsome peaks (Clark, 1981; Gray, 1997; Lobell and Asner, 2002). Such ef-fects can clearly deteriorate the mineralogical interpretation of thespectra.

We veried the necessary drying time for Vis/NIR spectrometry bysaturating the surface of a rock core with water at room temperatureand acquiring a 1 s spectrum every 7.5 s on the same spot until thecore appeared dry and the spectra stabilized for several minutes. Thechosen sample was a non-porous basaltic lava from Matagami. Fig. 2illustrates that for this representative sample, a drying time of approxi-mately ve minutes was necessary for the spectra to stabilize; at thisstage the core also appeared fully dried visually. From this experimentand other observations, a conservative drying time of ten minutes waschosen.

No drying time tests were performed for XRF measurementsbut we assume that the effect of added water is worse for Vis/NIRspectrometry, therefore ten minutes of drying is sufcient for XRFmeasurements.

2.5. Inuence of core temperature

Vatandoost et al. (2008) used a MSCL for physical property mea-surements on rocks, and suggested that cores be brought in the labo-

ratory two to three hours before being analyzed, so that they are

96 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 931110

5

10

15

20

25

300 500 700 900 1100 1300 1500 1700 1900 2100 2300 2500

04:01

03:31

00:00

03:31

04:01

03:17logged at a constant temperature. Since space is restricted in the INRSmobile laboratory, we carried out simple tests to verify if this precau-tion is indeed necessary.

For density and magnetic susceptibility, the range of investigatedcore temperatures was 15 to 35 C, and we conclude that over thisinterval, these two parameters are not inuenced in a geologically sig-nicant manner. Specically, a whole core box, i.e. about 4.5 m of NQcore, of weakly magnetic (1.01.3103 SI), homogeneous-lookingMatagami basalt was tested at 22 different depths at 15 C, 25 C and35 C. Ambient temperature in the laboratory remained approximatelyconstant during the test. Nomajor differences appear on the proles forthe different core temperatures (Fig. 3). The average of the uncorrectedgamma densities is the same regardless of core temperature (Table 1).The averagemagnetic susceptibility of the basalt was found to be slight-ly temperature-dependant. For example, the average susceptibility at15 C is 1.6% higher than the average at 25 C, whereas the average at35 C is 2.1% lower (Table 1). However, t tests reveal that these differ-ences are not statistically signicant at the 95% condence level. Theequivalent non-parametric test (Wilcoxon) suggests a statistically sig-nicant difference.

A box of highly magnetic gabbro (~70150103 SI) was alsomeasured at many points for two contrasting temperatures (23 Cand 37 C) to further investigate the effect of core temperature onmagnetic susceptibility. Details of this experiment are reported byRoss et al. (2011a). The difference in average magnetic susceptibilitybetween the two measurement series was only 0.2%, but this differ-ence is statistically signicant at the 95% condence level using boththe t test and the Wilcoxon test. However, as with the basalt, such

Fig. 2. Inuence of addedwater on the surface of a core on Vis/NIR spectrometry. The testwas performed on Matagami diamond drill hole BRC-08-72, at down-hole depth 82.1 m.At the beginning of the drying experiment (t=0), the surface of the core was saturatedwith water. Notice the important water absorption peak near 1920 nm, which is absentfrom the dry core spectrum at the end (7 min 18 s). Many other differences are obvious,such as the average reectance, and the height of the peak at 600 nm. For this sample, veminutes was needed for the rock to appear fully dry and for the spectra to stabilize. Noisein the spectra is duemostly to shortmeasurement times (1 s) and the presence of uores-cent lighting in the laboratory during the test.very small differences are not geologically signicant, since the mag-netic susceptibility varies over several orders of magnitude betweendifferent lithologies.

We also tested the inuence of core temperature on VIS/NIR spectra.A sample containing chlorite and epidote was measured twice at42.3 C and twice at 23.0 C, without moving the sample. The spectraall look very similar (not shown). The wavelengths of the twomost im-portant absorption peakswere determined: FeOH andMgOH. The FeOHpeak was at 2257 and 2258 nm (high temperature) and then at 2256and 2258 nm (low temperature). The MgOH peak was at 2347 and2341 nm (high temperature) and then at 2344 and 2343 nm (low tem-perature). In other words, core temperature does not inuence thewavelength position of major absorption features with the equipmentused and the temperature range tested.

Consequently, no formal procedure of bringing the cores inside thelaboratory a certain time in advance was adopted, although in typicallogging sessions, most core boxes actually have time to thermallyequilibrate with air in the laboratory before they are logged.

2.6. Choice of measurement points

Choosing representative measurement points on the core is an im-portant decision inuencing data quality. The logger control softwaresupplied by Geotek Ltd. offers either a user-dened constant distancebetweenmeasurements (e.g., 20.0 cm) or a variable samplingmodein which the user species the relative depth (from the section top) ofeach measurement point.

The constant interval mode was investigated rst to see if thistime-saving strategy would work. Time would have been saved both(i) by not choosing, noting down and entering individual measurementpoints in the software prior to logging, and (ii) by selecting a samplinginterval equal to the distance between the two slowest instruments/sensors on the MSCL in order for these instruments to operate simulta-neously (see Vatandoost et al., 2008). However, the Matagami explora-tion core is fractured (in the best conditions one break per meter,typically much more), and measurements must not be made in ornear gaps, for obvious reasons (density would be articially lower,etc.). TheMSCL software provides an option to skip somemeasurementpoints when fractures are detected automatically by the position laser,but experience showed that not all fractures were detected, and thatlow-angle fractures were problematic. Therefore we quickly switchedto the variable sampling mode.

In variable samplingmode, the user selects themeasurement pointsone by one on the core, using a predened interval as a general guide(e.g., approximately every 20 cm), and a tapemeasure to determine rel-ative depth. This takes time, but allows the geoscientist not only toavoid gaps but also to exercise judgment on which point is representa-tive of a given lithological interval. For example, isolated veins or alteredspots can be avoided, and so can anomalously amygdaloidal zones in alava ow. However, if the whole section is altered, amygdaloidal orfull of veinlets, then such features are representative andmeasurementspoints are selected on cores that display the features.

The following sections describe the sensors and instrumentsinstalled on the INRS's MSCL and the tests performed for each param-eter, when applicable.

3. Line-scan imaging

Line-scan imaging creates images of amuch better quality that couldbe achieved using a normal digital camera. The distortion-free (length-wise) and largely reection-free images can be used to rene the visuallog of a drill hole, or to compare the othermeasured properties with theaspect of the core. This allows many anomalies in the physical or geo-chemical properties to be explained (e.g., by the presence of veinlets

or alteration).

00.5

1

1.5

2.92 2.94 2.96 2.98 3.00 3.02 3.04

35oC

15oC

97P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 931112In addition, because the illumination is fully controlled, the imagescontain quantitative redgreenblue (RGB) information about the corethat could be used for image analysis. The average color of a segmentof core will be a function of protolith, alteration and mineralization.

2.5

3

3.5

4

4.5

5

25oC

Fig. 3. Results of the test examining the physical properties of a box of core at three core temand standard deviations. Horizontal error bars represent the estimated precision of the me

Table 1Summary of a series of uncorrected gamma density and magnetic susceptibility mea-surements with the MSCL at three different temperatures in Matagami diamond drillhole BRC-08-72, box #5.

Core temperatureC

Density (uncorr)AverageaS.D.b

g/cm3

Magnetic susceptibility ()AverageaS.D.b

105 SI

15 3.000.03 1221025 3.000.02 120835 3.000.02 1179

a Average of a series of measurements at the following relative down-hole depths inmeters (measured from the top of the box): 0.14; 0.36; 0.54; 0.74; 0.94; 1.12; 1.36;1.65; 1.87; 1.99; 2.25; 2.45; 2.83; 2.99; 3.16; 3.34; 3.54; 3.74; 3.94; 4.14; 4.34; and 4.52.b S.D.=standard deviation of this series of measurements.25oC3.06 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5The Vis/NIR spectrometer also produces RGB values but only one suchvalue per spectrometer measurement (e.g., every 30 cm), so theline-scan images aremuch richer. Howeverwe have not pursued this re-search path yet.

3.1. Principle of measurement and practical details

The GeoScan color line-scan camera, designed by Geotek Ltd., is athree charge-coupled device (CCD) unit using three 2048 pixel CCD ar-rays (Geotek Ltd., 2010). Adjustable lighting is provided by uorescenttubes that illuminate the core evenly from both sides of the image line.In practice, an optimal illumination setting is chosen for each diamonddrill hole (or each district). When images are being acquired, the coremoves continuously at low speed through the MSCL.

The scanning produces a continuous image of each 1.5 m-long sec-tion, which we concatenate for each core box (~4.5 to 6 m of coredepending on core size). Longer portions of core can be amalgamatedbut the image le sizes become unduly large. The absolute down-holedepth in meters is automatically added to the margin of the image,

35oC

15oC

peratures (Matagami diamond drill hole BRC-08-72, box #5). See Table 1 for averagesasurements. See text for discussion.

allowing the user to visualize exactly where each measurement wastaken (Fig. 4).

Three resolutions are available: 100 pixels per cm (254 pixels perinch), 200 pixels/cm and 400 pixels/cm. During normal logging in theINRS laboratory, the 100 pixels per cm resolution setting is used, sincethis is sufcient to obtain a good image of the core (e.g., Fig. 4); the scan-ning rate is then about two minutes per meter of core.

4. Magnetic susceptibility

Volumetric magnetic susceptibility the parameter that should beused in geophysical modeling is a dimensionless constant that cor-responds to the degree of magnetization of a rock in response to amagnetic eld (Hunt et al., 1995). This parameter is mostly controlledby the abundance of minerals such as magnetite or pyrrhotite in the

a b c

rillte dn 54

98 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111Fig. 4. Examples of routine 100 pixels per cm line-scan images of the core fromMatagami dabundant epidote and later quartz or calcite veinlets, between 448.80 and 449.00 m absoluto volcanic or hydrothermal processes, as well as someminor disseminated suldes, betwee

from the main lens between 572.60 and 572.80 m (half-core).hole BRC-08-72 obtainedwith theMSCL camera. (a) Gabbroic intrusion locally invaded byown-hole depth (whole core). (b) Chloritized Bracemac Rhyolite showing brecciation due2.54 and 542.74 m (whole core). (c)Massive suldes (sphalerite, pyrite and chalcopyrite)

rock. Potential eld geophysical surveys can be better interpretedwhen rock physical properties (density and magnetic susceptibility)are known for each major lithological unit in the region of interest,

99P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111rather than pulled out of a textbook. Magnetic susceptibility has nu-merous other uses; for example, it can be included in multivariatestatistical analyses to separate different lithologies.

4.1. MSCL measurement method

The MSCL measures magnetic susceptibility with the MS2C CoreLogging Sensor from Bartington Instruments Ltd. (Oxford, England),which is now connected to the MS3 meter from the same rm. TheMS2C is a loop-type sensor through which the core passes in itsnon-magnetic core boat (Fig. 1b). Initially the MS2 meter was used,but some rocks can have magnetic susceptibilities higher than theupper limit of this meter, even for half cores, so we upgraded to thenew MS3 meter, which is suitable even for iron ore.

TheMS2C1 sensor applies a low frequency (0.565 kHz), low intensity(~80 A/m) alternating magnetic eld. The oscillator frequency is modi-ed due to the sample's presence in the loop, and the new frequency isused to calculate the magnetic susceptibility (Bartington InstrumentsLtd., not dated). Once the core is in the sensor, measurements aremade in 1.1 s with the MS2 meter, or any chosen time with the MS3meter (1 s is typically used). Different loop diameters are available fordifferent core sizes; the INRS laboratory owns two different loops suit-able for most core sizes (e.g., BQ, NQ, HQ and 3-inch).

4.2. Instrument drift, inuence of air temperature

Magnetic susceptibility measurements are dependent on air tem-perature (Geotek Ltd., 2010). To cancel any instrument drift or temper-ature inuence, themeter is zeroed automatically before each new corebox, or series of two boxes, arrives in the sensor. The default distance forzeroing is 10 cm between the MS2C sensor and the core boat's left end,but tests showed that 20 cm was more prudent if very magnetic rocksare logged.

4.3. Calibration, precision, and accuracy

Bartington Instruments Ltd. calibrates each MS2C sensor beforedelivery; in principle, this calibration is stable through time (GeotekLtd., 2010). To verify periodically that the magnetic susceptibilitymeasurement system is still functioning correctly, a calibrationcheck core is provided. This consists of a cylinder lled with slightlymagnetic concrete. The value measured must be within 5% of thenumber written on the cylinder to ensure that a catastrophic calibra-tion error has not occurred (Bartington Instruments Ltd., not dated).We have made this check regularly over a two year period and thecorrespondence between the measured value and the printed valuewas always better than 1.5%; this is an indication of the accuracy ofthe system for susceptibilities of about 58103 SI (depending onthe sensor). To check the accuracy of the system at higher values,we ordered a second check piece from Bartington Instruments Ltd;tests are ongoing and the results will be reported elsewhere.

Precision of measurements can be evaluated by taking a series ofreadings on the check pieces, without moving them. With thelow-susceptibility check pieces, using the MS3 meter and 1 s readings,the relative standard deviation (RSD) of a series of 31 measurementsis 0.01% for each of our MS2C sensors. For the high-susceptibilitycheck piece tested under the same conditions, the RSDs are even lower.

1 Bartington Instruments Ltd. also offers a dual frequency sensor, model MS2B,which allows the user to compute a frequency-dependent magnetic susceptibility.However, this sensor only accepts 25.4 mm cylindrical cores, which must be inserted

manually. Therefore, this sensor would not be appropriate for integration into a MSCL.4.4. Correction for core and loop diameters

The reading displayed by the magnetic susceptibility meter is anuncorrected value (uncor). A correction factor must be applied to ac-count for the diameter of the core (d) vs. the internal diameter of theloop sensor, where d is used as an indication of the core volume. Thisproduces the corrected volumetric magnetic susceptibility () (GeotekLtd., 2010).

4.5. Correction for cut cores

Rock cores that are mineralized or potentially mineralized are typi-cally split mechanically or cut in half lengthwise with a rock saw by ex-ploration companies, for assay purposes. Only half of the core istherefore available for logging in certain intervals. In the case of splitcores, their irregular geometry prevents any kind of simple geometricalcorrection to be made to magnetic susceptibility or density measure-ments; this basically makes split cores unusable for MSCL logging. Thegeometry of cut cores is much simpler and corrections can be madefor the missing volume of rock. The actual thickness of cut cores variessignicant from one piece to the other, so the thickness of the halfcore is determined with a vernier caliper at each measurement spot.The correction factor is a function of the missing vertical surface of thehalf core, assuming its thickness is constant (see Vatandoost et al.,2008).

4.6. Comparison with the KT-10 device

Portable devices such as the KT-10, commercialized by TerraplusInc. of Richmond Hill, Ontario and other vendors, are often used tomeasure magnetic susceptibilities of rocks in the eld or on cores,by government surveys and exploration companies. Such devicesare relatively inexpensive and measurements only take a few sec-onds. Therefore we performed a comparison of the MSCL magneticsusceptibility data with KT-10 measurements to see if they matched.This test was performed at a time when the MS2 meter was beingused, so highly susceptible samples were problematic and are exclud-ed from the analysis. Eight NQ core boxes from Matagami diamonddrill hole BRC-08-72 were selected for this test, representing the var-ious typical lithologies of the area. Measurements were made every10 cm, with the core in core boats, by both methods (MSCL andKT-10 with the 5 cm core width setting). Measurement points lyingon or near fractures were excluded.

The data were divided into low susceptibility measurements(b5103 SI with the MSCL) and high susceptibility measurements(>20103 SI); there were no values measured between 5 and20103 SI. The low values are all from whole cores, in volcanicrocks and weakly magnetic gabbros (boxes #4, #65, #97, #121,#135, and #146), and there were 230 valid measurement couples(Fig. 5). The correlation between the two series (KT-10 vs. MSCL) is96%, with a good visual t, despite systematically lower valuesobtained with the KT-10. A linear regression between the two serieshas a slope of 0.81 when forced through the origin (Fig. 5). This indi-cates that the KT-10 underestimates magnetic susceptibility by 24%on average, for these weakly magnetic rocks. The difference betweenthe two methods may be due to different coil congurations: whenthe core lies horizontally, the MS2C coil is coaxial with the core,whereas the KT-10 coil is placed at on the core, with a verticalaxis. Also, the operating frequencies are not the same (0.565 kHz forthe MS2C sensor on the MSCL and 10 kHz for the KT-10), whichmay inuence the comparison.

For the subset of more magnetic rocks, fewer data are available, andthe rocks tested are of a different nature: magnetic gabbros (wholecores) and massive suldes (cut cores). For the gabbros (box #107,n=45), the correlation between the two series is 90%, and the average

underestimation by theKT-10 is about 33%. For themassive suldes, the

beam is quite small, which means that small-scale heterogeneities ofthe rocks will be apparent in the gamma density data, superimposedon random measurement errors.

5.2. Calibration

TheMSCL records counts per second (cps) from the gamma detector.This is converted to an uncorrected density using a calibration curvewhich is visualized on a plot of uncorrd vs. ln (cps) (e.g., Fig. 6). Calibra-tion is performed two to three times a day during routine logging,depending on the duration of the logging session. Detector drift can bedue to unwanted long term (hours) air temperature variations or toprolonged exposure to gamma rays (Vatandoost et al., 2008).

Calibration involves making long counts on a stepped aluminumcore (2.71 g/cm3) within a core boat. This machined aluminum cali-bration piece is different for each core caliber (e.g., NQ vs. BQ) and com-prises cylindrical sections of different diameters. The maximumdiameter of the calibration piece corresponds to the core caliber, andthe sections of smaller diameters represent various mixtures of airand aluminum within an imaginary cylinder corresponding to thecore caliber. For marine sediments, the imaginary cylinderwould be re-

100 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111number of data points is very small (box #127, n=18), and four sam-ples had magnetic susceptibilities in excess of the MS2 meter's capabil-ities. For the massive sulde sampling points that did not exceed theMS2 meter upper limit, the correlation with the KT-10 measurementsis good, and most measurement couples lie close to the 1:1 line on a bi-nary plot (not shown). The reason for the different behavior of themag-netic gabbro (large underestimation) vs. the massive suldes (goodcorrespondence) is not known.

0

1

2

3

0 1 2 3MS2

KT-

10

1:1

0.5

1.5 2.50.5

1.5

2.5

Fig. 5. Comparison of magnetic susceptibility measurements every 10 cm in severalcore boxes from BRC-08-72 by two methods: the MS2 system mounted on the MSCL,and the handheld KT-10 device. This is the subset of the data with MS2 susceptibilitiesof 5103 SI or less.5. Gamma density

The traditional method ofmeasuring the density of non-porous rockcores is the immersion technique (also known as hydrostaticweighing): each piece is weighted in both air and water, and the drydensity can be deduced (by opposition to the wet density, which re-quireswater-saturation of the sample). This immersion technique is ob-viously impractical for long drill holes, if the sampling interval is to bekept short, since each measurement takes several minutes. It wouldbe a lot worse for wet density measurements. The MSCL uses amuch quicker and reasonably precisemethod, based on the attenuationof gamma rays.

5.1. MSCL measurement method

Measurement of density using gamma-ray attenuation on coresmov-ing on a track is amethod thatwas developed in the 1960s (Evans, 1965).On the current MSCL, a source of 137Cs (10 mCi) located above thelogger's track produces a narrow vertical beam of gamma rays(0.662 MeV)which crosses the core (and core boat) from top to bottom.The photons not attenuated by the core and the core boat are detected onthe other side, using scintillation of a NaI crystal 2 in. thick and 2 in. indiameter (Geotek Ltd., 2010). At these energies, the Compton effect is re-sponsible formost of the deviated photons (Evans, 1965). Gamma atten-uation is a function of core diameter (d) and uncorrected rock density(uncorr), which itself depends on the rock composition. The correctionthat needs to be made to obtain the true density (corr) is explained inSection 5.6. The volume of rock actually interacting with the gamma

9.4 9.5 9.6 9.7 9.8 9.9 100Fig. 6. Example of a calibration curve for uncorrected gamma density (uncorr) mea-surements with the MSCL. Numbers in cm represent the diameter of segments onplaced by an actual core liner lled with water (Best and Gunn, 1999;Blum, 1997). During calibration, each portion of the aluminum piece ismeasured in the gamma density setup for 120 s, which yields a numberof points on the uncorrd vs. ln (cps) diagram (Fig. 6). A linearregression istted through these points. A long exposure time is neededduring calibration to ensure maximal precision. Note that the calibra-tion points on the graph use a theoretical average density of theair-aluminum mixture, the nominal core diameter, and the measuredcps. To nd the density of an unknown sample, one considers the ln(cps) measured on the sample (typically over much shorter exposuretimes), the nominal or measured core thickness, and deduces uncorr.

This calibration method was suggested to us by Geotek Ltd., andseems designed mostly for sediment cores (e.g., Best and Gunn, 1999).Using it on rocks is difcult for several reasons. First, in the Matagamimining camp for example, most samples have densities larger thanthat of aluminum (as opposed to porous sediment cores), which meansthat one is extrapolating beyond the limits of the calibration curve.Second, it would appear that the smallest segments of the aluminum cal-ibration pieces are too small for NQ and BQ coreswith a 5 mmcollimatoron the gamma source. For the NQ piece, at least the 1.5 cm diameter seg-ment must be ignored and the problem is obviously worse for BQ core.

2

4

6

8

10

12

14

2the aluminum calibration piece (see text for explanation). CPS=counts per second.

5.6. Corrected gamma density for unmineralized whole NQ cores

Two factors lead to systematic errors in our uncorrected densitymeasurements: an imperfect calibration method (Section 5.2), and theinuence of the rock composition. Different chemical elements havedifferent Compton attenuation coefcients, so the geochemical compo-sition of a rock sample will inuence its uncorrected gamma density. Ingeneral, rocks composed of common silicate minerals have a narrowrange of Z/A values (atomic number/atomic mass) and attenuation co-efcients; for example average granite has a Z/A of 0.4969 versus0.4938 for an average gabbro (Hallenburg, 1984). Contrast this with avalue of 0.4818 for pure aluminum and it is clear that most of thesystematic error on uncorrected gamma densities is coming from theuse of Al in the calibration piece. Furthermore, when working onunmineralized volcanic and intrusive rocks from a limited geographicregion and age range, with a common tectonic and metamorphichistory, it is reasonable to assume that the true density of these rockswill vary systematically with their compositions, due primarily to igne-ous differentiation. In other words, two rocks with the same densityshould not have completely different compositions (and Z/A values).So a density-dependant adjustment to the uncorrected gamma densi-ties will implicitly take into account the rock composition.

To obtain such a correction curve, we compare uncorrected gamma

2.69

2.70

2.71

2.72

2.73

0 1 2 3 4 5 6 7C

12o

C12

o

C12

o 12 C12o

C22

o C22

o C22

o

5.02

C1 2

o

32

Fig. 7. Tests showing the inuence of the air temperature in the laboratory on theuncorrected gamma density (uncorr) measurements. The tests were done on aluminumso there is no need to correct the densities. (a) During the rst day, the temperaturewas increased by over 10 C over about 3.54 h, which resulted in an increase in mea-sured aluminum density. The temperature was then maintained for a time and later de-creased, but the density continued to increase for several hours. (b) During the secondday, the temperature was maintained within a narrow range. The blue curves show theaverages ofmany10 smeasurements over about 10 min. The error bars represent a hypo-thetical precision of 0.0085 g/cm3. See text for further explanations.

101P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111The empirical way in which these calibration problems are solved isdiscussed below.

5.3. Precision of measurements

The precision of gamma density measurements is a function ofcounting time and of the primary ux of gamma rays. Using the samegamma source collimated at 5 mm, Vatandoost et al. (2008) deter-mined that 8 s of counting were sufcient to reach a 0.01 g/cm3 preci-sion. We veried this during precision tests on an aluminum piece,using 10 s integration times. The aluminum piece was put under thegamma beam and kept there untouched for two days (the gammaux was interrupted overnight). Continuous measurements weremade during about ten minutes, every half hour, resulting in 25 seriesof ~5075 data points (each measurement lasting 10 s). The standarddeviation of the series varied between 0.006 and 0.010 g/cm3, withthe value 0.008 found twelve times and the value 0.009 found tentimes. Therefore we can conrm that the 1 precision of 10 s gammadensity measurements with a 5 mm collimator is slightly better than0.01 g/cm3, or a 0.4% RSD on aluminum.

5.4. Inuence of air temperature

The tests described in Section 5.3 were performed not only to deter-mine precision of the gamma density measurements but also to docu-ment the inuence of long term (hours) changes in air temperatureon the gamma detector, since this is a known drift-causing problem(Geotek Ltd., 2010; Vatandoost et al., 2008). During the rst test, theair temperature in the laboratory was voluntarily increased from18 C to 29 C over 3.54 h, maintained there for ~2.5 h, and thenbrought back to 24 C over one hour. The heating part of the scenariois comparable to what could happen in the laboratory during a hotsunny summer day in eastern Canada, in the absence of air conditioning.The response of the detector to this abrupt heating was impressive andis shown on Fig. 7a. The gamma detector had yet to thermally equili-brate with the ambient air at this stage, since over the rest of the exper-iment, although the air temperature was kept stable and thendecreased, the average measured aluminum density continued to goup steadily, for a cumulative increase in measured density of aboutfour times the precision of individual measurements (Fig. 7a). Thislarge drift is of course unacceptable, which is why the air temperaturemust be kept constant in the laboratory, and multiple calibrations areperformed daily to monitor any drift. The uncorrected gamma densitiesare then calculated by interpolating between the calibrations.

During the second day of these tests, the temperature was kept asconstant as possible, mostly in the 2122 C range, over nearly sevenhours. This represents a typical one work-shift day of logging whenall goes well. Minor temperature uctuations occurred over time-scales of minutes or tens of minutes, but this is unimportant as thesteel and lead shielding of the gamma detector is very thick andthese short-term thermal variations do not perturb the interior ofthe detector. The very satisfactory results are displayed on Fig. 7b.

5.5. Adjustment for cut cores

For half-cores, the same calibration curve is used as for whole coresfor simplicity, but an adjustment is subsequently made to compensatefor the missing core volume. This adjustment is simply d/t where d isthe nominal core diameter and t is the core thickness measured with avernier caliper. For example, for a perfect half core, d/t=2 so theuncorrected gamma density must be doubled to compensate for themissing thickness. This simplistic adjustment assumes that the gammabeam consists of a line crossing an object of constant thickness, whenin fact the beam is a cone crossing a cut cylinder; additional work is

needed to improve gamma density measurements on cut cores.2.69

2.70

2.71

2.72

2.73

2.74

2.75

2.76

C32

oC02

o

C62

o

C92

o

C92

o

C92

o

C62

o

C42

o

C92

o

C9 2

o

C92

o

C72

o

C42

o

C5

.12o

2.74

2.75

2.76

0 1 2 3 4 5 6 7

Co

Co Co

a

bdensity measurements with immersion densities on the same samples.

Whole cores from seven of the eight NQ core boxes used for theMSCL-KT-10 comparison were utilized (boxes #4, #65, #97, #107,#121, #135, and #146 from BRC-08-72). These boxes cover the rangeof unmineralized lithologies, and therefore expected densities, for theMatagami area. The immersion density of 103 numbered whole corepieces was determined at INRS using the conventional method de-scribed by Ross et al. (2011a).

The uncorrected gamma densities were acquired with the MSCLevery 2 cm, yielding between one and eighteenmeasurements per num-bered core piece, depending on their lengths. The average of threethickness measurements per piece (top, middle, bottom) was used incalculating uncorrected gammadensities. The totalwas 646 gammaden-sitymeasurements. TheMSCL datawas averaged for each piece of core toallow a direct comparison with the immersion densities (Fig. 8). Thescatter on this plot is due to two factors: (i) random errors of the 10 sgamma density measurements; and (2) internal heterogeneity of therock samples, which means that gamma densities depend on wherethe measurement is made on the sample. Averaging a number ofgamma densities for each core piece partly removes such effects, but re-call that some pieces are represented by only a few gamma densities inthis test.

The correlation between the two series is 98%. The MSCL overesti-mates the rock densities, but this can be corrected. An empirical correc-tion based on a linear regression through the Fig. 8 data allowscalculation of the corrected gamma density for the Matagami rocks(corr):

measurement volume is not the same for gamma densities vs. immer-sion densities). However, rocks are not homogeneous samples at themillimeter/sub-mm scale, so the remaining mismatch between thegamma densities and the immersion densities are due both to therandom errors and to sample heterogeneity, with the latter factorhaving the greater inuence (Fig. 9).

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

2.752.7 2.8 2.85 2.9 2.95

102 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111corr 0:9666uncorr 0:0038:

This empirical correction is valid for uncorrected gamma densitiesin the range 2.83.2 g/cm3, for whole NQ cores only. Two examplesillustrate the effect of this correction: uncorr=2.80 g/cm3 becomescorr=2.71 g/cm3 (decrease of 0.09 g/cm3), and uncorr=3.20 g/cm3

becomes corr=3.10 g/cm3 (decrease of 0.10 g/cm3).This correction removes much of the systematic error from the

gamma densities, so that only the random error remains. Therefore,on hypothetical homogeneous samples, the accuracy of correctedgamma densities would be similar to the estimated precision of themethod (it is not possible to estimate accuracy directly because the

Fig. 8. Comparison of average uncorrected gamma densities (uncorr), obtained withthe MSCL, with immersion densities, using whole cores from Matagami diamond drillhole BRC-08-72 (unmineralized volcanic and intrusive rocks, NQ size, n=103). A cor-rection can be made to the gamma densities based on the linear regression shown. See

text for details.Fig. 9. Prole of density vs. depth by two methods: corrected gamma densities (blackline with shading) and immersion densities (red rectangles) for a core box of WatsonRhyolite in Matagami diamond drill hole BRC-08-72. Also shown for comparison areuncorrected gamma densities (dashed gray line). The width of the shading representsthe estimated precision of gamma density measurements. The vertical extent of redrectangles illustrates the rst and last gamma density measurement for each corepiece. These core pieces are actually slightly longer, but gamma densities cannot bemeasured too close to the edges. The horizontal extent of the red rectangles is

arbitrary.

103P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111For half cores, andmineralized rocks (whole cores and half cores), re-search is ongoing to develop robust correctionmethods.Mwenifumbo etal. (2005) show that in sulde-rich rocks with densities greater than3.5 g/cm3, the assumption of a nearly-constant Z/A does not hold, andphotoelectric absorption canbecome signicant. Therefore, the empiricalcorrection for whole cores presented here does not apply to mineralizedsamples.

6. Visible light and near infrared (Vis/NIR) spectrometry

Vis/NIR spectrometry (Clark, 1999) is a method that allows detec-tion, semi-quantication, and characterization of certain minerals(composition, crystallographic variations). These can be primary min-erals in the rock, or more commonly secondary minerals such as meta-morphic or alteration minerals. For mineral exploration, the maininterest of Vis/NIR spectrometry is the information that can be extractedon hydrothermal alteration minerals (e.g., Thompson et al., 1999).

Minerals that can be detected belong to four families: silicates,carbonates, sulfates, and oxides. Alunite, amphiboles, anhydrite, biotite,buddingtonite, carbonates (ankerite, calcite, dolomite, magnesite,siderite), chalcedony, chlorite, clays (dickite, halloysite, kaolinite,montmorillonite, smectite), copper oxides, cristobalite, diaspore,epidote, goethite, gypsum, hematite, jarosite, opal, phlogopite, prehnite,pyrophyllite, pyroxenes, serpentine, talc, topaz, tourmaline, vesuvianite,wollastonite, white mica (illite, muscovite, paragonite, phengite),zeolites and zunyite are among the minerals that can be identied(Clark, 1999; Huntington et al., 1997; S. Pontual, pers. commun., 2011;Thompson et al., 1999). Some of these minerals can be very ne-grained and not visible or identiable with a hand lens.

Beyond initial mineral detection, the technique can give indicationson mineral composition. For example, semi-quantication of the Fe:Mgratio of chlorites (Yang et al., 1997), or identication of the varieties ofwhite mica and their Al contents (Cudahy, 1997; Herrmann et al.,2001; Huntington et al., 1999; Scott and Yang, 1997; Tappert et al.,2011; Thompson et al., 2009; Yang et al., 1997, 2011) have explorationapplications for several ore deposit types (volcanogenicmassive sulde,epithermal, iron oxide copper gold, etc.).

6.1. Principle of measurement

The technique, when applied to rock cores, is non-destructive andrequires no sample preparation except ensuring that the surface ofthe core is clean and dry. Penetration of the incoming photons in therock is in the range 30100 m (Pontual, pers. commun., 2011), soVis/NIR spectra are essentially derived from the surface of the sample.

Our LabSpec 2600 Vis/NIR spectrometer from Analytical SpectralDevices (ASD) Inc. of Boulder, Colorado, measures between 350 and2500 nm. Visible light has wavelengths between ~380 and 750 nm(Fig. 10). Infrared radiation between 1300 and 2500 nm is called shortwave infrared (SWIR) in the eld of remote sensing, but is consideredpart of the near infrared (NIR) in the eld of spectrometry (Clark,1999). Here we use the latter convention but note that many previousauthors have utilized SWIR in reports about geological applicationsof this technique.

A light source is attached to the LabSpec 2600 instrument, at the endof a ber optic cable. This high intensity contact probe, also suppliedby ASD Inc., is put in direct contact with the core, illuminating a circleabout 1 cm across. The probe sends back the reected light to the spec-trometer. The ratio between reected light and incident light is calledthe reectance; typical Vis/NIR spectra are plotted as reectance againstwavelength, and themain interesting features of the spectra are the ab-sorption peaks (Fig. 10).

Spectral resolution for the LabSpec 2600 is 3 nm at a wavelengthof 700 nm, 6 nm at 1400 nm and also 6 nm at 2100 nm (ASD Inc.,2007), which is equivalent to that of the TerraSpec Explorer instru-

ment from the same manufacturer. The TerraSpec Explorer is a eldspectrometer that is more commonly used on geological samples.Sampling interval for the LabSpec 2600 is 1.4 nm from 350 to1000 nm and 2 nm between 1000 and 2500 nm. Scanning time is0.1 s and the basic setup averages ten such measurements (integra-tion time of 1 s). Since the signal-to-noise ratio increases with thesquare root of the number of scans used in the averaging (ASD Inc.,2007), smoother spectra can be obtained by increasing the integra-tion time to 5 s or more. The use of uorescent lights should beavoided in the laboratory since such lights add noise to the Vis/NIRspectra.

6.2. Calibration

The LabSpec 2600 Vis/NIR spectrometer should be calibrated everyhour after the initial warm-up period (ASD Inc., 2007). A Spectralonwhite reference tile is put in contact with the probe to obtain a 100% re-ectance reading. In addition the spectrometer obtains a dark currentmeasurement (0% reectance). Spectralon is a special polymermanufactured by Labsphere of North Sutton, New Hampshire, whichis nearly 100% reective for the wavelengths of interest. Gloves shouldbeworn to avoid contaminating the Spectralon tile. Because our labora-tory quickly becomes dusty despite frequent vacuuming, we gentlyclean the reference tile everyday using very ne (1000 grit) sandpaperunder a water stream, and keep the tile protected when not in use.

6.3. Mineralogical interpretation of spectra

The most widespread method of mineralogical interpretation forVis/NIR spectrometry is the visual comparison of each spectrumwith a reference library covering a range of minerals (e.g., Clark etal., 2007; Thompson et al., 1999). Criteria used for mineralogical in-terpretation are the general shape of the spectra, and the study of ab-sorption peaks: depth, width, position (minimum wavelength)(Thompson et al., 1999). This visual method has a subjective ele-ment to it (Canet et al., 2010), since the quality of the interpretationdepends on the person performing it. On the other hand, subtle fea-tures can be detected by an experienced interpreter, which may bemissed by a computer algorithm. However, interpreting each spec-trum separately is impractical for large datasets such as those pro-duced by an MSCL, as such datasets contain thousands of spectra foreach drill hole.

Automated mineralogical identication can be performed by com-puter software. We use the software The Spectral Geologist (TSG),Core version, commercialized by AusSpec International of Australiaand New Zealand. TSG was initially developed by the Australian Com-monwealth Scientic and Research Organization (CSIRO) based onmany years of research. TSG Core can handle on the order of 100,000spectra simultaneously and has a function called The Spectral Assis-tant (TSA)which proposes either onemineral or amixture of 23min-erals for each spectrum in a dataset (Berman and Bischof, 1997;Huntington et al., 1997). This algorithm works on the 13002500 nmportions of the spectra only (although a function for visible light analy-sis has been recently added). Note that the mixture of minerals pro-posed by the algorithm explains the spectral signature of the sample,but the semi-quantitativemineral proportions are notmass or volumet-ric proportions since other minerals such as quartz or feldspar, likely tobe present, are not spectrally active in the Vis/NIR wavelengths.

It is almost certain that automated spectral interpretation(unmixing) will produce some errors ofmineral identication: the al-gorithm is not to be trusted blindly. A proportion of such errors can beeliminated manually by inspecting spectra which are interpreted assurprising minerals, etc. The main strength of automated mineralidentication (apart from its great speed) is in dealing with spectrawhich come from spatially contiguous samples, as these tend to have

similar mineral assemblages. By considering an entire drill hole

104 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111Visible

380

750

1300simultaneously, mineralogical associations possibly corresponding toalteration zones can be identied (Huntington et al., 1997).

6.4. Compositional variations and mineral proportions

In addition to identifyingwhich spectrally activeminerals are presentin each sample, one interest of Vis/NIR spectrometry is the ability to givecompositional indications in minerals such as micas or chlorites. Thisrequires measuring the wavelength positions of certain absorptionpeaks, such as the AlOH peak for white mica (Fig. 10) or the FeOHpeak for chlorites. TSG Core can perform these measurements automat-ically and the numerical data can be exported and plotted in spatialcontext. This can be used to produce, for example, a map, section or

400 800 1200 160

Ref

lect

ance

1400600 1000

ret

aw

ne

erg

Fig. 10. Examples of Vis/NIR spectra for some hydrothermal alteration and metamorphic minhave been vertically offset for clarity, and the main diagnostic absorption peaks are identievery similar, but notice the slight wavelength shift in their AlOH absorption peaks.OC3SWIR

25003D model outlining the different varieties of white mica around avolcanogenic massive sulde deposit.

When two minerals are abundant in a sample, spectral parameterscan be used to estimate their relative proportions. For example, theratio of the AlOH peak depth to the MgOH (or FeOH) peak depthin samples containing both white mica and chlorite is a function ofthe relative abundances of these minerals (e.g., Huston et al., 1999;Thompson et al., 2009).

7. Handheld XRF analyzers

A complementary method to traditional geochemistry is the use ofportable X-ray orescence (pXRF) devices (e.g., Ge, 2008; Glanzman

0 2000 24001800 2200

Illite/smectiteGDS4

ParagoniteGDS109

Chlorite(Mg-rich)SMR-13

HornblendeNMNH117329

CalciteCO2004

MuscoviteGDS107

HOlA

ret

aw

ret

aw

HOl A

MgO

HFe

OH

MgO

H

erals, taken from the U.S. Geological Survey splib06a library (Clark et al., 2007). Spectrad. The two varieties of white mica illustrated (muscovite and paragonite) are spectrally

and Closss, 2007; Peter et al., 2009). The smallest of these areknown as handheld analyzers. These analyzers give immediate insitu (non-destructive) measurements and spatial resolution can bedown to centimeters if needed.

The INRS mobile lab currently includes two Delta Premium ana-lyzers from Olympus Innov-X of Woburn, Massachusetts. One of thesehas been integrated to the MSCL, but time can be saved by using thetwo instruments separately from theMSCL, since the geochemical anal-yses take 6090 s each (much longer than other measurements). Since

pXRF technology is still new, and each device is different, we haveconducted an extensive test program to determine precision and accu-racy of the measurements. While precision was found to be acceptable,the accuracy of the measurements on intact (uncrushed) rocks is poor.We have developed empirical corrections for some elements based oncomparison with traditional geochemistry. The topic of in situ pXRFmeasurements on drill cores and how to correct them is complexenough to justify a separate paper, so the results of our investigationson this topic will be reported elsewhere.

48No

49o

50o

74Wo78o80o 76o

oirat

nO Qu

bec

ARCHEANPROTEROZOICGranitoids

N

Matagami Chibougamau

Rouyn-NorandaTimmins

ARCHEAN

ABITIBIOntario

Qubec

RivireBell

Rivi

reAl

lard

Persvrance

Isle-Dieu

Orchan

Norita

PD-2

PD-1

CaberMatagami

BracemacMcLeod

BellAllard

OrchanOuest

Cavelier

SOUTH FLANK

NORTH FLANK

WEST CAM

McIvor Pluton

intrusions

intrusions

Fault

Diabase(Prot.)

a b

c

ologe B

105P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111P

Fig. 11. (a) Location of the Abitibi greenstone belt in eastern Canada. (b) Simplied gecamp. (c) Simplied geological map of the Matagami area, showing the location of th

Grid is UTM Nad 83, zone 18.Lynx

ical map of the Abitibi greenstone belt showing the location of the Matagami miningracemac volcanogenic massive sulde deposit, modied from Roy and Allard (2006).

8. Application: preliminary results from the Matagamimining camp

The mobile laboratory is being used in applied research projects incollaboration with industry and governments. Data acquisition is nowcompleted for the rst such project in the Matagami area, where thecores from ten diamond drill holes totaling over 7000 m havebeen logged (Ross et al., 2011b, 2012). Here we present the resultsfor one of these holes, BRC-08-72, as an illustration of what multi-sensor core logging can bring to geoscience research and mineralexploration.

8.1. Geological setting

The Matagami mining camp in the northern Abitibi Subprovince(Superior Province) contains numerous volcanogenic massive suldedeposits (e.g., Pich et al., 1993; Fig. 11) and has good potential foradditional discoveries. Diamond drill hole BRC-08-72 crosses theBracemac volcanogenic massive sulde deposit (Fig. 12). This depositwill be mined by Xstrata Zinc Canada and Donner Metals Ltd as part

of the Bracemac-McLeod mine starting in 2013, with proven and prob-able reserves of 3.7 Mt @ 9.6% Zn, 1.3% Cu, 28 g/t Ag and 0.4 g/t Au(GENIVAR, 2010).

The Archean submarine volcanic sequence in the BracemacMcLeodarea starts with the Watson Lake Group, which contains a dacite over-lain by a rhyolite (Adair, 2009). The Key Tufte, a volcano-exhalativehorizon, sits above this felsic sequence and marks the position of mostof the VMS deposits in the Matagami camp (e.g., Pich et al., 1993).The main mineralized lens at Bracemac is located at this level and thesuldes are underlain by a semi-concordant chlorite-sericite alterationzone. The volcanic rocks overlying the Key Tufte belong to theWabassee Group, which here starts with the Bracemac Rhyolite andthen consists of a thick pile of intermediate to mac lavas intercalatedwith minor tuftes. At Bracemac, some of these upper tuftes markthe stratigraphic position of stacked VMS lenses (Fig. 12). Numerous in-trusions, mainly mac in composition, cross-cut the volcanic sequenceand in some cases, the ore bodies.

Diamond drill hole BRC-08-72, which is 662 m-deep, crosses thebottom two ore lenses at Bracemac as well as the whole volcanic se-quence (except the undrilled Watson Dacite) and the intensely alteredzones under the mineralized lenses (Fig. 12).

osit,

106 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111Fig. 12. Vertical cross-section through the Bracemac volcanogenic massive sulde dep

vertical exaggeration. Modied from Adair (2009).looking WNW, showing the trace of the studied diamond drill hole (BRC-08-72). No

8.2. Results and interpretation

Taking into account the overburden portion of the hole (no core)and the few missing core boxes, about 623.5 m of NQ core was avail-able. For physical properties, 1207measurements were made, resultingin an average measurement spacing of ~52 cm (Fig. 13). For XRF geo-chemistry, 1159 measurements were realized (Fig. 14). Some 1207Vis/NIR spectrometric measurements were made with an integrationtime of 1 s, but here we show instead 1135 more reliable measure-ments at 5 s (Fig. 15).

8.2.1. Physical propertiesDensity and magnetic susceptibility are greatest in the massive

suldes, as was expected (Fig. 13). Magnetic susceptibility is also mostly

high in quartz-gabbros compared to other intrusive and volcanic rocks.The physical properties obtained at Matagami will be useful on theirown to plan and interpret geophysical surveys, carry out constrainedinversions of the geophysical data, etc. But they are also useful for multi-variate statistical analyses, including those that aim to improve litholog-ical discrimination, in combination with immobile trace element ratiossuch as Ti/Zr and Al/Zr (Fig. 13). Work on such statistical analyses is on-going (B. Fresia, MSc in progress).

8.2.2. GeochemistryThe Ti/Zr andAl/Zr ratios are especially useful to conrmor interpret

lithological distinctions. The mac-intermediate volcanic rocks andgabbros and have higher Ti and lower Zr concentrations than the under-lying felsic volcanic rocks (Figs. 13, 14). Within the lavas there is a

ic mft in

107P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111Fig. 13. Downhole prole of the physical properties (corrected gamma density; volumetrcorrected immobile element ratios (Ti/Zr; Al/Zr) in BRC-08-72. The graphic summary at le

the core. Andesite should be interpreted as andesite or basalt (undifferentiated). Smoothinagnetic susceptibility; mean reectance in the visible light portion of the spectrum) andthis and subsequent gures is distilled from Xstrata Zinc Canada's original description of

g (3-point moving average) was applied to facilitate data display at this scale.

108 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111decrease in average Ti, average Zr, and average Ti/Zr, below about230 m down-hole depth; this is thought to correspond to the down-ward change from tholeiitic basalts to transitional andesites, known re-gionally from traditional geochemistry.

The original geological log of the core assumes that the rocks belowthe bottom ore zone consist of variably alteredWatson Rhyolite. But theimmobile element ratio proles outline several mac dikes in this por-tion of the core (especially at ~581599 m). Other unreported dikes areinterpreted below themiddle ore zone (below 200 m depth) and at thetop of the Bracemac Rhyolite, just under the Bracemac Tufte (Fig. 13).

The geochemical analyses can also be used to study hydrothermal al-teration. For example, a signicant gain in iron due to chlorite alterationis observed below the bottom ore zone. Iron is also higher in the andes-ite below themiddle ore zone (compare the gains in Fe on Fig. 14 to thehigh Ishikawa index values on Fig. 15). The highest Zr andAl values of the

Fig. 14. Downhole prole of the pXRF geochemistry in BRC-08-72. The data has not been corrPremiumXRF analyzerswere used inminingmode on different portions of the core and the datdrill hole are found in the interval marked highly altered near 600 m.Studies of alteration based on the pXRF data and traditional geochemis-try are ongoing.

8.2.3. MineralogyThe mineralogical prole for BRC-08-72 is dominated by Fe-Mg

chlorite (Fig. 15). Some of this chlorite is metamorphic, especially inthe mac to intermediate rocks. There is no obvious systematic differ-ence in chlorite composition between the very abundant, obviouslyhydrothermal black chlorite under the bottom ore zone, and chloritein the rest of the drilled sequence.

White mica is a common mineral group in the prole, and is moreabundant under the mineralized intervals, although not immediatelybelow the mineralization. Below the bottom ore zone, the whitemicas are muscovitic in composition, whereas they tend to be more

ected to correspond to traditional geochemistry on this plot. Two Olympus Innov-X Deltaa were leveled. Smoothing (3-pointmoving average)was applied to facilitate data display.

109P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111phengitic in the rest of the hole. This suggests that the compositionof white micas may be used as a vector toward ore at Matagami,although more work is needed to conrm this.

Mac to intermediate rocks more commonly display epidote,carbonate and amphibole relative to felsic rocks (Fig. 15). Epidoteand carbonate are typical greenschist grade metamorphic minerals,whereas the amphiboles have not been evaluated.

Fig. 15. Downhole prole of the mineralogy derived from infrared spectrometry for BRC-08-7main group whereas superimposed red squares are the subsidiary mineral groups. Blue lozenOther=otherminerals; N.M.D.=nomineral determined. Data acquisition time: 5 s per spectrite composition is based on thewavelength of the FeOHpeak. The Ishikawa alteration index, bafrom Xstrata Zinc Canada and INRS).8.3. Usefulness of the data

Measurements of physical properties in drill core will make it pos-sible to convert geophysical models into geological models. More-over, high-resolution geochemical and mineralogical measurementson volcanic and intrusive rocks, used together with physical proper-ties, will lead to a better understanding of the volcanic stratigraphy,

2. In the plot of mineral groups automatically extracted by TSG Core, black lines show theges represent white micas for which the estimated relative abundance is less than 15%.rum.Whitemica composition is based on thewavelength of the AlOHpeak, whereas chlo-sed on traditional geochemistry, is shown for reference (calculated using unpublished data

110 P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111volcanic architecture and hydrothermal alteration in the Matagamiarea. This will contribute to ongoing academic, industrial and govern-mental geoscience investigations. The different parameters which aremeasured are valuable on their own, but the data can also be analyzedprotably with multivariate statistical methods.

9. Comparison of the INRS logger with other logging systems

There are two other loggers or logger families in use or formerly inuse on rock cores about which signicant information is publicallyavailable, both from Australia. Here we explain what the commonpoints and differences are between the INRS logger and these othersystems.

9.1. A physical properties logger in Tasmania

Vatandoost et al. (2008) present information about a Geotek MSCLused at the University of Tasmania as part of anAMIRA geometallurgicalproject. The logger was used over the period 20052009 and about8 km of core was logged (M. Roach, written commun., 2012). The pa-rametersmeasuredwith this logger were density, magnetic susceptibil-ity, electrical resistivity, and P-wave velocity; there was also coreimagery being performed, using two camera systems (Vatandoost etal., 2008). The focus of the work wasmineralized (sulde-rich) core in-tervals, with the stated aim being petrophysical characterization ofmetalliferous deposits. The authors noted that axedmeasurement in-terval of 9 cm was optimal in terms of logging efciency, but they alsopointed out spikes in their logs due to broken sections or gaps in thecore. Presumably such spikes were removed during data processing.In our logging protocol, we have instead chosen to avoid gaps and bro-ken core altogether as explained in Section 2.6.

9.1.1. Comparison with the INRS loggerBoth loggers were manufactured by Geotek Ltd and work on the

same general principles. The magnetic susceptibility and density sen-sors are exactly the same (although to solve the problem of the upperlimit of the magnetic susceptibility sensor, the University of Tasmaniahad their MS2C sensor modied by Bartington Instruments Ltd.,whereas INRS had the opportunity to purchase a MS3 meter instead).The University of Tasmania system features two extra sensors forphysical properties (resistivity, P-wave) but does not perform geo-chemical or mineralogical measurements. The INRS logger did not in-clude a P-wave sensor because seismic surveys are uncommon ineastern Canada for mineral exploration purposes, and because wewanted the logger to have a narrow width in order to t in the trailer.The INRS logger did not include the non-contact resistivity sensoreither because of Vatandoost et al. (2008)'s comment that this sensorwas unsuitable for highly conductive mineralized rock samples.

9.2. Hyperspectral loggers

The Commonwealth Scientic and Industrial Research Organization(CSIRO) of Australia has developed three hyperspectral logging systems:HyLogger, HyChips and the TIR Logger. The rst two systems use Vis/NIRspectrometry whereas the third, which is a prototype, utilizes thermalinfrared radiation (Cudahy et al., 2009; Thompson et al., 2009). The sys-tem most comparable to the MSCL at INRS is the HyLogger. It operateswith exploration drill cores in their original trays, moving those trayswith a robotic system so that cores are digitally imaged andVis/NIR spec-tra are acquired (Huntington et al., 2006). CSIRO's website describes ver-sion 2 of this system, which can log 700 m of core per day (CSIRO,2011). The previous version the only one we can nd published litera-ture on (e.g., Huntington et al., 2006; Tappert et al., 2011) had a rela-tively low spectral resolution (16 nm; Huntington et al., 2006); version2 has a spectral resolution of 4 nm (Huntington et al., 2010). Advantages

of the HyLogger are its great logging speed and the high spatialresolution (125 observations per meter for versions 1 and 2, i.e. 8 mmspacing, also described as imaging spectroscopy or hyperspectralscanning by Huntington et al., 2010). CSIRO plans to add thermal infra-red measurements to the HyLogger (in the future version 3), to detectminerals such as feldspar, quartz, etc.

Competitors to theHyLogger include (i) the Hyperspectral Core Im-ager Mark II from the company Corescan Pty Ltd, of Australia; (ii) theSpecCamportable imaging spectrometer from Spectra-Map Ltd of En-gland. We are not aware of scientic publications about these systemsor their applications.

9.2.1. Comparison with the INRS loggerCSIRO's Hyloggers (versions 1 and 2) acquire Vis/NIR spectra and dig-

ital images. These parameters are also measured by the INRS logger, butthe spatial resolution of the spectralmeasurements is typicallymuch lesson the INRS logger, and the logging speed is slower. However, the INRSlogger acquires a number of important parameters not available withthe HyLogger, namely the physical properties and geochemistry.

10. Conclusions

In this paperwe have described amulti-sensor core logger specicallydesigned and adapted to work on rock cores from mineral explorationdrilling. The logger is notmeant to replace the geologist but instead is uti-lized to acquire rock properties measurements useful for geological andgeophysical interpretations. The measurements are non-destructive andcarried out at core storage sites. Using a semi-automated logger allowsthe different types of measurements to be co-located, which facilitatesdata analysis; it also allows a high spatial resolution (small measurementspacing) which is essential to capture quickly changing rock propertiesdue to down-hole lithological variations.

Other loggers have been used on exploration drill cores, but the onedescribed here is the rst to integrate magnetic susceptibility, density,geochemistry, Vis/NIR spectrometry (alterationmineralogy) and imaging.We believe this innovation makes the data especially useful for a widerange of applications. One of the applications illustrated here for theMatagami mining camp is the improvement of down-hole lithologicaldiscriminations when the rocks are ne-grained and/or hydrothermallyaltered. Such improved lithological discriminations can aid better under-standing the geology of an area, which can lead to better targeting for fu-ture drilling. Another application of the INRS logger is to create a completevirtual archive of the core, so that if the core is eventually lost, at least ahigh-quality image is preserved and numerous rock properties havebeen recorded at a high resolution for the benet of future generations.

Acknowledgements

The mobile laboratory was funded by an infrastructure grant fromthe Canadian Foundation for Innovation and the Government of Quebec(ministre de l'ducation, du Loisir et du Sport). Initial work wasperformed using research funding from ministre des RessourcesNaturelles et de la Faune (Qubec) and the Fonds qubcois de recherchesur la nature et les technologies (FQRNT). Xstrata Zinc Canada, DonnerMetals Ltd. and Breakwater Resources provided access to cores and lo-gistical support in Matagami. We thank all of these organizations andespecially the following individuals: M. Allard, M. Dessureault, S.Lacroix, P. Pilote and G. Roy. The LogView software from the GeologicalSurvey of Canada has been used to draft certain gures. B.Morris and ananonymous reviewer made helpful comments on an earlier version ofthe manuscript.

References

Adair, R., 2009. Technical report on the resource calculation for the Bracemac-McLeoddiscoveries, Matagami Project, Qubec (National Instrument 43101 Report).Report Filed on SEDAR. Donner Metals Ltd. (www.sedar.com, on March 2, 2009,

194 pp.).

ASD INC., 2007. LabSpec user manual. ASD Document 600531 Rev H. (72 pp.(downloaded from http://support.asdi.com on 3 January 2011)).

Bartington Instruments Ltd., not dated. Operation manual for MS2 magnetic suscepti-bility system, version OM0408, 71 pp. (downloaded from www.bartington.comon 26 December 2010).

Berman, M., Bischof, L., 1997. The algorithms underlying The Spectral Assistant soft-ware. Exploration and Mining Report 427R.Commonwealth Scientic and Industri-al Research Organization, Australia (114 pp.).

Best, A.I., Gunn, D.E., 1999. Calibration of marine sediment core loggers for quantitativeacoustic impedance studies. Mar. Geol. 160, 137146.

Blum, P., 1997. Physical properties handbook: a guide to the shipboard measurementof physical properties of deep-sea cores. Technical Note, 26. Ocean Drilling Pro-

Huntington, J.H., Cudahy, T.J., Yang, K., Scott, K.M., Mason, P., Gray, D.J., et al., 1999.Mineral mapping with eld spectroscopy for explorationnal report. ExplorationandMining Report 419R.Commonwealth Scientic and Industrial Research Organi-zation, Australia (35 pp.).

Huntington, J.F., Mauger, A.J., Skirrow, R.G., Bastrakov, E.N., Connor, P., Mason, P., et al.,2006. Automated mineralogical core logging at the Emmie Bluff iron oxide coppergold prospect. Primary Industry and Resources of South Australia (PIRSA). MESA J.41, 3844.

Huntington, J., Whitbourn, L., Mason, P., Berman, M., Schodlok, M.C., 2010. HyLoggingvoluminous industrial-scale reectance spectroscopy of the earth's subsurface.Proceedings of ASD and IEEE GRS; Art, Science and Applications of ReectanceSpectroscopy Symposium (Boulder, Colorado, 14 pp.).

Huston, D.L., Kamprad, J., Brauhart, C., 1999. Denition of high-temperature alterationzones with PIMA: an example from the Panorama VHMS district, central Pilbara

111P.-S. Ross et al. / Ore Geology Reviews 53 (2013) 93111altered rocks: an example from the Acoculco Caldera, Eastern Trans-MexicanVolcanic Belt. J. Geochem. Explor. 105, 110.

Clark, R.N., 1981. The spectral reectance of watermineral mixtures at low tempera-tures. J. Geophys. Res. 86, 30743086.

Clark, R.N., 1999. Spectroscopy of rocks and minerals, and principles of spectroscopy.In: Rencz, A.N. (Ed.), Manual of Remote Sensing. Remote Sensing for the EarthSciences, 3. John Wiley and Sons, New York, pp. 358.

Clark, R.N., Swayze, G.A., Wise, R., Livo, E., Hoefen, T., Kokaly, R., et al., 2007. USGS digitalspectral library splib06a. Digital Data Series, 231. U.S. Geological Survey. (http://speclab.cr.usgs.gov/spectral.lib06).

CSIRO, 2011. HyLogging systems.Website http://www.csiro.au/Organisation-Structure/Flagships/Minerals-Down-Under-Flagship/hylogging-systems.aspx (accessed May15, 2012 (last updated on October 14, 2011)).

Cudahy, T.J., 1997. Geometry of the spectral-mineralogical alteration halo at BirthdaySouth, Fimiston Gold Deposit, Western Australia. Exploration and Mining Report421R.Commonwealth Scientic and Industrial Research Organization, Australia(69 pp.).

Cudahy, T.J., Hewson, R., Caccetta, M., Roache, A., Whitbourn, L., Connor, P., et al., 2009.Drill core logging of plagioclase feldspar composition and other minerals associatedwith Archaean gold Mineralisation at Kambalda Western Australia, using a bi-directional thermal infrared reectance system. Reviews in Economic Geology.Society of Economic Geologists, pp. 223235.

Evans, H.B., 1965. GRAPE a device for continuous determination of density andporosity. Proceedings of the 6th annual SPWLA logging symposium. Society ofProfessional Well Log Analysts, Dallas, Texas, pp. B1B25.

Ge, L., 2008. Geochemical prospecting. In: Potts, P.J., West, M. (Eds.), Portable X-rayFluorescence Spectrometry: Capabilities for In Situ Analysis. The Royal Society ofChemistry, Cambridge, U.K., pp. 141173.

Ge, L., Lai, W., Lin, Y., 2005. Inuence of and correction for moisture in rocks, soils andsediments on in situ XRF analysis. X-Ray Spectrom. 34, 2834.

GENIVAR, 2010. Technical Report and Feasibility Study for the Bracemac-McLeodProject, Matagami Area, Quebec. GENIVAR report to Xstrata Zinc Canada. ProjectNo.: AV121288. Report Filed on SEDAR (www.sedar.com, by Donner Metals Ltd.on October 18, 2010, 316 pp.).

Geotek Ltd., 2010. Multi-Sensor Core Logger (MSCL) Manual, version 0510. (176 pp.(downloaded from www.geotek.co.uk on 26 December 2010)).

Glanzman, R.K., Closs, L.G., 2007. Field portable X-rayuorescence geochemical analysis itscontribution to onsite real-time project evaluation. In: Milkereit, B. (Ed.), Proceedingsof Exploration 07: Fifth Decennial International Conference on Mineral Exploration,pp. 291301.

Gray, D.J., 1997. Spectral reectance studies of the impact of water on mineral spectra.Exploration and Mining Report 444R.Commonwealth Scientic and Industrial Re-search Organization, Australia (82 pp.).

Hallenburg, J.K., 1984. Geophysical Logging for Mineral and Engineering Applications.PennWell Books, Tulsa, Oklahoma.

Herrmann, W., Blake, M., Doyle, M., Huston, D., Kamprad, J., Merry, N., et al., 2001. Shortwavelength infrared (SWIR) spectral analysis of hydrothermal alteration zonesassociated with base metal sulde deposits at Rosebery and Western Tharsis,Tasmania, and Highway-Reward, Queensland. Econ. Geol. 96, 939955.

Hunt, C.P., Moskowitz, B.M., Banergee, S.K., 1995. Magnetic properties of rocks andminerals. In: Ahrens, T.J. (Ed.), Rock Physics & Phase Relations: A Handbook ofPhysical Constants, pp. 189204.

Huntington, J., Mason, P., Berman,M., 1997. Geological evaluation of The Spectral Assistant(TSA) formineralogical identication. Exploration andMiningReport 417R.Common-wealth Scientic and Industrial Research Organization, Australia (74 pp.).Craton. AGSO Res. Newsl. 30 (not paginated).Lobell, D.B., Asner, G.P., 2002. Moisture effects on soil reectance. Soil Sci. Soc. Am. J. 66,

722727.Mwenifumbo, C.J., Salisbury, M., Elliott, B.E., Pug, K.A., 2005. Use of multi-channel

gammagamma logs to improve the accuracy of log-derived densities of massivesuldes. Petrophysics 46, 346353.

Peter, J., Mercier-Langevin, P., Chapman, J.B., 2009. Application of eld-portable X-rayuorescence spectrometers in mineral exploration, with examples from the AbitibiGreenstone Belt. Proceedings of the 24th International Applied GeochemistrySymposium, vol. 1, pp. 8386 (Fredericton, New Brunswick).

Pich, M., Guha, J., Daigneault, R., 1993. Stratigraphic and structural aspects of the vol-canic rocks of the Matagami mining camp, Quebec; implications for the Norita oredeposit. Econ. Geol. 88, 15421558.

Ross, P.-S., Bourke, A., Fresia, B., 2011a. Analyse multiparamtrique haute rsolutionde carottes de forage dans la rgion de Matagami Partie 1, Mthodologie et per-formance. Report GM 65521.Ministre des Ressources naturelles et de la Faune,Qubec (72 pp.).

Ross, P.-S., Bourke, A., Fresia, B., Debreil, J.-A., 2011b. Analyse multiparamtrique haute rsolution de carottes de forage dans la rgion de MatagamiPartie 2,rsultats prliminaires. Report GM 65522.Ministre des Ressources naturelles etde la Faune, Qubec (32 pp.).

Ross, P.-S., Bourke, A., Fresia, B., Debreil, J.-A., 2012. Analyse multiparamtrique haute rso-lution de carottes de forage dans la rgion de Matagami 20102012, Rapport nal. Re-port GM 66441.Ministre des Ressources naturelles et de la Faune, Qubec (128 pp.).

Rothwell, R.G., Rack, R.R., 2006. New techniques in sediment core analysis: anintroduction. In: Rothwell, R.G. (Ed.), New Techniques in Sediment Core Analysis:Geological Society of London, Special Publications, pp. 129.

Roy, G., Allard, M., 2006. Matagami, une approche cible sur de nouveaux concepts[abstract]. Qubec Exploration 2006. Document DV 200603. Ministre des Ressourcesnaturelles et de la Faune, Qubec, p. 13.

Scott, K.M., Yang, K., 1997. Spectral reectance studies of white micas. Exploration andMining Report 439R.Commonwealth Scientic and Industrial Research Organization,Australia (35 pp.).

Tappert, M., Rivard, B., Giles, D., Tappert, R., Mauger, A., 2011. Automated drill core log-ging using visible and near-infrared reectance spectroscopy: a case study fromthe Olympic Dam deposit, South Australia. Econ. Geol. 106, 289296.

Thompson, A.J.B., Hauff, P.L., Robitaille, A.J., 1999. Alteration mapping in exploration:application of short-wave infrared (SWIR) spectroscopy. Soc. Econ. Geol. Newsl.39, 1627.

Thompson, A., Scott, K., Huntington, J., Yang, K., 2009. Mapping mineralogy with reec-tance spectroscopy: examples from volcanogenic massive sulde deposits. In:Bedell, R., Crosta, A.P., Grunsky, E. (Eds.), Remote Sensing and Spectral Geology.Reviews in Economic Geology. Society of Economic Geologists, pp. 2540.

Vatandoost, A., Fullagar, P., Roach, M., 2008. Automated multi-sensor petrophysicalcore logging. Explor. Geophys. 39, 181188.

Weber, M.E., Niessen, F., Kuhn, G., Wiedicke, M., 1997. Calibration and application of ma-rine sedimentary physical properties using amulti-sensor core logger. Mar. Geol. 136,151172.

Yang, K., Huntington, J.F., Phillips, R.N., Gemmell, J.B., Fulton, R., 1997. Spectral signa-tures of hydrothermal alteration in the volcanic rocks at Hellyer, Tasmania. Explo-ration and Mining Report 306R.Commonwealth Scientic and Industrial ResearchOrganization, Australia (58 pp.).

Yang, K., Huntington, J.F., Gemmell, J.B., Scott, K.M., 2011. Variations in compositionand abundance of white mica in the hydrothermal alteration system at Hellyer,Tasmania, as revealed by infrared reectance spectroscopy. J. Geochem. Explor.108, 143156.gram. http://dx.doi.org/10.2973/odp.tn.26.1997.Canet, C., Arana, L., Gonzlez-Partida, E., Pi, T., Prol-Ledesma, R.M., Franco, S.I., et al.,

2010. A statistics-based method for the short-wave infrared spectral analysis of

A multi-sensor logger for rock cores: Methodology and preliminary results from the Matagami mining camp, Canada1. Introduction2. General description of the logger and preliminary tests2.1. Principle of operation and logging velocity2.2. Core preparation2.3. Core thickness measurements2.4. Influence of added water2.5. Influence of core temperature2.6. Choice of measurement points

3. Line-scan imaging3.1. Principle of measurement and practical details

4. Magnetic susceptibility4.1. MSCL measurement method4.2. Instrument drift, influence of air temperature4.3. Calibration, precision, and accuracy4.4. Correction for core and loop diameters4.5. Correction for cut cores4.6. Comparison with the KT-10 device

5. Gamma density5.1. MSCL measurement method5.2. Calibration5.3. Precision of measurements5.4. Influence of air temperature5.5. Adjustment for cut cores5.6. Corrected gamma density for unmineralized whole NQ cores

6. Visible light and near infrared (Vis/N