Accepted Manuscript

The characterization, mobility, and persistence of roaster-derived arsenic in soils at

Giant Mine, NWT

Mackenzie J. Bromstad, Lori A. Wrye, Heather E. Jamieson

PII: S0883-2927(16)30216-5
DOI: 10.1016/j.apgeochem.2017.04.004
Reference: AG 3861

To appear in: Applied Geochemistry

Received Date: 18 August 2016

Revised Date: 9 April 2017
Accepted Date: 11 April 2017

Please cite this article as: Bromstad, M.J., Wrye, L.A., Jamieson, H.E., The characterization, mobility,
and persistence of roaster-derived arsenic in soils at Giant Mine, NWT, Applied Geochemistry (2017),
doi: 10.1016/j.apgeochem.2017.04.004.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

The characterization, mobility, and persistence of roaster-derived

arsenic in soils at Giant Mine, NWT

PT
Mackenzie J. Bromstad, Lori A. Wrye, and Heather E. Jamieson
Geological Sciences and Geological Engineering, Queens University, Kingston, ON CANADA
K7L 3N6

RI
Abstract

SC
Approximately 20,000 tonnes of arsenic (As)-bearing emissions from roasting gold (Au)-

bearing arsenopyrite ore were aerially released from 1949-1999 at Giant Mine, near Yellowknife,

U
Canada. Soil samples collected within 4 km of the former roaster from sites undisturbed by
AN
mining or other human activity contain up to 7700 mg/kg total As. Total As concentrations are

highest within a few cm of the surface, and particularly enriched in soil pockets on rock outcrops.
M

Scanning electron microscopy and synchrotron microanalysis show that roaster-derived arsenic

trioxide (As2O3) has persisted in shallow soils in the area. Roaster-generated maghemite and
D

hematite are also present. These anthropogenic forms of As are much more common in near-
TE

surface soils than natural As-bearing minerals. Comparison of the proportions of As, Sb, and Au

concentrations in outcrop soil samples and historic As2O3-rich dust captured by emission controls
EP

suggest most of the roaster-derived As in soils at Giant was likely deposited before 1964.

Topographic restriction by rock outcrops and a dry, cold climate likely contribute to the
C

persistence of As2O3 in outcrop soils.

AC

Limited post-depositional mobility of roaster-generated As is evident from textural

relationships in thin section, the presence of As in weathering products, the variation of As:Sb

with depth, and the elevated As concentrations in soil pore waters sampled by suction lysimeters.

Synthetic gastric and lung fluids extracted an average of 34% and 18% of the total As in three

adjacent soil samples. Given that As2O3 is one of the most toxic and bioaccessible forms of As,
 ACCEPTED MANUSCRIPT

the persistence of roaster-generated As in shallow soils may be a human health risk at this site.

Keywords:
Arsenic trioxide; arsenic; soils; arsenic speciation; Giant Mine; synchrotron

PT
1. Introduction

1.1 Site background

RI
The quartz-carbonate vein- and brittle-ductile shear zone-hosted gold (Au) deposit at Giant

Mine is located in the Kam Group of the Archaean-age Yellowknife Greenstone Belt (Canam

SC
2006). Most Au in the Giant deposit is considered refractory, mainly hosted within the crystal

structure of arsenopyrite (FeAsS) and to a lesser degree arsenian pyrite (FeS2). Roasting of iron

U
(Fe)-sulfide-rich floatation concentrate at approximately 500C oxidized the Fe-sulfides to
AN
nanocrystalline, micro-porous Fe-oxides, principally maghemite and hematite. The porosity of

these roaster-derived Fe-oxides exposed refractory Au to subsequent cyanide leaching (Fawcett

M

and Jamieson 2011; INAC 2007; Walker et al. 2005, 2015; Wrye 2008). Roasting of arsenopyrite
D

also produces SO2 emissions and As-rich vapors that precipitate as arsenic trioxide (As2O3)-rich
TE

dust from the oxidation of As(-I) to As(III):

2FeAsS + 5O2 = Fe2O3 + As2O3 + 2SO2 (Equation 1)

EP

After three years of uncontrolled roaster emissions at Giant (1949-1951), concerns for

worker health and the accidental poisoning deaths of several local children spurred the
C

development of As emissions controls and later efficiency increases (Hutchinson et al. 1982,
AC

Sandlos and Keeling 2012, Sinclair 1951, Walker et al. 2005). The first generation of As2O3

capturing technology, an electrostatic precipitator (ESP), was implemented in 1952 (Walker et al.

2015). Other technological advances in roaster efficiency and emissions capturing technology,

including the installation of a baghouse, resulted in a dramatic drop in emissions after 1958

(approximately 85% of emissions were released pre-1958); however, As2O3-rich dusts were
 ACCEPTED MANUSCRIPT

emitted over the area surrounding Giant Mine until the roaster ceased operation in 1999 (Walker

et al. 2005, Wrye 2008).

This study aims to understand the spatial extent, nature, mobility and age of roaster-derived

As contamination in soils on the Giant Mine property. These questions were investigated through

PT
geochemical and micro-analytical analysis of soil samples from locations isolated from surface

mine activities, mostly focusing on outcrop soil pockets. Co-existing soil pore water sampling

RI
was also completed at a subset of locations, and gastric and lung bioaccessibility testing, done in

SC
partnership with the United States Geological Survey (USGS), was completed at one outcrop soil

location.

U
1.2 Arsenic trioxide in soils
AN
Although arsenolite and claudetite, the two polymorphs of As2O3, have been reported both

as rare weathering products of arsenic-bearing sulfide minerals and from hot springs and
M

fumaroles, their presence in soils is most commonly due to the impact of waste from mining and

mineral processing, and the application of arsenical pesticides and herbicides (Foster and Kim
D

2014; Majzlan et al. 2014, and references therein; Morin and Calais 2006). Majzlan et al. (2014)
TE

describe the most common form of anthropogenic arsenic soil contamination in mining

environments as the release of arsenolite through ore roasting, and this typically results in high
EP

concentrations in near-surface soils. However, in most cases reported, arsenic trioxide has altered

rapidly to As-bearing Fe-oxyhydroxides, hydrated ferric arsenates, or sorbed species on clay

C

minerals (Morin et al. 2002, Cancs et al. 2005, 2008, Arai et al. 2006, Drahota et al. 2009,
AC

Haffert and Craw 2009).

1.3 Arsenic in soils near Giant mine

A previous study of a suite of soil samples from the Yellowknife area beyond the mine

property reported that concentrations of As in mineral soils, humus, and leaf litter vary from up to

30 mg/kg near unmineralized bedrock, to 300 mg/kg over mineralized bedrock (Kerr 2006),
 ACCEPTED MANUSCRIPT

whereas average As content in soils for all of Canada is 6.6 mg/kg (Reimann et al. 2009).

Naturally occurring As in the Yellowknife Greenstone Belt is derived from arsenopyrite-bearing

bedrock and its weathering products (Kerr 2006, Wrye 2008). The presence of roaster-derived

As2O3 and roaster Fe-oxides in soils on the Giant mine property was originally confirmed by

PT
Wrye (2008).

Roaster dust captured during the emissions process and stored underground is the closest

RI
proxy available for estimating the composition of roaster emissions at Giant Mine. The dust contains

SC
60 wt.% As on average (INAC 2007). After As2O3, which makes up approximately 80% of the dust,

the most common As-bearing compounds are roaster-generated Fe-oxides (INAC, 2007; SRK

U
2002a). Fawcett and Jamieson (2011) found 88% of As in one dust sample occurring as As(III), with
AN
the rest as As (V), generally consistent with a mixture of arsenic trioxide and roaster Fe-oxides.

Antimony (Sb) occurs in minor amounts as an impurity in As2O3 and as (As,Sb)2O3 (SRK 2002b).
M

Gold can also be present in the dust within roaster Fe-oxides, with concentrations ranging from 2

to 76 mg/kg documented in ESP and baghouse dust (SRK 2002b). Several changes in processing
D

efficiency, emission controls, and ore quality over mine life (SRK 2002a) correspond with distinct
TE

chemical changes in overall roaster dust composition (Table 1) (Canam 2006). The most notable

change in bulk roaster dust chemistry occurs in dust produced pre- and post-1964, when the last
EP

and most efficient emissions control was installed (see Table 1). Generally, older dust (pre-1964)

had slightly less As and an order of magnitude more Au than newer (post-1964) dust (Table 1).
C

Arsenic trioxide contains 76 wt. % As, and reagent grade As2O3 solubility measures
AC

approximately 10 g/L -16 g/L As under standard state conditions (Pokrovski et al. 1996), while

arsenopyrite solubility is less than 0.00999 g/L at 25C (Craw et al. 2003). However, As2O3 dust

at Giant is less soluble than reagent-grade As2O3. Experimental As2O3 solubility measurements on

select historic roaster dust samples at Giant range from 8.3 g/L to 10.8 g/L, with one outlier of older
 ACCEPTED MANUSCRIPT

dust at 15 g/L at 25C (SRK 2002b). The solubility difference between As2O3 at Giant and

reagent grade As2O3 may be due to Sb content in some Giant As2O3 (Riveros et al. 2000).

Roaster-generated Fe-oxides in tailings at Giant have been documented containing from

<1 wt% up to 7 wt.% As (Walker et al. 2005) as both As(III) and As(V). Most roaster Fe-oxides

PT
are composed of concentrically zoned, nanocrystalline maghemite (Fawcett and Jamieson 2011).

While maghemite is normally metastable relative to hematite at the Earths surface (Majzlan et al.

RI
2003), evidence shows roaster Fe-oxides produced by the roaster at Giant are stable under surface

SC
oxidizing conditions, as the relative proportions of As(III) and As(V) in roaster Fe-oxides

exposed for over 50 years in a historic subaerial tailings deposit are similar to those in day-old

U
calcine (Walker et al. 2005). Kinetic barriers, grain size, changes in free energy due to chemical
AN
composition (Majzlan et al. 2003), and nanocrystallinity (effecting phase stability and reaction

kinetics) might prolong its stability (Fawcett and Jamieson 2011).

M

Rock outcrop constitutes up to 70% of the surface cover in the Yellowknife area (Kerr and

Wilson 2000) and covers 30% of the Giant Mine site (Wrye 2008). Outcrop depressions trap
D

water and promote evaporation during runoff after rainfall and freshet at Giant (Spence and Woo
TE

2002). Our previous work showed that elevated (up to 3280 mg/kg) concentrations of As were

found in soils within depressions in rock outcrops (Wrye 2008). Thus, large outcrops were
EP

optimal sample locations for the purposes of this research.

2. Methods
C
AC

2.1 Field methods

2.1.1 Soil and water samples

Forty outcrop soils within the dominant wind directions from the roaster were sampled in

summer 2010 via grab sampling, mostly targeting soils likely to be high in As based on previous

work (Wrye 2008) and undisturbed by mining activities and infrastructure. Nine soil sample
 ACCEPTED MANUSCRIPT

locations were also sampled for coexisting pore water. Six additional soil samples were taken via

soil core sampling; however, only one core sample was taken in an outcrop soil pocket. See

Figure 1 for sample locations relative to the roaster, and dominant wind directions.

Specific outcrop soil hollows were chosen based on topography, usually favoring those

PT
with few or no visible drainage points. Samples were collected by hand with a trowel and plastic

zipper-seal freezer bags, cleaning the trowel with deionized water and paper towels between

RI
samples. Large pieces of leaf and organic debris were removed before collection, but due to

SC
immaturity outcrop soils tend to be quite heterogeneous, so removing smaller debris was

impractical. Moreover, given the historic impact of aerial emissions of soils, it was thought that

U
removal of surface material may remove evidence of that impact. The soil core samples were
AN
collected with 5 cm diameter hollow aluminum tubes attached to a drivehead, hammered into the

ground with a sledgehammer, cut to appropriate lengths with a pipe cutter, and sealed on the ends
M

with duct tape. At the end of each day soil samples were placed in a freezer.

Soil pore water samples were collected with suction lysimeters after the method described
D

by Rollo and Jamieson (2006). Six-inch (15 cm) and 12-inch (30.5 cm) Soilmoisture Equipment
TE

Corp model 1920F1 suction lysimeters were installed according to manufacturer instructions,

collecting water over 5-7 days through porous ceramic cups via a vacuum-induced hydraulic
EP

gradient. The lysimeter at site 8 was re-installed after the first sample was retrieved; water from

the first install at site 8 and the only installation at site 6 were filtered through 0.45 m filter
C

paper upon collection but other samples were considered to be filtered to 0.45 m by the porous
AC

ceramic cup. Upon sample collection, pH, ORP, conductivity, and temperature measurements

were collected in the field with a YSI multimeter probe calibrated each day. Up to 50 mL of

water were acidified for As and Sb metals analysis preparation at a ratio of 1 mL 20% nitric acid

to 60 mL sample. Sample in excess of 50 mL was bottled but not acidified, to be tested for

common cations and anions. Water samples were placed in a cooler and transported to Taiga
 ACCEPTED MANUSCRIPT

Environmental Laboratories at the end of each day.

Three soil samples were collected from different areas of one outcrop soil location

northwest of the roaster (site 8) and were tested for As bioaccessibility in simulated gastric and

PT
lung fluids. The location chosen for the three bioaccessibility samples was selected based on high

As soil pore water results obtained from suction lysimeter sampling during the first of three

RI
weeks of sampling. The bioaccessibility samples came from three distinctly different sections of

one soil-filled outcrop hollow. Sample locations were chosen to include different soil types,

SC
relative depths to bedrock, and relative drainage levels (detailed in section 4.3 and Figure 6).

Samples for bioaccessibility testing were collected with a metal shovel, cleaned between

U
samples with de-ionized water and paper towels, and placed in heavy duty plastic sample bags
AN
measuring approximately 30 cm x 45 cm. Sample bags were tightly sealed with duct tape and

frozen for transport to Queens University for processing.

M

2.2 Analytical methods

D

2.2.1 Chemical analysis of soil and water samples

TE

A subsample of each frozen soil sample was freeze dried and either 15 g or 0.5 g of

sample (15 g was sent when possible to compensate for inhomogeneity), along with both 15 g and
EP

0.5 g of a standard reference material (SRM 2711- NIST Montana II Soil) were sent to ACME

labs for ultra-trace aqua regia and inductively-coupled plasma-mass spectrometry (ICP-MS)
C

analyses. Samples were neither crushed nor homogenized. Four sample duplicates, two standards
AC

added by the lab (STD DS7 and OREAS45PA) in addition to SRM 2711, and three blanks were

analyzed along with all samples.

For total nitrogen and carbon analysis, samples were ground to a fine powder using a

mortar and pestle. Approximately 0.3 g -1 g of each sample was combusted and analyzed for CO2

via infrared detection in a LECO CNS-2000 Organic Carbon and Nitrogen analyzer. An organic
 ACCEPTED MANUSCRIPT

soil standard, high in carbon and nitrogen, was run after every five samples. A mineral soil

standard, low in both carbon and nitrogen, was run after every ten samples. Blanks were run at

the beginning and the end of each run.

All water samples were analyzed in Yellowknife at Taiga Environmental Laboratories

PT
within a week of their collection. All samples, including one duplicate, were analyzed for total As

and total Sb concentrations using ICP-MS. For samples with water left over after As and Sb

RI
analysis, major cations (Ca2+, Mg2+, K+, Na+) and anions (Cl-, SO42-, F-, NO2-, NO3-) were

SC
analyzed using an ion chromotography scan. One duplicate was run for cation and anion analysis.

Samples for bioaccessibility were air-dried. Samples with high clay content were manually

U
broken apart before sieving with a Retsch Sieve Shaker model AS 200 for 15-25 minutes to
AN
obtain the <250 m and <20 m size fractions needed for gastric and lung bioaccessibility testing.

Six samples were tested, two size fractions from each of the three samples taken for
M

bioaccessibility testing. Testing was performed by the USGS after Morman et al. (2009).
D

2.2.2 Mineralogical analysis of soil samples

TE

Fifteen thin sections (Tables 2a,b, and 4) were prepared according to the methods described

by Walker et al. (2005), then examined under a petrographic microscope (transmitted and
EP

reflected light) to characterize samples and choose possible As-host targets for Environmental

Scanning Electron Microscope (ESEM) and synchrotron microanalysis. Those selected for
C

synchrotron work were lifted from their glass slides by soaking in acetone and mounted on
AC

Kapton tape (Walker et al. 2005).

Although sulfide minerals and roaster Fe-oxides can be recognized easily in thin section,

transparency, minute grain size, and masking by organic matter make non-reflective As hosts

such as As2O3 extremely difficult to find optically. Most soil-borne As hosts (but only those

touching the thin section surface) were detected via back-scatter electron (BSE) detector on an
 ACCEPTED MANUSCRIPT

ESEM. Since As is the dominant heavy element in Giant soils and BSE detectors preferentially

highlight heavier elements, the BSE is well suited for As detection in these materials. High

degrees of magnification made identifying <10 m diameter As hosts and contextual relationships

possible. Uncoated thin sections were examined under low vacuum with the BSE detector on the

PT
MLA 650 FEG ESEM at Queens University and on a Cambridge S200 with EDS system SEM

under ESEM mode at the Geological Survey of Canada, Ottawa.

RI
Beamline X26A at the National Synchrotron Light Source (NSLS), part of Brookhaven

SC
National Laboratory (BNL) in Upton, NY was used for all synchrotron-based analyses. All

synchrotron analyses done for this study used a focused, monochromatic beam with a spot size of

U
5 m V x 9 m H. The beam at X26A was focused from 400 m x 400 m using two 100 mm
AN
long dynamically bent silicon mirrors in Kirkpatrick-Benz geometry. The beam was

monochromatized using two silicon, channel-cut crystal monochrometers. Detectors are at a 90

M

angle to the incident beam. Samples were placed in a sample holder at a 45 angle to the incident

beam, meaning that the beam is an ellipse in shape when it reaches the sample (Walker et al.
D

2005).
TE

Micro X-ray Radiation Fluorescence (XRF) mapping, micro X-ray Absorbance Near-

Edge Structure (XANES), and micro X-ray Diffraction (XRD) analyses were performed to
EP

characterize the As hosts. Micro-XRF fly-scan mapping was used to document element

concentration correlations of targets, and provided a tool to select precise targets for XRD and
C

XANES.
AC

Micro-XRD works somewhat similarly to conventional powder XRD but on a micron

scale, identifying microcrystalline As host minerals present in individual <20 m grains. Two-

dimensional (2-D) XRD images were collected from a Bruker SMART 1500 CCD area detector

with 1024 x 1024 pixel resolution for 60s at 17479 eV and wavelength 0.7039 . Two-

dimensional images were analyzed in the computer program Fit2D (Hammersley 1998), which
 ACCEPTED MANUSCRIPT

was calibrated using standard reference materials -Al2O3 (standard reference material 647a) and

silver behenate (AgC22H43O2). Debye-Scherrer ring patterns are smooth for roaster Fe oxides,

corresponding to randomly orientated nanometer sized particles, typical of the roaster Fe oxides

at Giant. Arsenic trioxide Debye-Scherrer patterns vary from spotty rings to bright spots,

PT
corresponding to submicron to micron-sized crystallites with insufficiently random orientation to

produce smooth rings (Manceau et al. 2002, Walker et al. 2005). Two-dimensional images were

RI
integrated into 1-D diffraction pattern images in Fit2D and then analyzed in the peak matching

SC
software X-Pert High Score Plus against the International Center for Data Diffraction (ICDD

2010) reference pattern database.

U
Oxidation state is important for differentiating natural and anthropogenic As sources in
AN
Giant Mine soils, since natural arsenopyrite and its weathering products occur mainly as As(-I)

and As(V), while anthropogenically-derived As2O3 and roaster Fe oxides occur as As(III) and
M

As(V). Micro-XANES analyses were used to determine the As oxidation state on micron-scale

hosts by observing small energy shifts along the As K absorption edge (11800 eV to 11970 eV),
D

using prepared standards for As(-I), (III), and (V) (arsenopyrite, As2O3, and scorodite
TE

respectively, after Walker et al. 2005). The scorodite standard was run periodically throughout

beam time after any thin section was analyzed with XANES to keep track of energy shifts.
EP

Micro-XANES spectra were analyzed in the program ATHENA (Ravel 2008). Standard and

unknown spectra were normalized and pre-edges and post-edges corrected to remove background.
C

After this, unknown absorption edges were determined and any calibrations necessitated by
AC

changing scorodite drift standards were performed. ATHENA also was used for linear

combination fitting to fit unknown spectra to the three standards and to assign percentage values

to each As oxidation state present.

3. Results
3.1 Total As concentration in soils

3.1.1 Outcrop soil samples

 ACCEPTED MANUSCRIPT

Arsenic concentrations were considered separately for outcrop soil samples and soil

cores, since sample location types, collection methods, depths, and surface areas covered can vary

greatly between soil cores and outcrop samples. Arsenic concentrations in outcrop soil samples

ranged from 156 mg/kg to 5760 mg/kg, with an average of 1550 mg/kg. Only 3 of 40 samples had

PT
As concentrations below the GNWT (2003) guideline for Giant Mine industrial soils of 340

mg/kg [Figure 2a,c, and Table 1S (Supporting Material)], and all samples contained As

RI
concentrations at least one order of magnitude greater than the national guideline for soil of 12

SC
mg/kg (CCME 2007). The range of As values returned for some outcrop soils is comparable to

the range of As values found in tailings at Giant, with 24 of 40 samples containing greater than

U
800 mg/kg As (Bromstad and Jamieson 2012). Although As concentrations generally decreased

with increasing distance from the roaster (Figure 2A,C), and were generally highest within the
AN
dominant downwind directions from the roaster (Figure 1, 2A), these two factors alone do not

fully explain As concentration patterns.

M

Outcrop soils at Giant tend to be very immature, and contain partially weathered
D

fragments of regolith and un-decomposed organic matter. Carbon content ranged from 3% to
TE

48%, with an average of 23%.

3.1.2 Soil cores

EP

Arsenic concentration results from soil cores are presented in Figure 2a,b and Table 2S

(Supporting Material). All core samples were subsampled following the natural soil layers
C

observed, with the uppermost sample generally being the organic-rich humic layer. This resulted
AC

in differences in sample depth intervals (Table 2S). All soil cores returned greater As

concentrations in the top horizon than in any deeper samples, as was the case for our previous

work (Wrye 2008), consistent with an aerial As source. For example, the shallowest, organic-rich

material from C-25 (0-6cm) returned the highest As concentration, 7680 mg/kg As, of any sample

in this project. This shallow sample had a dramatically higher As concentration than the more
 ACCEPTED MANUSCRIPT

mineral-rich soil below (6-15cm), with only 323 mg/kg As (Table 2S). Other soil core As

concentrations ranged from 11 mg/kg to 701 mg/kg As, averaged 177 mg/kg As, and represent

soil depths up to 37 cm.

3.1.3 Duplicate Analyses

PT
The results from duplicate samples reflect the heterogeneity of many outcrop soils and

RI
shallow core soils visually observed during sampling. Even with this soil heterogeneity, relative

percent differences (RPD) were <20% for As in all samples. In most cases Au and Sb RPD were

SC
also <20%; the highest RPD for Au and Sb were for samples tested with only 0.5g of material (5

samples). Arsenic RPD for 0.5g and 15g duplicates of standard reference materials are <15%.

U
Some RPD values for Sb and Au in 0.5g and 15g standard duplicates were >15%; however,
AN
overall concentrations of Sb and Au were low, likely negatively influencing RPD data. All

absolute differences between Au and Sb concentrations across size fractions were <20 ppb (Au)
M

and <6ppm (Sb) (full QA/QC data is in Bromstad (2011)).

D

3.2 Micro-analysis: Arsenic hosts

TE

In this study, arsenic trioxide, arsenopyrite, and roaster-generated iron oxides are referred to

as primary As hosts in soils, regardless of natural or anthropogenic origin. All As hosts resulting from
EP

weathering of these primary As hosts post-deposition are referred to as secondary or as weathering

products.
C

3.2.1. Arsenic Trioxide (As2O3)

AC

Based on examination of 19 thin sections, As2O3 is by far the most abundant manually

microscopically identifiable As host observed in Giant Mine outcrop soils, followed by roaster-

derived Fe-oxides. Synchrotron-based XRD results confirmed that As2O3 is present as

arsenolite, not claudetite, in agreement with Walkers (2006) analysis of roaster dust. Arsenic

trioxide in outcrop soils does not micro-diffract well, and the spotty Debye- Scherrer rings in 2D
 ACCEPTED MANUSCRIPT

XRD images (Figure 3C) indicate that As2O3 crystals found in outcrop soils are often large

relative to the diameter size of the synchrotron beam, i.e. micron to sub-micron crystallites. The

orientation of crystallites could also be insufficiently random to produce full rings (Manceau et

al. 2002, Walker et al 2005). Micro-XANES analysis confirmed the presence of As(III).

PT
As2O3 grains observed in this study range from <5m diameter sub-rounded particles to

grains approximately 50m in diameter with more complex and angular edges and pores, and are

RI
often irregular and intricately patterned (see Figure 3). X-Ray Fluorescence spectra collected for

SC
over 30s on the ESEM caused visible spot damage, but we determined, by collection of -XRD

before and after SEM-induced spot damage, that this melting and recrystallization does not

U
change mineral form or oxidation state (Wrye 2008). The micro-focused synchrotron beam did
AN
not cause noticeable changes to the surface. Faint changes in texture near grain rims were

observed in many cases (Figure 3), as was one reaction rim (containing Fe and possibly As) on an
M

As2O3 grain encapsulated in a mass of organic matter (Figure 3). In samples with abundant

organic carbon, irregular to polygonal clusters of As2O3 grains were often found associated with
D

organic matter (Figure 3). However, many As2O3 grains were not attached to other material.
TE

Euhedral grains, when present, are consistent with descriptions of As2O3 in ESP dust in

underground chamber B236 (deposited 1958-1962) by Riveros et al. (2000); these were euhedral
EP

due to their sublimation directly from the vapor phase, with somewhat porous inter-growths of

two to three crystals.

C

3.2.2 Roaster-generated Fe-oxides

AC

Roaster-generated Fe-oxides examined were generally consistent with those described by

Walker et al. (2005) in terms of textures, As oxidation state ranges, and mineral form. Textures

were either spongy or concentric, and eight grains analyzed with XRD were nanocrystalline

maghemite, with minor hematite and magnetite (see Figure 4a). Arsenic in roaster-generated Fe-

oxides occurs primarily as mixed As(III) and As(V), with considerably varying proportions from
 ACCEPTED MANUSCRIPT

grain to grain.

3.2.3 As-Fe oxyhydroxide weathering rims

Three grains of pyrite with As-bearing Fe-oxyhydroxide rims were identified (Figure 4b).

PT
Similarly enriched Fe-oxyhydroxide weathering rims on pyrite have been found in crushed rock

fill near the townsite at Giant and in exposed historic tailings from Giant (Walker 2006) but have

RI
not previously been documented in soils. Micro-XRF Fe-As mapping (Figure 4b) shows that the

grain rims are enriched in As relative to the pyrite centers. The As in these Fe-oxyhydroxide rims

SC
occurs primarily as As(III), with some As(V) and no As(-I).

U
3.2.4 Other As-bearing Fe-rich grains AN
In addition to As2O3 roaster-generated iron oxides, and As-bearing Fe-oxyhydroxide rims

on sulfides, seven instances of As hosted with Fe (oxyhydr)oxides were observed, with As and Fe
M

in varying proportions (roughly determined via element counts in XRF maps). Most grains

measured <20 m, with the exception of one larger (<100 m) mass on the edge of a large Fe-
D

oxide grain (Figure 4).

TE

Micro-XANES results for these grains indicated a mix of two to three As oxidation states

co-existing, with As dominantly as As(V), with As(-I) present in quantities greater than As(III).
EP

The dominance of As(V) suggests these grains are the result of weathering of As2O3 or

arsenopyrite. While As(-I) in XANES results could be remnant of weathering arsenopyrite,

C

which may be naturally derived, it could also be resultant of the highly penetrative synchrotron
AC

beam intersecting sulfides buried within the thin sections.

Most of these grains micro-diffracted weakly or not at all, meaning they are either coarsely

crystalline or amorphous; one sample produced a usable -XRD diffraction pattern matching a

mixture of goethite and lepidocrocite (Figure 4c). While we have not specifically identified the

other grains discussed in this section, the combination of visual textural evidence, XANES, and
 ACCEPTED MANUSCRIPT

XRF results with Fe:As count ratios estimated from XRF maps suggests there are likely

different As hosts represented in addition to goethite and lepidocrocite. Two possibilities include

an amorphous hydrated Fe-dominated species with more minor As content, and a <10m,

coarsely crystalline Fe-As phase with greater As content, possibly an Fe-arsenate mineral. These

PT
interpretations are based on a lack of diffraction in XRD results, visual textures, and slope

comparisons of Fe-As XRF point count plots.

RI
Arsenopyrite was not identified in any of these samples. It was identified in one soil sample

SC
taken near the roaster in an earlier study (Wrye 2008).

3.3 Co-existing soil pore waters

U
Arsenic in soil pore waters varied widely among the nine sample sites (Table 2 and Figure
AN
5). Concentrations for As ranged from 6.5 g/L to 2085 g/L. The Canadian Council of Ministers

for the Environment (CCME) guideline for As in water for the protection of aquatic life is 5 g/L
M

(CCME 2001).
D

Pore waters containing the highest concentrations of dissolved As correspond to soils with
TE

large quantities of As2O3 observed in thin section. Higher solid concentrations of As correlate

with exponentially higher dissolved As concentrations (Figure 5); microanalysis of soils from
EP

sites with the four highest aqueous As concentrations (6, 8, 19, 24.5) showed As2O3 at all sites,

with qualitatively more As2O3 grains present than other identifiable As hosts for the two most As-
C

rich water sites (Figure 5, circled in red). The two sites (8, 24.5) with qualitatively more As2O3
AC

identified in thin section have lower pH values in co-existing pore waters (Table 2). Field

observations of sites 8 and 24.5 included high organic content and much wetter and deeper

outcrop soil pockets than sites 6 and 19, which may affect these results.

3.4: Bioaccessibility in simulated gastric and lung fluid

Arsenic trioxide is known to be one of the most bioaccessible forms of As to humans

 ACCEPTED MANUSCRIPT

(Plumlee and Morman 2011, Ruby et al. 1999). An average of 34% of the total As in the <250

m fraction of the three samples tested was found to be soluble in simulated gastric fluid, and an

average of 18% of the total arsenic in the <20 m fraction was soluble in simulated lung fluid

(Figure 6). The three samples were collected within 5m of one another at outcrop soil sample

PT
location 8; the difference in As concentrations illustrate local variation in outcrop soils at this

scale, whereas the results from the duplicate bioaccessibility leach sample indicate high

RI
reproducibility from the split of a single sieved sample (Figure 6E). Arsenic trioxide was

SC
confirmed via SEM in thin sections of outcrop soil sample 8, the sieved gastric fraction of B-3,

and both sieved fractions of B-2. In all samples, higher total As concentrations occur in the finer

U
fraction, prepared for lung bioaccessibility testing, suggesting that many As2O3 particles may be

<20m (Table 3). The shallowest sample with the poorest drainage (B-2) contained the highest
AN
As concentrations, an order of magnitude higher than the next most As-rich sample. Despite very

different As concentrations among the three sample locations, neither lung nor gastric
M

bioaccessibility varied greatly, suggesting that As hosts are likely similar from site to site.
D

4. Discussion
TE

4.1 Textural, microanalytical, and aqueous evidence of post-depositional As mobility

The presence of rims (Figure 3) and other textures suggests that some As2O3 particles have
EP

changed since deposition. The correlation of increased dissolved As concentrations with the

increased presence of As2O3 particles observed in thin section suggests some As2O3 is dissolving
C

and has become locally mobile at Giant. The As-bearing Fe-oxyhydroxide rims occurring on As-
AC

poor sulfide grains further evidence local As mobility in Giant Mine soils, and the confirmation

of As(V)-bearing weathering products suggests some pore water As(III) has oxidized. The As(III)

As(V) oxidation reaction is acid-generating (Nordstrom and Archer 2003) and would lower soil

pore water pH unless sufficient carbonate were present to neutralize it; thus, As(III) As(V)

oxidation may contribute to the lower pH values observed in pore water samples with the highest
 ACCEPTED MANUSCRIPT

amounts of coexisting As2O3 (Table 2).

While smaller and more compact As2O3 grains and clumps of grains were observed in thin

section (Figure 3D), possibly the remnants of once larger grains, many As2O3 textures at Giant

maintain large, angularly porous, fragile-looking structures (Figure 3E,F,G). Textural evidence

PT
suggests that the dry and cold conditions in the Yellowknife area lead to very slow weathering of

arsenic trioxide. The persistence of As2O3 in Giant soils contrasts with its apparent instability in

RI
other soil studies (e.g. Yang and Donahoe 2007, Yue and Donahoe 2009, Fitzmaurice et al 2009;

SC
Haffert and Craw 2008).

Although roaster-derived Fe oxides and As-bearing rims on sulfides often have relatively

U
small percentages of As in comparison to As2O3, their Fe content and distinct textures make
AN
spotting them with the ESEM BSE detector possible. Finding hosts with very low concentrations

of both As and other heavy elements, as is the case with organics or clays with sorbed As, is
M

challenging with manually operated SEM. This study may consequently underestimate these

secondary As hosts, and the evidence they could provide for As mobility. Advances in
D

methodology like automated SEM software technology, such as Mineral Liberation Analysis
TE

(MLA) (Buckwalter-Davis 2013), may remove this obstacle to identifying all As hosts, and

provide a way to quantitatively analyze As hosts in Giant Mine soils. Others have applied SEM-
EP

MLA to the characterization of As-hosting phases in lake sediments near Giant Mine and been

able to determine the relative proportions of arsenic trioxide, arsenopyrite, realgar, and other
C

hosts (Howell 2014; Schuh et al. 2015, Van Den Berghe 2015).
AC

4.2 Geochemical evidence of As mobility

4.2.1 Changes in As:Sb ratios and As oxidation (with depth)

Since the arsenic trioxide produced from the Giant roaster also contains Sb (Riveros et al.

2000), As and Sb concentrations is soil were expected to correlate. Hutchinson et al. (1982)
 ACCEPTED MANUSCRIPT

reported an R value (correlation) of 0.95 for As and Sb in 27 shallow soil samples from 1972 and

1974 taken within approximately 10 km of Giant. For 2010 outcrop soil samples, correlation

between As and Sb was poorer, with an R value of 0.48 (Table 4).

Riveros et al. (2000) reported that residual material from leaching experiments on Giant

PT
mine roaster dust was more enriched in Sb than pre-experiment material, suggesting that Sb

content in arsenic trioxide decreased solubility. Reagent grade Sb2O3 is also less soluble than

RI
As2O3 (CANMET 2000 quoted by SRK 2002b). Figure 7A shows higher As:Sb ratios in soil pore

SC
waters compared to co-existing soils, suggesting preferential leaching of As over Sb from arsenic

trioxide, assuming arsenic trioxide is the main source of mobile As in soils. The relatively poor

U
correlation of Sb with As for all 40 outcrop soil samples in this study (Table 4) may be the result
AN
of different mobility of the two elements during the 60 years since the most intense period of

roaster deposition. Hutchinsons samples were taken only 24 years after that period, and their
M

higher Sb:As correlation may reflect reduced opportunity for element migration.

Core sample results show As:Sb ratios increasing with sample depth in all six core
D

samples, irrespective of sample location or site type (Figure 7D). This suggests As is more mobile
TE

than Sb in the soil column. If both As and Sb primarily originate at the surface, and soils are not

physically disturbed, preferential downward propagation of As in the soil column may be

EP

governed by greater As dissolution and spreading through pore water.

XANES results from our previous work (Wrye 2008) using a broadened beam (either 1 mm
C

x 0.349 mm or 600 m x 600 m, compared to the 9m x 5 m beam used in our -XANES

AC

analyses) support As being mobile within the soil column. These broad-beam results provide

information on the relative proportions of As(-I), As(III), and As(V) for the combination of all As

hosts present in the samples under the broadened beam. Three to four soil samples taken from

each of four soil cores, measuring 30 cm to 40 cm in total length, were analyzed by this method.

Generally, As(III) and As(V) coexist in the shallowest samples, with 20-80% As(III) at 2cm
 ACCEPTED MANUSCRIPT

depth. In contrast, As(V) dominates the deepest samples with 100% of As occurring as As(V)

(Wrye 2008). This is consistent with aerial deposition of As2O3 at the surface, and a decrease in

the proportion of As2O3 with depth as a result of dissolution, and an increase in the proportion of

As(V) in weathering products.

PT
4.2.2 Possible role of organic matter

Figures 7 shows that greater organic content correlates with lower As:Sb ratios in soils

RI
(7B), and higher As:Sb ratios in pore water (7C). The As:Sb correlation improves when samples

SC
are grouped based on carbon content (Table 4).

Arsenic can sorb to organic matter under certain conditions (Sadiq 1997), and it has been

U
proposed as a viable As host at Giant Mine by Meunier et al. (2011) and Wrye (2008). Dissolved
AN
organic matter can mobilize weakly sorbed As and decrease adsorption of aqueous As by

competing with it for sorption sites on mineral surfaces and through aqueous complexation
M

(Campbell and Nordstrom 2015, Wang and Mulligan 2009, Bauer and Blodau 2006, Redman et

al. 2002). However, the interplay of organic matter, As, and Fe content is complex (Campbell and
D

Nordstrom 2015) and this likely extends to the situation at Giant.

TE

No competition like that between As and organic acids has been found for Sb, and

comparative studies have found less Sb than As associated with organics in soil (Wilson et al.
EP

2010). At Giant mine, dissolved As and Sb have been reported to be competing for sorption sites

on Fe- and Mn-oxides in sediments, with As out-competing Sb (Fawcett et al. 2011). However,
C

Fe and Mn content in outcrop soils is usually quite low in comparison to carbon (majority of Fe
AC

<2%, Mn <200 mg/kg). With comparatively low Fe-oxyhydroxide content in organic-rich soils,

organics in soil may govern As adsorption (Saada et al. 2003). Organics in outcrop soil pore

waters (humic acids) could possibly help to solubilize As even more compared to Sb, and to keep

dissolved As concentrations high relative to Sb by competing with As for sorption sites.

However, having larger volumes of organic matter present in an outcrop hollow may relate to
 ACCEPTED MANUSCRIPT

having wetter outcrop hollows; locations with the most carbon often supported extremely heavy

moss cover over wet soil, meaning very high carbon content could sometimes be a poor proxy for

soil wetness. In this case, one might expect to find more advanced As dissolution due to the

presence of water, exaggerating the As:Sb ratios, since As is expected to be more soluble than Sb.

PT
4.3: The influence of topography on As mobility

Outcrop soil hollows may concentrate soil-borne As through wash-down, with rain and

RI
snowmelt sweeping roaster-derived As deposited on outcrop surfaces into topographic lows with

SC
restricted drainage (Figure 6). Restrictive topography appears to have both prevented the

migration of As beyond individual outcrop hollows, and to have concentrated As within hollows.

U
For example, bioaccessibility samples B-1, B-2, and B-3 were all taken within 5m of one another
AN
but within different outcrop hollow conditions (Figure 6). Sample B-2 came from the shallowest

(<5cm), driest, most clearly topographically constrained soil, had thinner moss cover, and
M

contained an order of magnitude more As than the next most As-rich sample, B-1, for both

<250m and <20m fractions (Table 3). Sample B-1 contained more As than sample B-3, the
D

sample taken from the deepest soil (>50cm) with the best drainage. However, both of these
TE

samples were more similar to each other than to sample B-2, with relatively moist to wet soil,

grassy-looking vegetative cover, and soil depths >20cm; relative soil porosity and wetness may
EP

play a role in As mobility.

Two important implications for human exposure and remediation relate to the outcrop soil
C

data. Firstly, if the extreme differences between shallow and deep soil As seen in soil core results
AC

are also generally present at outcrop soil locations (i.e., sample C-25), all our outcrop soil sample

results may underestimate As concentrations in the shallowest (<5cm) soils due to dilution with

deeper soil. Kerrs (2006) detailed depth-stratified sampling of mineral soil cores and overlying

humus layers, leaf litter, and tree bark around Yellowknife clearly demonstrates this principle;

soils likely effected by roaster emissions displayed a marked increase in As, Au, and Sb in the
 ACCEPTED MANUSCRIPT

shallowest materials (Kerr 2006).

Secondly, as seen in sample B-2, in extremely shallow soil pockets with constricted or no

drainage, soil-borne As may be both concentrated in outcrop hollows by wash-down, and also

have limited vertically mobility within the soil column. In very shallow outcrop soil pockets (<3-

PT
5cm deep), the lack of any vertical migration of As to deeper soils may lead to greater near-

surface As enrichment in soil pockets with small total soil depths, relative to nearby soil pockets

RI
with greater total soil depths; this means that two outcrop soil pockets located in close proximity

SC
to one another could pose very different levels of potential risk, when considering very shallow

soils. Any depth and topography component to the As story in Giant Mine soils is especially

U
important when considering remediation of the most shallow and topographically constrained As-
AN
rich outcrop soils.

4.4: The age of roaster-derived As in Giant Mine soils

M

Most of the As in the soils sampled for this study is likely roaster-derived. This is based

partly on the identification of arsenic trioxide and roaster-generated Fe oxides in thin section and
D

the absence of arsenopyrite, and partly on the relative enrichment of As in near-surface soils, and
TE

the relationship between As concentrations and direction and distance from the roaster. Kerr

(2006) completed extensive soil sampling around Yellowknife and the only As beyond
EP

background concentrations that was attributed to natural sources was in soil directly overlying

obviously mineralized rocks. No such circumstance was observed during the sampling for this
C

study.
AC

To determine if the roaster-derived As in Giant soils is the product of relatively recent

roaster emissions, or the remnants of the early days of low emission controls, we examined

correlations in As, Au, and Sb amongst outcrop samples, historic roaster dust samples, and

regional background soils. Four outcrop soil samples (O-3, O-27, O-28, O-31) were removed

from the data set for this analysis to limit uncertainty about roaster-derived vs. other
 ACCEPTED MANUSCRIPT

anthropogenic As sources (i.e., waste rock, tailings, mill) due to their proximity to other

contamination sources.

Arsenic, Sb, and Au data is plotted in Figure 8. Although most emissions were released

before 1958 (Bromstad and Jamieson 2012), a change in roaster equipment in 1964 decreased the

PT
amount of Au released in roaster dust and emissions, without much change in As and Sb. Gold in

oxidized areas (i.e. surface environments) is not easily solubilized (Boyle 1979) and is likely to

RI
be immobile. Natural Au in soils around Giant Mine is elemental and any transport or

SC
transformation since the last glaciation has likely been physical, not chemical (Kerr and Knight

2007). Gold may be present in roaster-generated Fe oxides, which are expected to be stable in

U
near-surface soils (Walker et al. 2005).
AN
Both the As:Au and Sb:Au data clearly form a trend between pre-1964 roaster dusts, 2010

Giant Mine outcrop soil samples, and regional soils (data in Table 5). Assuming roaster dusts are
M

a reasonable proxy for roaster emissions, outcrop soil samples appear to have a distinct pre-1964

roaster dust signature. Since 86% of roaster emissions were released pre-1964, this is not
D

surprising. The As:Au (and As:Sb) data are more scattered than the Sb:Au data, consistent with
TE

the already discussed higher mobility of As relative to Sb (Figure 8).

Since these analyses were based on total As concentrations in soils, the continued presence
EP

of pre-1964 roaster-derived As does not preclude the local partial transformation and oxidation of

As2O3.
C
AC

5.0 Conclusions

Evidence reviewed thus far suggests the bulk of roaster-derived As still present in Giant

Mine outcrop soils was deposited over 50 years ago. Its continued presence is likely due to slow

reaction kinetics (i.e., Qi and Donahoe 2008, Haffert and Craw 2009, Yue and Donahoe 2009,

Yang et al. 2005), a dry and cold climate, topography, and the Sb-related decreased solubility of

As2O3 at Giant Mine compared to pure As2O3 (Riveros et al. 2000).

 ACCEPTED MANUSCRIPT

While As2O3 particles are still present in soils, evidence exists supporting slow, limited

change and mobility of roaster-derived As. Partial dissolution textures shown by As2O3 grains

and As-bearing weathering rims on nearby pyrite grains are present. Pore waters associated with

As2O3-bearing soil have elevated concentrations of dissolved As. Arsenic and Sb are more highly

PT
correlated in soil samples from the 1970s than in soils from 2010, and As:Sb ratios increase with

sample depth for core samples. Bulk XANES results show heightened levels of As(V) relative to

RI
As(III) in some soils, possibly reflecting a weathering-related As(V) host that we have not been

SC
able to verify reliably with the methodology in this study, but which may be As sorbed to organic

material or Fe-oxyhydroxide weathering rims. Selective Sequential Extraction (SSE) results

U
suggest a portion of As in Giant soils is now sorbed to organic matter (Wrye 2008) and XAS

results from Meunier et al. (2011) show As(V) hosted in a 70% organic carbon soil, supporting
AN
grain scale evidence for possible dissolution and reaction textures on As2O3. Since As

bioaccessibility was found to be closely related to the identity of As hosts, better understanding of
M

As weathering products present in Giant soils would help define potential risk associated with
D

exposure to these soils.

TE

Topographic constraints, soil depth, type, vegetation, and saturation all contribute to soil

physical and chemical heterogeneity at Giant, and with the potential for high As concentrations in
EP

near-surface soils, sampling soils at standard depth horizons in the future would greatly aid in

understanding the presence of As2O3 and mobility of As with depth. Some of our outcrop grab
C

soil sample results may represent a dilution of higher As concentrations in the shallowest regions
AC

of grab soil samples.

A comprehensive remediation plan for the Giant Mine site was filed in 2007 (INAC 2007)

and the regulator provided a response in 2013 (MVRB 2013). The complexity of the site,

including the management of 237,000 tonnes of As2O3 buried in the underground workings, and

the requirement to involve other parties, notably the local indigenous communities and NGOs
 ACCEPTED MANUSCRIPT

(Non-Governmental Organizations), and the nature of the environmental assessment process have

all contributed to the long timeline. The remediation plan calls for removal or covering of

approximately 960,000 m3 of contaminated soils to industrial site-specific standard of 340 mg/kg

total As, which was established by the City of Yellowknife in 2003 (GNWT 2003). A residential

PT
guideline of 160 mg/kg was established at the same time. These criteria are much higher than the

Canadian Soil Quality Guideline of 12 mg/kg (CCME 2007), and were based on the expectation

RI
that exposure would be limited by the cold climate and that the average natural background

SC
concentration of arsenic in and around Yellowknife was determined to be 150 ppm. The

industrial guideline was also based on the expectation of little or no public access (GNWT

U
2003, Risklogic 2002). Given more recent evidence reported in this paper that the roaster stack

emissions have introduced widespread soil contamination by As2O3, and the construction of a
AN
highway bypass adjacent to some of the highly contaminated outcrop soil pockets, the evaluation

of natural background and future public access may need to be re-examined.

M

6.0 Acknowledgements
D

We are grateful for the assistance of personnel at Giant Mine who helped with site access
TE

for field sampling, as well as the work of field assistant Cirisse Stephen. Thank you to Geoff

Plumlee and Suzette Morman of the USGS for conducting our bioaccessibility testing, and to
EP

AANDC for access to the Giant Mine site.

This study was funded by the Natural Sciences and Engineering Research Council of
C

Canada, Indian and Indigenous Affairs Canada, and the Northern Scientific Training Program.
AC

7.0 References

Arai, Y., Lanzirotti, A., Sutton, S. R., Newville, M., Dyer, J., & Sparks, D. L. (2006). Spatial and

temporal variability of arsenic solid-state speciation in historically lead arsenate

contaminated soils. Environmental Science & Technology, 40(3), 673-679.

Bauer M. and Blodau C. 2006. Mobilization of arsenic by dissolved organic matter from iron
 ACCEPTED MANUSCRIPT

oxides, soils and sediments. Sci. Total Environ. 354, 179-190.

Boyle, R. 1979. The geochemistry of gold and its deposits (together with a chapter on

geochemical prospecting for the element). Geological Survey of Canada, Bulletin 280,

584pp.

PT
Bromstad, M.J., Jamieson, H.E. 2012. Giant Mine, Yellowknife, Canada: Arsenite waste as the

legacy of gold mining and processing. Chap 2. In: The Metabolism of Arsenite, Santini, J.

RI
and Ward, S. (eds.) CRC Press, pp 25-41.

SC
Bromstad, M.J. 2011. The characterization, persistence, and bioaccessibility of roaster-derived

arsenic in surface soils at Giant Mine, Yellowknife, NT. M.Sc. Thesis, Queens University,

U
Kingston, Ontario, Canada.
AN
Buckwalter-Davis, M.J. 2013. Automated mineral analysis of mine waste. M.Sc. Thesis, Queens

University, Kingston, Ontario, Canada.

M

Campbell, K.M., Nordstrom, D.K. 2015. Arsenic speciation and sorption in natural environments.
D

Chapter 3 In: Arsenic: Environmental geochemistry, mineralogy, and microbiology. Eds:

TE

Bowell, R.J., Alpers, C.N., Jamieson, H.E., Nordstrom, D.N., Majzlan, J. Reviews in

Mineralogy and Geochemistry, 79, 185-210.

EP

Canadian Council of Ministers of the Environment (CCME). 2001. Canadian Water Quality

Guidelines for the Protection of Aquatic Life: Arsenic. Canadian Environmental Quality
C

Guidelines XX
AC

Canadian Council of Ministers of the Environment (CCME). 2007. Canadian soil quality

guidelines for the protection of environmental and human health: Summary tables.

Updated September, 2007. In: Canadian environmental quality guidelines, 1999,

Canadian Council of Ministers of the Environment, Winnipeg.

Canam T. W. 2006. Discovery, Mine Production, and Geology of the Giant Mine. In Gold in the

Yellowknife Greenstone Belt, Northwest Territories: Results of the EXTECH III

 ACCEPTED MANUSCRIPT

Multidisciplinary Research Project (eds. C. D. Anglin, H. Falck, D. F. Wright and E. J.

Ambrose). Geological Association of Canada, Mineral Deposits Division, pp. 188-196.

Cancs, B., Juillot, F., Morin, G., Laperche, V., Alvarez, L., Proux, O., Hazemann J.-L., Proux,

O., Brown, G.E. Jr., and Calas, G. (2005). XAS evidence of As (V) association with iron

PT
oxyhydroxides in a contaminated soil at a former arsenical pesticide processing plant.

environmental science & technology, 39(24), 9398-9405.

RI
Cancs, B., Juillot, F., Morin, G., Laperche, V., Polya, D., Vaughan, D. J., Hazemann J.-L.,

SC
Proux, O., Brown, G.E. Jr., and Calas, G. (2008). Changes in arsenic speciation through a

contaminated soil profile: a XAS based study. Science of the Total Environment, 397(1),

U
178-189.

Craw D., Falconer D., Youngson J. H. 2003. Environmental arsenopyrite stability and
AN
dissolution: theory, experiment, and field observations. Chem.Geol.199, 71-82.

Drahota, P., Rohovec, J., Filippi, M., Mihaljevi, M., Rychlovsk, P., erven, V. and Pertold, Z.
M

2009. Mineralogical and geochemical controls of arsenic speciation and mobility under
D

different redox conditions in soil, sediment and water at the Mokrsko-West gold deposit,
TE

Czech Republic. Science of the Total Environment, 407(10), pp.3372-3384.

Fawcett S. E., Jamieson H. E. 2011. The distinction between ore processing and post-depositional
EP

transformation on the speciation of arsenic and antimony in mine waste and sediment.

Chem.Geol. 283, 109-118.

C

Fawcett S.E., Jamieson H.E., Nordstrom D.K., McCleskey R.B. 2011. Arsenic and antimony
AC

geochemistry of mine wastes and associated waters and sediments at the Giant Mine,

Yellowknife, Northwest Territories, Canada. Appl. Geochem. In press.

Fitzmaurice A. G., Bilgin A. A., O'Day P. A., Illera V., Burris D. R., Reisinger H. J., Hering J. G.

2009. Geochemical and hydrologic controls on the mobilization of arsenic derived from

herbicide application. Appl. Geochem. 24, 2152-2162.

Foster, A.L., and Kim C.S. 2014 Arsenic speciation in solids using X-ray Absorption
 ACCEPTED MANUSCRIPT

Spectroscopy. Chapter 5 In: Arsenic: Environmental geochemistry, mineralogy, and

microbiology. Eds: Bowell, R.J., Alpers, C.N., Jamieson, H.E., Nordstrom, D.N., Majzlan,

J. Reviews in Mineralogy and Geochemistry, 79, 257-367.

Government of Northwest Territories (GNWT) 2003. APPENDIX 4: Remediation Criteria for

PT
Arsenic in the Yellowknife Area Soils and Sediment. In Environmental guidelines for

contaminated site remediation (GNWT).

RI
Haffert L., Craw D. 2008. Mineralogical controls on environmental mobility of arsenic from

SC
historic mine processing residues, New Zealand. Appl. Geochem. 23, 1467-1483.

Haffert L., Craw D. 2009. Field quantification and characterization of extreme arsenic

U
concentrations at a historic mine processing site, Waiuta, New Zealand. New Zealand J.

Geol. Geophys. 52:3, 261-272.

AN
Hammersley A.P. 1998. ESRF Internal Report, ESRF98HA01T, FIT2D V9.129 Reference

Manual V3.1.
M

Howell, D. 2014. Arsenic in Lakes Surrounding Yellowknife: Anthropogenic or Naturally

D

Derived? BSc thesis, Queens University.

TE

Hutchinson T.C., Aufreiter S., Hancock G.V. 1982. Arsenic pollution in the Yellowknife area

from gold smelter activities. J. Radioanalytical Chem.71, 58-73.

EP

Indian and Northern Affairs Canada (INAC) 2007. Giant mine remediation plan. Report of the

Giant mine remediation team-Department of Indian Affairs and Northern Development as

C

submitted to the Mackenzie Valley Land and Water Board (MVLWB). 260 pps.
AC

International Centre for Diffraction Data (ICDD) 2010. Power Diffraction Database (PDF-2).

Release 2010. International Centre for Diffraction Data, Newton Square, Pennsylvania,

USA.

Jamieson, H.E. 2014. The legacy of arsenic contamination from mining and processing refractory

gold ore at Giant Mine, Yellowknife, Northwest Territories, Canada. In: Arsenic:

Environmental geochemistry, mineralogy, and microbiology. Eds: Bowell, R.J., Alpers,

 ACCEPTED MANUSCRIPT

C.N., Jamieson, H.E., Nordstrom, D.N., Majzlan, J. Reviews in Mineralogy and

Geochemistry, 79, 533-553

Kerr D. E. 2006. Surficial geology and exploration geochemistry, Yellowknife area. In Gold in

the Yellowknife Greenstone Belt; Northwest Territories: Results of the EXTECH III

PT
Multidisciplinary Research Project (eds. C. D. Anglin, H. Falck, D. F. Wright and E. J.

Ambrose). Geological Association of Canada, Mineral Deposits Division, pp. 301-324.

RI
Kerr D.E. and Knight R.D. 2007. Gold grains in till, Slave Province, Northwest Territories and

SC
Nunavut. Geological Survey of Canada Open File 5463, 8pps.

Kerr D. E. and Wilson P. 2000. Preliminary surficial geology studies and mineral exploration

U
considerations in the Yellowknife area, Northwest Territories. Current Research 2000-C3,

8.
AN
Mackenzie Valley Review Board (MVRB). 2013. Report of Environmental Assessment and

Reasons for Decision: Giant Mine Remediation Project. Reasoned decision report by the
M

MVEIRB concerning the EA surrounding the Giant Mine Remediation Project. 245 pp.
D

Majzlan, J., Drahota, P., and Filippi, M. 2014. Paragenesis and crystal chemistry of arsenic
TE

minerals. In: Arsenic: Environmental geochemistry, mineralogy, and microbiology. Eds:

Bowell, R.J., Alpers, C.N., Jamieson, H.E., Nordstrom, D.N., Majzlan, J. Reviews in
EP

Mineralogy and Geochemistry, 79, 17-184.

Majzlan J., Navrotsky A. and Schwertmann U. 2003. Thermodynamics of iron oxides: Part III.
C

Enthalpies of formation and stability of ferrihydrite ( Fe(OH)3), schwertmannite

AC

( FeO(OH)3/4(SO4)1/8), and -Fe2O3. Geochim. Cosmochim. Acta 68, 1049-1059.

Manceau A., Marcus M.A., and Tamura N. 2002. Quantitative speciation of heavy metals in soils

and sediments by synchrotron x-ray techniques. Reviews in Mineralogy and Geochemistry,

49, pp. 341-428.

Meunier L., Koch I., and Reimer K. J. 2011. Effects of organic matter and ageing on the
 ACCEPTED MANUSCRIPT

bioaccessibility of arsenic. Environmental Pollution 159, 2530-2536..

Morin, G. and Calais, G. 2006. Arsenic in soils, mine tailings and former industrial sites.

Elements 2, 97-101.

Morin, G., Lecocq, D., Juillot, F., Calas, G., Ildefonse, P., Belin, S., Brisois, V., Dillmann P.,

PT
Chevallier C., Gauthier C., Sole, A., Petit, P.-E., and Borensztajn, S. (2002). EXAFS

evidence of sorbed arsenic (V) and pharmacosiderite in a soil overlying the Echassieres

RI
geochemical anomaly, Allier, France. Bulletin de la Socit Gologique de France, 173(3),

SC
281-291.

Morman S.A., Plumlee G.S. and Smith D.B. 2009. Application of in vitro extraction studies to

U
evaluate element bioaccessibility in soils from a transect across the United States and

Canada. Appl. Geochem. 24, 1454-1463.

AN
Nordstrom, D.K. and Archer, D.G., 2003. Arsenic thermodynamic data and environmental

geochemistry. In Arsenic in ground water (pp. 1-25). Springer US.

M

Plumlee, G.S. and Morman S.A. 2011. Mine Wastes and Human Health. Elements 7, 399-404.
D

Pokrovski G., Gout R., Schott J., Zotov A. and Harrichoury J. 1996. Thermodynamic properties
TE

and stoichiometry of As (III) hydroxide complexes at hydrothermal conditions.

Geochimica et Cosmochimica Acta 60, 737-749.

EP

Qi Y. and Donahoe R. J. 2008. The environmental fate of arsenic in surface soil contaminated by

historical herbicide application. Sci. Total Environ. 405, 246-254.

C

Ravel, B. (2008) ATHENA (0.8.56) [software]. Retrieved from:

AC

http://cars9.uchicago.edu/ifeffit/BruceRavel/Horae

Redman A.D., Macalady D.L., and Ahmann D. 2002. Natural organic matter affects arsenic

speciation and sorption onto hematite. Environmental Science and Technology 36 (13),

2889-2896.

Reimann C., Matschullat J., Birke M. and Salminen R. 2009. Arsenic distribution in the

environment: The effects of scale. Appl. Geochem. 24, 1147-1167.

 ACCEPTED MANUSCRIPT

Risklogic Scientific Services, Inc. (Risklogic). 2002. Determining natural (background) arsenic

soil concentrations in Yellowknife, NWT, and deriving site-specific human-health based

remediation objectives for arsenic in the Yellowknife area: Final Report, 34pp.

Riveros P.A., Dutrizac J.E., Chen T.T. 2000. Recovery of marketable arsenic trioxide from

PT
arsenic rich roaster dust. Environmental Improvements in Mineral Processing and

Extractive Metallurgy: Proceedings of the V International Conference on Clean

RI
Technologies for the Mining Industry. Volume II, 135-149.

SC
Rollo, H.A. and Jamieson, H.E. 2006. Interaction of Diamond Mine Waste and Surface Water in

the Canadian Arctic. Appl. Geochem. 21, 1522-1539.

U
Ruby, M.V., Schoof R., Brattin W., Goldade M., Post G., Harnois M., Mosby D.E., Casteel

S.W., Berti W., Carpenter M., Edwards D., Craigin D., and Chappell W. 1999. Advances in
AN
evaluating the oral bioavailability of inorganics in soil for use in human health risk

assessment: American Chemical Society, 33:2, 3697-3705.

M

Saada, A., Breeze D., Crouzet C., Cornu S. and Baranger P. 2003. Adsorption of arsenic (V) on
D

kaolinite and on kaolinitehumic acid complexes: Role of humic acid nitrogen groups.
TE

Chemosphere 51, 757-763.

Sadiq, M. 1997. Arsenic chemistry in soils: an overview of thermodynamic predictions and field
EP

observations. Water, Air and Soil Pollution 93, 117-136.

Sandlos, J. and Keeling A. 2012. Giant Mine: Historical Summary.

C

http://www.reviewboard.ca/registry/project.php?project_id=69
AC

Schuh, C.E., Jamieson H.E., Palmer M.J., and Martin A.J. 2015. Spatial variations in arsenic

geochemistry in sediments and their associated pore waters from lakes in the Yellowknife

region. In 2015 Yellownife Geoscience Forum Abstracts, Technical Program, pp 144.

SENES Consultants Limited (SENES). 2005a. Air quality monitoring at Giant Mine site-

Yellowknife: a baseline study. In: INAC (2007): Giant Mine Remediation Plan, supporting

document A11. 67pp.

 ACCEPTED MANUSCRIPT

Sinclair, G.E.B. 1951. No title. Memo to Major MacKay (Director of Indian Affairs Branch, Dept

of Citizenship and Immigration) referring to Yellowknife death of Frank Abel via arsenic

trioxide poisoning. Accessed online on November 23, 2014:

http://globalnews.ca/news/1682932/timeline-yellowknifes-giant-mine/

PT
Spence C. and Woo M. 2002. Hydrology of subarctic Canadian Shield: bedrock upland. J.

RI
Hydrology 232, 111-127.

SRK Consulting Engineers and Scientists (SRK) 2002a. Final Report: Giant Mine arsenic trioxide

SC
management alternatives. Report submitted to INAC, Yellowknife, Canada, 2002a,140pp.

Available at:

U
http://www.mvlwb.ca/mv/Registry/Forms/FolderView.aspx?RootFolder=%2fmv%2fRegist
AN
ry%2f2007%2fMV2007L8-0031%2ftrioxide&FolderCTID=&View={9F47DAB1-E4D7-

4B6D-A050-461E63EB7AFE}. Accessed 12 September 2011.

M

SRK Consulting Engineers and Scientists (SRK) 2002b. Giant Mine arsenic trioxide dust

properties In: SRK (2002b): Giant Mine arsenic trioxide management alternatives- Final
D

Report, supporting document 5. 63pp.

TE

Van Den Berghe, M. 2015. The mobility of arsenic in sediments and co-existing pore waters from

three small lakes west of Giant Mine, NWT. In 2015 Yellowknife Geoscience Forum
EP

Abstracts, Technical Program, pp 147-148

Walker, S.R. 2006. The solid-phase speciation of arsenic in roasted and weathered sulfides at the
C

Giant Gold Mine, Yellowknife, NWT: Application of synchrotron microXANES and

AC

microXRD at the grain scale. PhD Thesis, Queens University, Kingston, Ontario, Canada.

Walker S.R., Jamieson H.E., Lanzirotti A., Andrade C.F., and Hall G.E.M. 2005. The speciation

of arsenic in iron oxides in mine wastes from the Giant Gold Mine, N.W.T.: Application of

synchrotron micro-XRD and micro-XANES at the grain scale. Can.Mineral. 43, 1205-

1224.
 ACCEPTED MANUSCRIPT

Walker S.R., Jamieson H.E., Lanzirotti A., Hall G.E.M., Peterson R.C. 2015.The effect of ore

roasting on arsenic oxidation state and solid phase speciation in gold mine tailings. Appl.

Geochem., in press. doi:10.1144/geochem2013-238

Wang S., Mulligan C.N. 2009. Enhanced mobilization of arsenic and heavy metals from mine

PT
tailings by humic acid. Chemosphere 74, 274-279.

RI
Wilson S.C., Lockwood P.V., Ashley P.M., Tighe M. 2010. The chemistry and behaviour of

antimony in the soil environment with comparisons to arsenic: A critical review. Env.

SC
Pollution 158, 1169-1181.

Wrye L. 2008. Distinguishing between natural and anthropogenic sources of arsenic in soils

U
from the Giant Mine, Northwest Territories, and the North Brookfield Mine, Nova Scotia.
AN
M.Sc.Thesis, Queens University, Kingston, Ontario, Canada.

Yang L. and Donahoe R.J. 2007. The form, distribution, and mobility of arsenic in soils
M

contaminated by arsenic trioxide, at sites in southeast USA. Appl. Geochem. 22, 320-341.

Yang J., Barnett M., Zhuang J., Fendorf S.E., and Jardine P.M. 2005. Adsorption, oxidation, and
D

bioaccessibility of As(III) in soils. Env. Sci. Technol. 39, 7102-7110.

TE

Yue Z., Donahoe R.J. 2009. Experimental simulation of soil contamination by arsenolite. Appl.

Geochem. 24, 650-656

C EP
AC
 ACCEPTED MANUSCRIPT

Table 1: Summary of Giant Mine roaster ESP and baghouse dust composition

% of total
Year(s) As % As (mg/kg) Au (mg/kg) Sb % Sb (mg/kg) emissions
released
1949-1963

PT
46.4 464200 36.7 1.22 12200 86%
average
1964-1999
65.3 653100 3.22 1.11 11080 14%
average
Data from SRK 2002a. The majority of emissions were released by 1958 (85%), but data is presented here from

RI
1963, when the last change in emissions technology was implemented (Bromstad and Jamieson 2012). For more
complete information see chapter 4 of Bromstad (2011).

U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 2: Pore water results for As and Sb in comparison to soils

Site/Sample As(aq) As(s) Sb(aq) Sb(s) pH ORP volume Thin As2O32

g/L mg/kg g/L mg/kg (mV) (ml) section?1
Site 6 (0.45um filter) 163 825 80 34 6.76 148.7 56.5 Y Y,

PT
moderately
abundant
Site 8 (1st lysimeter) (0.45um 1440 1070 45 196 5.65 160 202 Y Y, very
filter) abundant

RI
Site 8 (2nd lysimeter) 2090 1070 55.5 196 5.25 201 228 Y Y, very
abundant

SC
Site 9 25.1 701 11.1 31 7.26 148 58 N n/a
Site 14 24.6 415 5.8 65 6.09 155 349 N n/a
Site 15 6.5 299 21.9 8 6.41 164 112 N n/a
Site 17 9.5 459 12.5 19 6.94 209 13 N n/a

U
Site 18 51.3 388 23.2 79 6.90 141 219 N n/a
AN
Site 19 180 808 4 9 7.18 195 4 Y Y,
moderately
abundant
Site 24.5 850 1120 13 148 5.25 207 235 Y Y, very
M

abundant
1
Refers to whether or not a thin section of the corresponding soil sample was examined. Thin sections were made
for four of the nine sites where waters were collected.
D

2
Refers to whether or not solid As2O3 was documented in thin section in the corresponding soil results.
Moderately abundant and very abundant are qualitative judgments of the relative abundance of As2O3 particles
in thin section compared to other As hosts. Functionally these judgments usually refer to the proportions of As2O3
TE

to roaster-derived iron oxides.

C EP
AC
 ACCEPTED MANUSCRIPT

Table 3: Gastric and lung bioaccessibility results

Gastric fluid (<250m fraction) Lung fluid (<20m fraction)
leached As, total As, % leached As, total As, % bioaccessibility
mg/kg mg/kg bioaccessibility mg/kg mg/kg
B-1 160 483 33 113 611 18
B-2 1370 4760 29 1180 5850 20

PT
B-3 89.5 238 38 43 270 16
B-3 duplicate 101 238 42 42 270 15
All sieved bioaccessibility samples were made into thin sections.

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 4: Correlations (R-values) for As and Sb in Giant Mine soil samples of different eras
Year collected Samples used As Sb correlation C content # samples
1972 & 1974 Hutchinson et al. (1982) 0.95 n/a 27
soils from nearby Con
and Giant Mines1
2
1999 - 2000 Kerr and Knight 2005, 0.74 n/a 127
Kerr 2001; till samples
within 25km of Giant

PT
2010 ALL 2010 outcrop soil 0.48 all 40
2010 subset 2010 outcrop soil 0.94 0-10% 8
2010 subset 2010 outcrop soil 0.92 10-15% 6
2010 subset 2010 outcrop soil 0.94 15-25% 11

RI
2010 subset 2010 outcrop soil 0.93 >25% 15
2010 subset 2010 outcrop soil 0.93 25-40% 10
2010 subset 2010 outcrop soil 0.98 >40% 5
1

SC
Sample distances range up to approximately 10 km from Giant Mine; sample depths were limited to the upper 6 cm
of soils.
2
Till samples, not including humus material, collected within 25 km radius of Giant Mine

U
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 5: Arsenic, Au, and Sb values for regional background soils and historic roaster dust
As (mg/kg) Au (g/kg) Sb (mg/kg)

PT
Yellowknife regional background till (silt and clay
fraction)1
over granite 5 - 10 0-1 0.1 - 0.2

RI
over volcanics 10 - 30 1 - 10 0.1 - 0.5
Yellowknife regional background humus1
over granite 4-8 1-2 0.2 - 0.4
over volcanics 16 - 79 4 - 49 4 - 7.7

SC
Yellowknife area rocks with mineralized quartz veins
near volcanic or sedimentary contacts1
volcanics 8 19 2.1
mineralized granitoids4 90 20 2

Yellowknife soil background2

U 150
AN
Pre-1964 roaster dust3 46.4% 36700 12200
Post-1964 roaster dust3 65.3% 3220 11080
Soil samples used in previous background calculations could have been affected by anthropogenic roaster
contamination. 1Data from Kerr (2006); 2Data from Risklogic (2002) and GNWT (2003). According to
M

Kerr (2006), locally high As concentrations in the immediate Yellowknife area, located only 5km from
Giant Mine, most likely indicate an anthropogenic As input; 3Data from SRK 2002c. 4Granitic rocks away
from the belt are likely to be significantly lower in As (Kerr 2006).
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

A B

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Figure 1: 2010 sample locations and wind direction. (A) Soil core (triangles), outcrop pocket
soil (circles), soil pore water (exes) and bioaccessibility soil sample locations (red cross).
Outcrop soil samples were taken away from mine activity, mainly within the dominant wind
directions of the roaster (yellow star), shown in (B). Windrose shows dominant wind directions
(to the southeast and to the northwest) from the Yellowknife Airport (SENES 2005).

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Figure 5: Pore water results. Comparison between soil pore water As concentrations (blue,
logarithmic scale) and solid As concentrations in outcrop grab soil samples (brown, linear scale),
for all 11 lysimeter installations; the site 8 lysimeter was installed twice. Soils from the sites with
the highest aqueous As concentrations (circled, 8, 24.5, 6, and 19) were made into thin sections
and examined under ESEM. Samples with the highest aqueous As are circled in red (8 and 24.5).

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Sample Site 8
A B (bioaccessibility, outcrop soil, and water samples)

PT
B-3
B-1
(lysimeter)

RI
B-2
Image from Wrye (2008)
OS 8
C N
Sample Site 8 Approximate Drainage (cartoon)

U SC
AN
(lysimeter)
M
D
TE
EP

D As(mg/kg) E Percent bioaccessibility

6000 50
5000
C

40
4000
30
AC

3000
20
2000
10
1000

0 0
B-1 B-2 B-3 OS 8 B-1 B-2 B-3 B-3*
Gastric As % bioaccessibility
Unsieved <250 m <20 m
Lung As % bioaccessibility
 ACCEPTED MANUSCRIPT

Figure 6: The influence of topography on As concentrations and bioaccessibility. (A)

Cartoon cross-section of the wash down effect. Rainfall and snowmelt may help concentrate
As2O3 deposited on outcrop surfaces in outcrop soil hollows and the soil nearby outcrop bases.
(B) Sample site 8 photograph, showing relative locations of three bioaccessibility samples (B1,
B2, and B3), an outcrop grab sample (OS 8), and the lysimeter pore water sampling installation.
(C) Cartoon of site 8, illustrating relative soil depth to outcrop (graduating gray colors) and
approximate runoff direction (black arrows; dotted lines are approximate topographic high points,

PT
controlling runoff direction). Soil samples were collected within 5m of each other, but in
extremely different depth outcrop hollows. (D) Differences in As concentration for unsieved
samples (yellow; OS 8 and B1), <250m sieved gastric bioaccessibility samples (dark blue; B1,
B2, B3), and <20m lung bioaccessibility samples (light blue; B1, B2, B3). (E) Bioaccessibility

RI
results for simulated gastric (dark blue) and lung (pale blue) fluids from the USGS. Sample B3*
is a duplicate analysis, demonstrating very good reproducibility for bioaccessibility testing.

U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Figure 7: As-Sb correlation. (A) As:Sb ratios for soils vs. coexisting pore waters; water samples
contain greater As:Sb than soils. (B) As:Sb ratios for soils vs. carbon content (proxy for organic
matter). (C) As:Sb ratio for pore waters vs. carbon content of coexisting soils. (D) As:Sb ratios
for the six soil cores sampled in 2010, divided by soil sampling horizon (specific depths vary by
sample; see Table 2b). Horizons are numbered from the surface downwards (i.e., horizon 1 is the
shallowest). Cores show a clear trend of increasing As:Sb ratios with depth for all samples.

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Sb (mg/kg)

R2 = 0.39

PT
A

RI
As (mg/kg)

SC
U
Au (g/kg)

AN
R2 = 0.83
M

B Sb (mg/kg)
D
TE
Au (g/kg)

EP

R2 = 0.39
C

C
AC

As (mg/kg)
 ACCEPTED MANUSCRIPT

Figure 8: The age of soil-borne roaster-derived As. Arsenic, Sb, and Au from outcrop grab
samples are plotted with background soil compositions from Kerr (2006) and pre- and post-1964
roaster dust compositions (Table 7). Data lines of best fit for outcrop soil samples and R2 values
are included. Both (B) and (C) plots indicate a pre-1964 dust source; historic As:Sb dust
compositions in (A) are too close to one another to differentiate dust sources, as As and Sb
content in dust did not change as much over time as Au. t=till, h=humus, l=leaf litter, v=soil over
volcanics, g=soil over granitoids.

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

A1 A2

PT
As-5+

RI
3+
As
-1
As XANES

SC
BSE Photon Energy
20m

B2 E

U
B1 AN
2D- XRD
M

BSE
10m
D

BSE
C1 reaction
rim
C2
F
TE
EP

Arsenolite 10m
2D XRD As2O3
10m
BSE
C

BSE
AC

D1 D2
G
rim

SEM
damage
10m
10m
BSE
BSE Plain and reflected light
 ACCEPTED MANUSCRIPT

Figure 3: Arsenic trioxide (As2O3) microgeochemistry. (A) Single As2O3 grain (A1) As2O3
BSE detector image and (A2) As3+ XANES result. Dashed lines indicate approximate peak
locations for As-1, As3+, and As5+. (B) As2O3 (white grains) from Wrye (2008), clustered in (B1) a
BSE image and (B2) the corresponding 2-D XRD image, matching peaks for the mineral
arsenolite. (C) As2O3 grain (white) within organic mass with slight Fe-bearing reaction rim. (C1)
BSE image, and (C2) spotty and streaky 2-D XRD image matching arsenolite. No other
encapsulated grains have been documented. (D) As2O3 cluster (white grains) associated with

PT
organic matter. (D1) BSE image, (D2) plain and reflected light microscope image. (E and F)
Large As2O3 grains with irregular hole and edge patterns; (E) has somewhat euhedral outer edges.
(G) BSE image of As2O3 grain with slight textural rim visible in one spot, and visible SEM beam
damage caused by collecting EDS spectra for >10 seconds.

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

A1 A3 Maghemite (Fe2O3),
A2 +/-magnetite (Fe3O4)

PT
XANES XRD (2-D)
-1
As : 18% XRD (1-D)

RI
3+
As : 19%
BSE 30 m 5+
As : 63%
2m
B2 B3

SC
B1

Fe-Sb-Pb-S
pyrite

U
pyrite
AN
As-bearing XANES
2m -1
As-bearing
As : 0%
Fe-oxyhydroxide 3+
Fe=green, As=purple
As : 72%
Fe-oxyhydroxide 5+
Plain and reflected light As : 28%
BSE
M

C1
C3 Goethite /
D

Lepidocrocite (Fe-
C2 oxyhydroxide)
TE
EP

XANES
-1
As : 20%
As3+: 15% XRD (2-D)
5+
As : 65%% XRD (1-D)
C

BSE 10 m
AC
 ACCEPTED MANUSCRIPT

Figure 4: Non-As2O3 As host micro-geochemistry. (A) Concentrically zoned roaster-generated

Fe-oxide. (A1) BSE photo and As XANES results showing a mixture of As(-I), As(III), and
As(V); (A2) 2-D XRD and (A3) XRD photos, identified as maghemite +/- magnetite. (B) As-
bearing Fe-oxyhydroxide rim on pyrite. (B1) Plain and reflected light petrographic microscope
image; (B2) BSE photo and XANES results showing the rim is predominantly As(III) with
some As(V); (B3) XRF fly-scan map showing As-rich rim (purple) surrounding As-poor sulfide
center (green). (C) As-bearing Fe-oxyhydroxide; (C1) BSE image with XANES results showing

PT
a mixture of As(-I), As(III), and As(V); (C2) 2-D XRD and (C3) 1-D XRD pattern matching
Fe-oxyhydroxides goethite and lepidocrocite.

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

As (mg/kg)

PT
RI
U SC
AN
M
D
As (mg/kg)

TE
C EP
AC

Distance from roaster (m)

Distance from roaster (m)
 ACCEPTED MANUSCRIPT

Figure 2: Arsenic concentrations in soils. (A) Map showing geographic distribution of As

concentrations for outcrop grab soil samples (circles) and soil cores (triangles). (B) Arsenic
concentration with soil depth in soil core samples, and sample distance from roaster. Horizons
correspond to relative sample distance to the surface; horizon 1 is nearest to the surface. Soil
horizon depths differ between samples depending on specific soil conditions (Table 2b). Amongst
core samples, C-25 is the only soil core taken from an outcrop soil pocket. (C) Arsenic
concentrations for outcrop grab soil samples, and sample distance from the roaster.

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Highlights (The characterization, mobility, and persistence of roaster-derived arsenic in

soils at Giant Mine, NWT)
Roaster-derived As2O3 persists in outcrop soils; some is slowly weathering.
Arsenic concentrations are usually greatest in near-surface soils.
Outcrop soil arsenic concentrations can rival or surpass those in tailings.
Most roaster-derived arsenic in soils at Giant was likely deposited before 1964.
Topography and climate help concentrate arsenic and promote As2O3 persistence.

PT
RI
U SC
AN
M
D
TE
C EP
AC