225

The Canadian Mineralogist

Vol. 43, pp. 225-239 (2005)

CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, BRITISH COLUMBIA:

A GEOLOGICAL ANALOGUE TO CARBON DIOXIDE SEQUESTRATION

LYLE D. HANSEN§ AND GREGORY M. DIPPLE

Department of Earth and Ocean Sciences, University of British Columbia, 3669 Stores Road,
Vancouver, British Columbia V6T 1Z4, Canada

TERRY M. GORDON
Department of Geology and Geophysics, University of Calgary, 2500 University Drive Northwest,
Calgary, Alberta T2N 1N4, Canada

DAWN A. KELLETT
Department of Earth and Ocean Sciences, University of British Columbia, 3669 Stores Road,
Vancouver, British Columbia V6T 1Z4, Canada

ABSTRACT

Listwanite, a carbonate-altered serpentinite commonly associated with gold and mercury mineralization, also represents a
natural analogue to CO2 sequestration via in situ carbonation of minerals. The reaction pathways and permeability structure
controlling listwanite formation are preserved and exposed at Atlin, British Columbia, where listwanite extends tens of meters
into surrounding wallrock. The overall mineralogical transformation is the same as that being considered for industrial sequestra-
tion of CO2. In nature, this reaction proceeds via subreactions that are fossilized as spatially distinct zones. Serpentine + olivine
+ brucite reacted with CO2 to form serpentine + magnesite, then magnesite + talc, and finally magnesite + quartz. These minera-
logical transformations are achieved isochemically, except for the volatile species H2O and CO2. Although the first stage of the
reaction only accounts for about 5–15% of the carbonation potential of serpentinite, it is very widespread, and therefore may have
sequestered a significant portion of the total bound CO2. Moreover, within intact bedrock, the progress of the magnesite + talc
reaction generates fracture permeability, which appears to have locally enhanced reaction. The first two reactions combined
account for about half of the carbonation potential for serpentinite and have a small associated increase in the volume of solids,
which limits porosity loss. They thus hold the greatest promise for in situ mineral carbonation. The carbonation reactions are
controlled by the activity of CO2 in the fluid phase. Proponents of industrial mineral-carbonation processes therefore may seek to
control the composition of the input gas to preferentially drive carbonation reactions that minimize porosity loss and maximize
permeability generation. Magnetite was progressively destroyed during carbonation, which allowed magnetic susceptibility to be
used as a proxy for carbonation-reaction progress. It facilitated the mapping of the permeability structure of these systems and
delineated subtle variations in reaction progress that might otherwise have gone unnoticed in the field.

Keywords: listwanite, serpentinite, talc, magnesite, CO2 sequestration, permeability, mineral carbonation, silica–carbonate al-
teration, carbon management, magnetic susceptibility, reaction progress, Atlin, British Columbia.

SOMMAIRE

La listwanite, une serpentinite carbonatée généralement associée à une minéralisation aurifère et mercurifère, représenterait
un analogue naturel de la séquestration du CO2 via la carbonatation in situ de minéraux. Le déroulement des réactions et la
structure de perméabilité régissant la formation de la listwanite sont préservés et visibles à Atlin, en Colombie-Britannique, où la
listwanite est structuralement liée à un système de fractures et de perméabilité, et s’étend sur des dizaines de mètres dans la roche-
mère intacte. La transformation minéralogique globale est la même que celle qui est à l’étude pour la séquestration industrielle du
CO2. Dans la nature, cette réaction procède selon des sous-réactions fossilisées sous forme de zones étalées dans l’espace. Ser-
pentine + olivine + brucite réagissent avec le CO2 pour former serpentine + magnésite, ensuite magnésite + talc, et enfin magnésite
+ quartz. Ces transformations minéralogiques se déroulent de façon isochimique, exception faite pour les espèces volatiles H2O
et CO2. Quoique la première étape de la réaction ne rend compte que d’environ 5–15% du potentiel en carbonatation de la
serpentinite, il s’agit d’une réaction très répandue, et donc qui aurait pu avoir piégé une proportion importante du CO2 total. De

§
E-mail address: lhansen@eos.ubc.ca
226 THE CANADIAN MINERALOGIST

plus, dans la roche-mère intacte, le progrès de la réaction magnésite + talc produit une perméabilité et des fractures, qui semblent
avoir promu la réaction localement. Les deux premières réactions ensemble rendent compte d’environ la moitié du potentiel de
carbonatation de la serpentinite, et présentent un faible augmentation du volume des solides, ce qui limite la perte en porosité. Les
deux premières étapes de la réaction possèdent donc le plus grand espoir pour une carbonatation in situ. Les réactions de carbona-
tation sont régies par l’activité du CO2 dans la phase fluide. Les exploitants des processus industriels devraient donc chercher à
controller la composition de la phase gazeuse ajoutée afin de forcer les réactions de carbonatation qui minimisent la perte en
porosité et maximisent la génération de perméabilité. La magnétite a progressivement été détruite au cours de la carbonatation, ce
qui nous a permis d’utiliser la susceptibilité magnétique comme indicateur du progrès de la carbonatation. Ceci a facilité la
cartographie de la structure des zones de perméabilité dans ces systèmes et a permis la délinéation de variations subtiles en
progrès de la réaction qui autrement auraient bien pu nous échapper sur le terrain.

(Traduit par la Réaction)

Mots-clés: listwanite, serpentinite, talc, magnésite, séquestration du CO2, perméabilité, carbonatation de minéraux, altération en
silice–carbonate, gérance du carbone, susceptibilité magnétique, progrès de la réaction, Atlin, Colombie-Britannique.

INTRODUCTION ation process is recorded in the form of mineralogical

zonation separated by reaction fronts, which permits
The presence of low-temperature carbonated ultra- each reaction to be examined in isolation. Reactions can
mafic rocks in nature suggests that conditions favorable therefore be examined individually for their potential in
for mineral carbonation exist at shallow levels of the CO2 sequestration and impact on porosity and perme-
crust. In mineral carbonation, CO2 is chemically bound ability.
within the structure of a carbonate mineral by reaction
with Mg2+or Ca2+-derived from silicate minerals. Fossil RELEVANCE TO INDUSTRIAL APPLICATIONS
analogues to mineral-carbonation systems are common
in ultramafic terranes throughout the world and have The carbonation of serpentine and forsterite (Mg-
produced suites of metamorphic rocks composed of ser- dominant olivine), to stable Mg-carbonate minerals, is
pentine, magnesite, talc and quartz-bearing assem- of environmental interest because its ability to fix an-
blages. These rocks are known as listwanite (Kashkai thropogenic carbon dioxide (Seifritz 1990). Globally,
& Allakhverdiev 1965) or as products of silica–carbon- mineral carbonation offers virtually unlimited capacity
ate alteration (Sherlock et al. 1993). Listwanite has his- and the promise of safe, permanent storage of CO2, with
torically been studied because it is commonly spatially little risk of accidental release (Guthrie et al. 2001).
associated with lode-gold mineralization (e.g., Ash Proposed industrial implementation of mineral carbon-
2001, Schandl & Naldrett 1992, Wittkopp 1983) and ation includes the capture of CO2 from a point source
mercury. Liswanite metamorphism has generally been such as a coal-fired power plant, and transport by pipe-
considered a highly metasomatic process that leads to line to a reaction facility for storage. Reaction could
wholesale changes in bulk-rock composition (Schandl occur within an industrial chemical reactor by carbon-
& Naldrett 1992, Sherlock et al. 1993). ation of ultramafic rock mined from a quarry. However,
Here we document the mineral reactions, geochemi- this process demands highly efficient and rapid trans-
cal alteration, and the development of permeability ac- formation of minerals (O’Connor et al. 2001). As an
companying listwanite formation near Atlin, British alternative, direct injection of CO2 into large subsurface
Columbia. Listwanite is controlled structurally by a pre- ultramafic formations allows for reaction times of tens
existing network of faults and fractures, which served to hundreds of years (Guthrie et al. 2001), in a process
as high-permeability pathways for percolation of CO2- known as in situ mineral carbonation.
bearing fluid. Evidence for carbonation, indicated by Numerous schemes of magnesium silicate carbon-
Mg-carbonate minerals, is present for tens of meters into ation have been tested in the laboratory (Goff & Lackner
wallrock adjacent to the controlling fractures. The per- 1998, Goldberg et al. 2001, O’Connor et al. 2001, Wu
meability structure of these systems is therefore similar et al. 2001, Zevenhoven & Kohlmann 2001), and all
to that of proposed in situ systems of mineral carbon- involve chemical reactions such as:
ation, which would utilize joints and fractures as a per-
meability network. Pervasive formation of listwanite Mg3Si2O5(OH)4 + 3 CO2 ⇒ 3 MgCO3
adjacent to the fracture systems is found to be Serpentine Magnesite
isochemical in all major chemical species except H2O + 2 SiO2 + 2 H2O (RA)
and CO2, and the overall mineralogical transformation Quartz
involves breakdown of olivine and serpentine (see be-
low for the relevant reactions). These deposits therefore Mg2SiO4 + 2 CO2 ⇒ 2 MgCO3 + SiO2 (RB)
serve as a geological analogue to mineral carbonation Olivine Magnesite Quartz
as an environmentally important process. The carbon-
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 227

To date, the rate of reaction in the laboratory has GEOLOGICAL SETTING

been reported to be up to about 80% conversion of ser-
pentine to silica and magnesite in 30 minutes at 155°C The ultramafic rocks at Atlin represent a tectonically
and 185 bar (O’Connor et al. 2001). However, this is emplaced upper mantle section of oceanic lithosphere
still too sluggish and the process too costly to accom- (Ash & Arksey 1990a), composed predominantly of
modate industrial CO2 output within industrial reactors. harzburgite and minor dunite, now mostly transformed
In situ mineral carbonation could proceed at a more lei- to serpentinite and listwanite. Aitken (1959), Ash &
surely pace; however, there are no reliable mineral car- Arksey (1990a) and Hansen et al. (2004) documented
bonation experiments in CO2 injection systems (Matter the structural controls of listwanite development
et al. 2002). Reaction-path modeling predicts that CO2 (Fig. 1). Listwanite overprints the foliation of the
injection into serpentine reservoirs (at 250 bars and serpentinite formed during ophiolite obduction, indicat-
60°C) would result in substantial CO2 sequestration in ing that listwanite formation postdates the emplacement
a few tens of years (Cipolli et al. 2004). Moreover, the of ophiolitic material onto the Stikine and Cache Creek
large increase in the volume of solids (21% for serpen- terranes (Ash 1994). The timing of listwanite genesis in
tine and 80% for olivine) associated with reactions RA the Atlin area is also constrained by 40Ar–39Ar age de-
and RB may reduce the sequestration capacity of the termination of chromian muscovite to be in the range
reservoir by destroying permeability at the injection site. ca. 168–172 Ma (Ash 2001). Further details on the

FIG. 1. Simplified geological map of the Atlin area, illustrating the distribution of listwanite along the basal décollement and a
joint-and–fracture permeability network, modified from Ash (1994) and Hansen et al. (2003).
228 THE CANADIAN MINERALOGIST

regional geology near Atlin are provided in Monger pervasive carbonate alteration extends from a few cen-
(1975, 1977a, b), Monger et al. (1978), Bloodgood et timeters to many tens of meters outward from the pri-
al. (1989), Ash & Arksey (1990b), Ash et al. (1991), mary structural controls into intact bedrock. The most
Mihalynuk et al. (1992) and Ash (1994). distal carbonation persists heterogeneously throughout
The temperature of formation of listwanite is con- the map area, and cannot be accurately represented at
strained by fluid-inclusion analysis of quartz–carbonate the scale of Figure 1, where only the more focused and
veins to be in the range of 210–280°C. Homogenization intense carbonation is mapped as listwanite. The reac-
temperatures [ThL–V(L)] of 210 to 240°C were measured tion halos thus extend beyond the area mapped as
on low-salinity (<5 wt% equivalent NaCl) fluid inclu- listwanite, but the spatial distribution of more intensely
sions that show no evidence of phase separation (An- carbonated rocks serves to illustrate the structural con-
drew 1985). Fracture-controlled listwanite likely formed trol of fluid infiltration. Mineralogical zonation within
under sublithostatic pressure on the fluid phase, which the halos is a record of the pathway of the carbonation
is assumed to be approximately 500 bars. Listwanite is reaction.
very common in ultramafic bodies; two other well-
known occurrences include the Timmins area of Ontario MINERALOGICAL ZONATION AND REACTION SEQUENCE
(e.g., Schandl & Naldrett 1992) and the Mother Lode
camp in California (e.g., Wittkopp 1983). Madu et al. The Atlin ultramafic rocks consist of depleted
(1990) and Schandl & Wicks (1991) reported similar mantle-derived harzburgite formed during the adiabatic
temperatures of formation from other occurrences of extraction of melt beneath a mid-ocean ridge (Ash &
listwanite. Arksey 1990a). The melt-extraction process and
serpentinization event(s) have left the residual
STRUCTURAL CONTROL harzburgite chemically heterogeneous prior to listwanite
metamorphism. The variation in modal orthopyroxene
Our field mapping confirms the structural control of is a record of the original heterogeneity of the
listwanite development along the basal décollement and harzburgite. Serpentinization reactions consumed oliv-
along a transecting network of sets of steeply dipping ine + orthopyroxene to form serpentine + magnetite. The
joints and fractures (e.g., Hansen et al. 2004, Fig. 1). extent of serpentization generally exceeds 70% by vol-
Although the joint and fracture systems served as the ume. Serpentine pseudomorphs after orthopyroxene are
primary conduit for infiltration of CO2-bearing fluid, common.
The mineral content of serpentinite and listwanite
was determined by X-ray diffraction (XRD) and optical
and scanning electron microscopy (Table 1). Serpentine
within uncarbonated samples was confirmed as
antigorite by XRD and Raman spectroscopy (Rinaudo
& Gastaldi 2003). Assemblages of uncarbonated
antigorite ± relict olivine, brucite and orthopyroxene
were transformed to assemblages containing antigorite,
magnesite, talc and quartz. The mineral assemblages
track the migration of three carbonation reactions sum-
marized in Figure 2. In sum, these reactions result in the
overall transformations recorded in reactions RA and RB.
The physical separation of the reaction fronts allows
each reaction to be examined in isolation. The bound-
aries separating each mineral assemblage are reaction
isograds in the sense of Carmichael (1970).
The most distal carbonation reactions are volumetri-
cally minor and involve the breakdown of relict olivine
via reaction R1a to antigorite and magnesite (Fig. 3a,
assemblage A2 of Fig. 2):

34 Mg2SiO4 + 20 CO2 + 31 H2O ⇒

Olivine
Mg48Si34O85(OH)62 + 20 MgCO3 (R1a)
Antigorite Magnesite

O’Hanley & Wicks (1995) documented the reaction

of olivine to lizardite and the subsequent transforma-
tion to antigorite at Cassiar, B.C. We have been unable
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 229

to confirm the presence of lizardite in samples subjected 45 MgCO3 + 17 Mg3Si4O10(OH)2 + 45 H2O (R2)
to R1a. Lacking evidence to the contrary, we therefore Magnesite Talc
consider all serpentine to be antigorite.
Trace amounts of brucite are present in some A1 as- which, combined with reaction R1, accounts for about
semblages, but brucite was not found in samples con- half of the overall carbonation-potential of the
taining magnesite. Carbonation of olivine-bearing serpentinite rocks at Atlin (up to ~20 wt%, Table 1, Fig.
samples may therefore also have included the reaction: 4a). It also marks a further increase in the CO2 content
of the fluid phase (Fig. 4b). The cores of large listwanite-
Mg(OH)2 + CO2 ⇒ MgCO3 + H2O (R1b) bearing systems consist of magnesite plus quartz (As-
Brucite Magnesite semblage A4, Fig. 2), which formed by reaction R3 at
the highest activity of CO2 in the fluid phase (Fig. 4b):
Coexisting products and reactants of reaction R1b
have not been observed in any samples collected from Mg3Si4O10(OH)2 + 3 CO2 ⇒ 3 MgCO3
Atlin. For simplicity, reactions R1a and R1b will be con- Talc Magnesite
sidered together as reaction R1. + 4 SiO2 + H2O (R3)
Reaction R1 generally accommodates less than ca. 5 Quartz
wt% CO2 (Table 1) and is limited in progress by the
abundance of olivine (± brucite) prior to carbonation. Thus, the overall transformation of serpentine and
Formation of magnesite at the expense of olivine olivine to quartz and magnesite is the same as that re-
(± brucite) can be driven by a modest increase in the corded in reactions RA and RB. Total carbonation of
activity of CO2 in the fluid phase with negligible de- serpentinite leads to CO2 contents in excess of 36 wt%
crease in activity of H2O (Fig. 4b). Early, distal carbon- on a whole-rock basis (Table 1). Reactions R1, R2 and
ation of brucite and olivine is consistent with the R3 record an increase in the activity of CO2 in an H2O-
observed higher reactivity of olivine and brucite rela- rich fluid (Fig. 4b).
tive to serpentine in mineral-carbonation experiments Reactions R1 to R3 produce magnesite of different
(Guthrie et al. 2001, Lackner et al. 1995). compositions that are discernible with back-scattered
The formation of magnesite plus talc (Assemblage electron imaging because of differences in mean atomic
A3) and the destruction of antigorite (Fig. 3b) mark a mass. Energy-dispersion spectroscopy confirms that
major front of carbonation and comprises most of the increasing mean atomic mass is due primarily to an in-
area mapped as listwanite in Figure 1. Carbonation pro- crease in the Fe content of magnesite. Magnesite M1
ceeded via the reaction: formed during reaction R1 exhibits a euhedral outline
with respect to serpentine and is overgrown by magne-
2 Mg48Si34O85(OH)62 + 45 CO2 ⇒ site M2, with a higher Fe content, associated with talc
Antigorite (Fig. 3b). The dark cores to magnesite in Figures 3c and
3d are likewise interpreted to record growth of magne-
site M1 overprinted by magnesite formed by reaction

FIG. 2. Simplified flow-chart for the reaction path of the Atlin listwanite system during
progressive carbonation of serpentinite.
230 THE CANADIAN MINERALOGIST

FIG. 3. Back-scattered electron images showing progressive carbonation during listwanite genesis. a) Reaction R1b: olivine to
serpentine and magnesite + magnetite. b) Reaction R2: antigorite to magnesite + talc. c) Reaction R3: magnesite + quartz
assemblage. d) Close-up of (c). M1, M2 and M3 are interpreted to represent magnesite generated during reactions R1, R2 and
R3, respectively. Note that magnesite M2 rims M1, and that M3 rims M2 and forms euhedral boundaries with quartz. Light grey
magnesite has a higher mean atomic mass and higher Fe content than dark grey magnesite.

R2. The dark rim of magnesite M3 in Figure 3d contains plies metasomatism. If the mineral reactions at Atlin
less Fe than magnesite M2 and has a euhedral outline in involve substantial mobility of major non-volatile spe-
contact with quartz. cies, then this system may not serve as a useful ana-
logue for mineral carbonation.
GEOCHEMICAL CHANGE DURING Whole-rock chemical compositions from Atlin (data
LISTWANITE FORMATION in the Appendix) are consistent with the hypothesis that
carbonation was not accompanied by chemical changes
The mineralogical transformations in reactions R1 to of major non-volatile chemical species. Decreases in
R3 can be achieved through (de)hydration–carbonation wt% MgO and SiO2, which combined make up about
reactions without modification of the major oxide 90% of the non-volatile component of the Atlin rocks,
chemical composition of serpentinite. Previous investi- correlate with an increase in CO2 content at a constant
gators of listwanite, however, have argued for perva- MgO:SiO2 ratio (Fig. 5). This trend is consistent with
sive chemical changes accompanying carbonation the immobility and passive depletion of Si and Mg dur-
(Schandl & Naldrett 1992, Sherlock et al. 1993, Buisson ing mass increase (CO2 addition). Serpentinite and
& LeBlanc 1985). Indeed, the term silica–carbonate al- listwanite overlap completely in composition if compo-
teration that is commonly applied to these systems im- sitions in terms of major oxides are renormalized to
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 231

FIG. 4. a) Comparison of the observed mineral content and whole-rock composition in a (MgO + FeO) – SiO2 – CO2 diagram
projected from H2O. b) Mineral stability in the system MgO–SiO2–CO2–H2O as a function of activity of H2O and CO2 in the
fluid phase, calculated using PTAX and the mineral database of Berman (1988). The dashed line is the metastable extension
of reaction R1a. The arrowed dashed line is the path of P(fluid) = 500 bars for a binary H2O–CO2 fluid, calculated with the
CORK equation of state (Holland & Powell 1991).

FIG. 5. a) Weight percent MgO plotted versus SiO2. Circle size is proportional to CO2 content. Data are presented in the Appen-
dix. The observed trend is consistent with the passive dilution of MgO and SiO2 by increase in CO2 content. b) MgO and SiO2
whole-rock compositions are renormalized to 100% (excluding H2O and CO2). Carbonated and uncarbonated samples plot on
calculated dunite and harzburgite compositions (using Fo90 and En90 compositions for olivine and orthopyroxene, respec-
tively). Note that the MgO and SiO2 contents of serpentinite and listwanite are indistinguishable. * Denotes an antigorite-rich
rock that contains 7 wt% magnetite (estimated from the data in the appendix) created during the complete serpentinization of
(Fo90 + En90)-rich rocks.

100% excluding H2O and CO2 (Fig. 5b). The variation Whole-rock geochemical compositions were tested
of SiO2:MgO ratio within the suite of rocks may be a against model sequestration-type reactions R1a, R1b, R2
function of the degree of initial serpentinization, or the and R3 using the method of Gordon (2003), a new
amount of initial orthopyroxene (Fig. 5b). geochemical mass-balance technique that quantitatively
232 THE CANADIAN MINERALOGIST

accounts for protolith heterogeneity. The six samples be obtained by simple manipulations using linear alge-
with assemblage A1 (Table 1) and the lowest CO2 con- bra. The basis contains linear combinations of elements
tent (<0.5 wt%; Appendix) were chosen to define the unaffected by the proposed mass-transfer reactions. Be-
protolith. The protolith for each carbonated rock was cause more than one reaction is involved in each group
modeled as a linear combination of these six samples so of samples, the possibility of various extents of reaction
that compositional heterogeneities in the protolith do not of R1a, R1b, R2 and R3 means that this method does not
propagate into the assessment of the carbonation. The constrain the C:H ratio of the mass-transfer vector.
remaining samples were divided into three groups. Given the composition of a carbonated rock, the
Seven samples without talc or quartz were considered mass-transfer vector must have an origin that can be
to have experienced only reactions R1a and R1b, four described by a linear combination of the vectors in
samples with talc but no quartz were considered to have protolith space and be orthogonal to vectors in invariant
experienced reactions R1a, R1b and R2, and the remain- space. This approach provides a vector equation, the
ing six quartz-bearing samples were considered to have least-squares solution of which provides a “best fit”
experienced the full suite of reactions. composition of the protolith and “best fit” mass-trans-
Reaction R1a requires that the model mass-transfer fer vector for each carbonated rock.
vector for the seven low-CO2 samples be zero for all The success of the model sequestration-type reac-
species except C and H. For these samples, the model tions in predicting the observed composition of a modi-
ratio of C added to H added is 20:62. Model reactions fied rock can be judged by comparing the residuals of
R1b, R2 and R3 have the ratios of C added to H added of the least-squares solution with the errors permitted by
1:–2, 1:–2 and 3:–2, respectively. A linear algebraic analytical uncertainties determined from replicate analy-
basis (Strang 1993) for the vector space orthogonal to ses. These results are shown in Figure 6. Positive re-
the model mass-transfer vectors (invariant space) can siduals indicate that the altered rock contains more of

FIG. 6. Ratio of residual error to standard error in whole-rock chemical composition of

carbonated samples. Residual error is the difference between compositions predicted by
reactions R1–R3 and measured compositions. Values of the ratio that exceed three stand-
ard errors (positive or negative) indicate an excess or deficiency of the species that
cannot be accounted for by the model sequestration-type reactions (R1a&b, R2 and R3).
In Table B of the Appendix, we list all residuals.
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 233

that element than permitted by the hypothesis. In gen- of the formation of magnetite during serpentinization
eral, the isochemical reaction model provides a good fit of harzburgite and dunite (e.g., Toft et al. 1990). There
to the data. Residuals for Si, Ti, Fe, Mg, Na, K, P, and is also evidence for magnetite generation during reac-
V return a good to excellent fit, generally within three tion R1 (Fig. 3a). Generally, magnetite forms rim over-
standard errors (see Appendix). The variation in the growths on chromite and as disseminated grains aligned
composition of these elements in listwanite can be fully in foliation planes and in fractures. Magnetite was sub-
explained by the variability of the protolith rocks and sequently destroyed during carbonation (Table 1) and
the dilution effect of mass addition. The model protolith is completely consumed by the final stages of reaction
subspace and simple sequestration-type reactions do not R2.
adequately describe the variation in Al, Mn, Ca, Cr, Ni Mass-balance calculations indicate that Fe is con-
and Zn, which all show large residuals. The failure of served during carbonation of serpentinite. Iron liberated
the model to explain the compositional variation in these by magnetite destruction must therefore be hosted
elements is most likely due to inadequate characteriza- within another mineral. The increase in Fe content of
tion of protolith variability. The residuals are generally magnesite in M2 is consistent with conservation of Fe
both positive and negative and show no systematic during magnetite destruction. There is a corresponding
variation with degree of carbonation. Moreover, Al, Cr, decrease in whole-rock magnetic susceptibility in
Ni and Zn are sparingly mobile in hydrothermal sys- samples recording reaction R2 (Fig. 7).
tems. There is no convincing evidence for chemical We have exploited the correlation of whole-rock
mobility beyond the volatile species, although a more CO2 content and magnetic susceptibility (Fig. 7) to de-
thorough examination of the protolith is required to ex- velop a semiquantitative measure of reaction progress
clude mobility of Al, Mn, Ca, Cr, Ni and Zn. that can be employed in the field. Magnetic susceptibil-
ity of outcrops and hand specimens was measured with
MAGNETIC SUSCEPTIBILITY an Exploranium KT–9 Kappameter. Although fully car-
bonated rocks contain >35 wt% CO2, virtually all mag-
Carbonation according to reaction R1 occurs many netic susceptibility appears to have been destroyed by
tens of meters from major fracture systems but is gener- about 20 wt% CO2, which corresponds to the comple-
ally not evident in the field. We have been unable to tion of reaction R2. The magnetic susceptibility of
map the distribution of magnesite formed after olivine serpentinite is heterogeneous and reflects the degree of
breakdown, for example, because this reaction is com- initial serpentinization. A least-squares fit to the data
monly not discernable in hand sample. The magnetic yields a semiquantitative relationship between magnetic
susceptibility of serpentinite is relatively high because susceptibility and CO2, and thus reaction progress:

FIG. 7. Magnetic susceptibility versus whole-rock wt% CO2. A linear fit was constrained
to intersect zero magnetic susceptibility at 20 wt% CO2, which approximately marks
the maximum CO2 content of rocks that record complete progress of reaction R2.
234 THE CANADIAN MINERALOGIST

Magnetic Susceptibility (10–3 S.I. units) = content calculated from magnetic susceptibility and Eq1
61.003 – 3.0544 (wt% CO2) (Eq1). (Fig. 9). Calculated CO2 content increases continuously
within the zone of rusty-weathering serpentinite, imply-
To test the utility of this relationship in mapping reac- ing a gradation in reaction progress across this zone.
tion progress, a 2 by 2 meter pavement of variably car- However, whole-rock geochemistry through the section
bonated serpentinite was mapped and analyzed in the indicates an abrupt change in CO2 content (Fig. 9). The
field for magnetic susceptibility (approximately 1550 gradation in magnetic susceptibility at the reaction front
analyses; Fig. 8). Fracture-controlled talc + magnesite could either reflect the resolution of the magnetic sus-
+ minor serpentine (R2) intersects the pavements in two ceptibility meter or the progressive destruction of mag-
discrete zones of reaction that are flanked by distal netite not directly linked to progress of carbonation
antigorite + magnesite assemblages (A2). The mineral- reaction. Regardless, magnetic susceptibility maps pro-
ogical zonation within the pavement corresponds to dis- vide semiquantitative estimates of reaction progress in
crete changes in magnetic susceptibility. The distinction the field and are invaluable in discerning the geometry
between serpentinite and rusty serpentinite in the field of reaction fronts associated with reaction R2.
is very subtle, yet this alteration is prominent in mag-
netic susceptibility. Measured whole-rock CO2 content
of samples along transect A–B is within 5% of the CO2

FIG. 8. a) Composite photograph of a 2 by 2 meter pavement-type outcrop on the western slope of Monarch Mountain (E
575887, N 6602098, WGS 84). b) Detailed geological map of the listwanite zone mapped at 1:20 scale. Sample locations and
section A–B correspond to those of Figure 9. c) Magnetic susceptibility map, made from ca. 1550 measurements, showing the
correlation of magnetic susceptibility with mineral content. d) Whole-rock wt% CO2 map calculated using Eq1.
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 235

VOLUME STRAIN ACCOMPANYING REACTION and olivine to magnesite plus serpentine accounts for
only about 5–15% of the carbonation potential for the
The carbonation reactions involve an increase in the serpentinite at Atlin. However, because products of re-
volume of solids (Table 2); each reaction therefore has action R1 are widespread, they may have sequestered a
the potential to create or destroy porosity and perme- significant portion of the total CO2 content of the
ability. The development of carbonate veins orthogonal listwanite system at Atlin.
to the outer limit of talc plus magnesite reaction in Fig- The carbonation of antigorite to magnesite and talc
ures 8a and 8b likely developed in an environment of binds large quantities of CO2 with a small associated
tension. The mechanical model of Jamtveit et al. (2000) gain in the volume of solid material. Moreover, within
predicts the development of high-permeability zones intact bedrock, the progress of this reaction may create
downstream of reactions that produce a net increase in a permeability front in advance of the reaction front,
solid volume (Fig. 10). Indeed, the pattern of carbon- further promoting fluid infiltration and reaction. Reac-
ation and vein formation in Figures 8a and 8b resembles tions R1 and R2, which combined account for about half
the patterns produced by the Jamtveit et al. model. If of the carbonation potential of serpentinite (Table 1),
vein formation outboard of the talc–magnesite reaction may hold the greatest promise for in situ carbonation of
front records tension and mechanical coupling of swell- minerals. Complete carbonation of serpentinite to mag-
ing during reaction R2, then the volume strain of the two nesite plus quartz is generally limited to the fractured
processes should balance, at least approximately. Vol- cores of the largest listwanite systems. Products of re-
ume strain was calculated by measuring vein thickness action R3 generally are not developed far into intact
(7.6 ± 3.8 mm) and density in the rusty serpentinite zone bedrock, which may have been incapable of accommo-
of Figure 8. The calculated increase in volume of 3.6 ± dating the large increase in solid volume associated with
1.8% compares well with the change in solid volume this reaction. Carbonation to magnesite plus quartz thus
for reaction R2 (2.3% in a serpentinite rock with 5% may limit the sequestration capacity of in situ mineral-
relict olivine). The distribution of tension gashes appears carbonation systems by destroying porosity and perme-
to control the outer extent of formation of rusty ability in the vicinity of injection sites. The reaction-path
serpentinite alteration and thus likely has enhanced per- models of Cipolli et al. (2004) predict significant con-
colation of reactive fluid into intact wallrock. version of serpentine within in situ mineral-carbonation
systems to carbonate and “chalcedony” (microcrystal-
IMPLICATIONS FOR CO2 SEQUESTRATION line quartz) within a few tens of years. Their model
assumes direct transformation of serpentine to “chalce-
Direct carbonation of olivine (± brucite) was previ- dony” and magnesite at 250 bars and 60°C on the basis
ously unrecognized at Atlin and records infiltration and of conditions of the Gruppo di Voltri serpentinite aqui-
carbonation of intact bedrock many tens to hundreds of fer at Genova. Our investigations at Atlin and Figure 11
meters from the primary system of fracture-controlled suggest that carbonation within the subsurface under
permeability. The relative ease with which fluid infil- these conditions would more likely proceed via a series
trated intact bedrock may be due to the inherently high of reactions (R1 to R3). Moreover, as indicated above,
reactivity of olivine and brucite and to the small increase
in solid volume associated with carbonation of small
amounts of brucite and relict olivine. The extent of re-
action R1 indicates that the primary permeability of
serpentinite at Atlin was sufficient for fluid infiltration
and progress of R1. However, the large increase in solid
volume of R1 may limit its usefulness in olivine-rich
bedrock. From Table 1, the direct carbonation of brucite

FIG. 9. Measured and calculated wt% CO2 across A–B from

Figure 8b. Wt% CO2 is calculated using Eq1 and data on
magnetic susceptibility (Fig. 8c). The range in calculated
CO2 content reflects the full range of four measurements of
magnetic susceptibility at each location.
236 THE CANADIAN MINERALOGIST

reactions R1 and R2 involve a relatively small increase

in the volume of solids, which will limit the destruction
of porosity, yet they still sequester half the CO2 as com-
plete carbonation to quartz plus magnesite. The stabil-
ity of carbonation reactions are controlled by the activity
of CO2 in the fluid phase; Figure 11 indicates that the
fluid chemistry favorable for driving R1 and R2 is a CO2-
rich aqueous fluid. Industrial processes of mineral car-
bonation could therefore be engineered to control the
input gas composition so as to preferentially drive car-
bonation reactions that minimize porosity loss and maxi-
mize permeability generation in the subsurface.

ACKNOWLEDGEMENTS

This research was funded by a Natural Sciences and

Engineering Council of Canada Discovery Grant, by the
Oil, Gas and Energy Branch of Environment Canada,
and by the Innovative Research Initiative for Green-
house Gas Mitigation, a program under the Climate
Change Action Plan 2000, and administered under En-
vironment Canada and Natural Resources Canada, Earth
FIG. 11. Mineral stability in the system MgO–SiO2–CO2–
Sciences Sector. It was conducted under the auspices of H2O as a function of activity of H2O and CO2 in the fluid
Activity 5.3a, “CO2 storage by mineral carbonation re- phase, calculated at the Gruppo di Voltri serpentinite aqui-
actions: kinetic and mechanical insights from natural fer at Genova, Italy (P = 250 bars, T = 60°C; Cipolli et al.
analogues” under the Earth Sciences Sector Climate 2004). The calculations were made using PTAX and the
Change Program, Project CC4500, entitled “Monitor- mineral database of Berman (1988).
ing methods and assessment of carbon sequestration
over Canada’s landmass”. We thank Bill Reynen, Envi-

ronment Canada for encouraging the initiation of this

project, and Mitch Mihalynuk, British Columbia Geo-
logical Survey Branch, for his help and expertise at
Atlin, including logistics, data sharing, and aerial pho-
tography. We have benefitted from many discussions
on and off outcrop with Bob Anderson of the Geologi-
cal Survey of Canada at Vancouver. Carmel Lowe of
the Pacific Geoscience Centre first suggested using
magnetic susceptibility to map alteration. Kiyoko
Kakano provided invaluable field and laboratory assis-
tance, and Sasha Wilson performed Raman analysis. We
thank A.E. Williams-Jones, L.P. Baumgartner, D.R.M.
Pattison and R.F. Martin for their insights and comments
during the review process. We would also like to give
special thanks to Dugald Carmichael for his invaluable
contributions to the science of metamorphic petrology.
Congratulations on your retirement, Dugald!

REFERENCES
AITKEN, J. D. (1959): Atlin map area, British Columbia. Geol.
Surv. Can., Mem. 307.

ANDREW, K. (1985): Fluid Inclusion and Chemical Studies of

FIG. 10. Distribution of fracture permeability in advance of a Gold–Quartz Veins in the Atlin Camp, Northwestern Brit-
reaction, predicted by the mechanical model of Jamtveit et ish Columbia. B.Sc. thesis, Univ. of British Columbia,
al. (2000). In the model, tension-fracture permeability re- Vancouver, B.C.
sults from an overall increase in solid volume during reac-
tion. Modified from Jamtveit et al. (2000).
 CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 237

ASH, C.H. (1994): Origin and tectonic setting of ophiolitic G UTHRIE , G.D., C AREY , J.W., B ERGFELD , D., B YER , D.,
ultramafic and related rocks in the Atlin area, British Co- C HIPERA , S., Z IOCK , H. & L ACKNER , K.S. (2001):
lumbia (NTS 104N). B.C. Ministry of Energy, Mines and Geochemical aspects of the carbonation of magnesium sili-
Petroleum Resources, Bull. 94. cates in an aqueous medium. Proc. First National Confer-
ence on Carbon Sequestration (Washington), Sess. 6C,
________ (2001): Relationship between ophiolites and gold– 14 p.
quartz veins in the North American Cordillera. B.C. Dep.
of Energy, Mines and Petroleum Resources, Bull. 108. HANSEN, L.D., DIPPLE, G.M. & ANDERSON, R.G. (2003): Car-
bonate altered serpentinites of Atlin, BC: a two stepped
________ & ARKSEY, R.L. (1990a): The Atlin Ultramafic analogue to CO2 sequestration. Geol. Soc. Am., Abstr. Pro-
Allochthon: ophiolite basement within the Cache Creek grams 35(5), 133-14, 329.
Terrane; tectonic and metallogenic significance (104N/12).
In Geological Fieldwork 1989. B.C. Department of Energy ________, ________, ________ & NAKANO, K.F. (2004): Geo-
and Mines, Pap. 1990–1, 365-374. logic setting of carbonate metasomatised serpentinite
(listwanite) at Atlin, British Columbia: implications for
________ & ________ (1990b): The listwanite – lode gold CO2 sequestration and lode-gold mineralization. Geol.
association in British Columbia. In Geological Fieldwork Surv. Can., Current Research (in press).
1989. B.C. Department of Energy and Mines, Pap. 1990–
1, 359-364. HOLLAND, T & POWELL, R. (1991): A compensated-Redlich–
Kwong (CORK) equation for volumes and fugacities of
________, MACDONALD, R.W. & ARKSEY, R.L. (1991): To- CO2 and H2O in the range 1 bar to 50 kbar and 100–
wards a deposit model for ophiolite related mesothermal 1600°C. Contrib. Mineral. Petrol. 109, 265-273.
gold in British Columbia. In Geological Fieldwork 1991.
B.C. Department of Energy and Mines, Pap. 1992–1, 253- J AMTVEIT, B., AUSTRHEIM, H. & M ALTHE-SORENSSEN , A.
260. (2000): Accelerated hydration of the Earth’s deep crust in-
duced by stress perturbations. Nature 408, 75-78.
BERMAN, R.G. (1988): Internally-consistent thermodynamic
data for minerals in the system: Na2O–K2O–CaO–FeO– KASHKAI, A.M. & ALLAKHVERDIEV, I. (1965): Listwanites:
Fe2O3–Al2O3–SiO2–TiO2–H2O–CO2. J. Petrol. 29, 445- their Origin and Classification. Akad. Nauk AZ SSR, Inst.
522. Geol., Baku. Russia (in Russ.).

BLOODGOOD, M.A., REES, C.J. & LEFEBURE, D.V. (1989): Ge- LACKNER, K.S., WENDT, C. H., BUTT, D.P., JOYCE, E.L., JR. &
ology and mineralization of the Atlin area, northwestern SHARP, D.H. (1995): Carbon dioxide disposal in carbonate
British Columbia (104N/11W and 12E). In Geological minerals. Energy 20, 1153-1170.
Fieldwork 1988. B.C. Department of Energy and Mines,
Pap. 1989–1, 311-322. MADU, B.E., NESBITT, B.E. & MUEHLENBACHS, K. (1990): A
mesothermal gold – stibnite – quartz vein occurrence in the
BUISSON, G. & LEBLANC, M. (1985): Gold in carbonatized Canadian Cordillera. Econ. Geol. 85, 1260-1268.
ultramafic rocks from ophiolite complexes. Econ. Geol. 80,
2028-2029. MATTER, J.M., TAKAHASHI, T., GOLDBURG, D., MORIN, R.H. &
STUTE, M. (2002): Secure, long-term geological carbon
CARMICHAEL, D.M. (1970): Intersecting isograds in the Whet- sequestration in mafic rocks: results from field and labora-
stone Lake area, Ontario. J. Petrol. 11, 147-181. tory experiments. Geol. Soc. Am., Abstr. Programs 34(5),
305.
CIPOLLI, G., GAMBARDELLA, B., MARINI, L., OTTONELLO, G. &
ZUCCOLINI, M.V. (2004): Geochemistry of high-pH waters MIHALYNUK, M.G., SMITH, M., GABITES, J.E., RUNKLE, D. &
from serpentinites of the Gruppo di Voltri (Genova, Italy) LEFEBURE, D. (1992): Age of emplacement and basement
and reaction path modeling of CO 2 sequestration in character of the Cache Creek terrane as constrained by new
serpentinite aquifers. Appl. Geochem. 19, 787-802. isotopic and geochemical data. Can. J. Earth Sci. 29, 2463-
2477.
GOFF, F. & LACKNER, K.S. (1998): Carbon dioxide sequester-
ing using ultramafic rocks. Environ. Geosci. 5(3), 89-101. MONGER, J.W.H. (1975): Upper Paleozoic rocks of the Atlin
terrane, northwestern British Columbia and south-central
GOLDBERG, P., CHEN, Z.Y., O’CONNOR, W.K., WALTERS, R.P. Yukon. Geol. Surv. Can., Pap. 74–47.
& ZIOCK, H. (2001): CO2 mineral sequestration studies in
US. Proc. First National Conference on Carbon Seques- ________ (1977a): Upper Paleozoic rocks of northwestern
tration (Washington), Sess. 6C, 10 p. British Columbia. In Current Research, Part A. Geol. Surv.
Can., Pap. 77–1A, 255-262.
GORDON, T.M. (2003): Algebraic generalization of the graphi-
cal Gresen and Pearce methods for identification of ________ (1977b): Upper Paleozoic rocks of the western Ca-
geochemical mass-transfer processes. Geol. Assoc. Can. – nadian Cordillera and their bearing on the Cordilleran evo-
Mineral. Assoc. Can. – Soc. Econ. Geol., Program Abstr. lution. Can. J. Earth Sci. 14, 1832-1859.
(CD-ROM) 28, 146 [ISSN 0701–8738, ISBN 0–919216–
86–2].
238 THE CANADIAN MINERALOGIST

________, RICHARDS, T.A. & PATERSON, I.A. (1978): The hin- SEIFRITZ, W. (1990): CO2 disposal by means of silicates. Na-
terland belt of the Canadian Cordillera: new data from ture 345, 486.
northern and central British Columbia. Can. J. Earth Sci.
15, 823-830. SHERLOCK, R.L., LOGAN, M.A.V. & JOWETT, E.C. (1993): Silica
carbonate alteration of serpentinite, implications for the
O’CONNOR, W.K., DAHLIN, D.C., NILSEN, D.N., RUSH, G.E., association of precious metal and mercury mineralization
WALTERS, R.P. & TURNER, P.C. (2001): Carbon dioxide in the Coast Ranges. Soc. Econ. Geol., Guidebook Ser. 16,
sequestration by direct mineral carbonation: results from 90-116.
studies and current status. Proc. First National Conference
on Carbon Sequestration (Washington), Sess. 6C, 10 p. STRANG, G. (1993): Introduction to Linear Algebra. Wellesley-
Cambridge Press, Wellesley, Massachusetts.
O’HANLEY, D.S. & WICKS, F. J. (1995): Conditions of forma-
tion of lizardite, chrysotile and antigorite, Cassiar, British TOFT, P.B., ARKANI-HAMED, J. & HAGGERTY, S.E. (1990): The
Columbia. Can. Mineral. 33, 753-773. effects of serpentinization on density and magnetic suscep-
tibility: a petrophysical model. Phys. Earth Planet. Interi-
RINAUDO, C., GASTALDI, D. & BELLUSO, E. (2003): Characteri- ors 65, 137-157.
zation of chrysotile, antigorite and lizardite by FT-Raman
spectroscopy. Can. Mineral. 41, 883-890. WITTKOPP, R.W. (1983): Hypothesis for the localization of gold
in quartz veins, Allegheny district. California Geol. 36(6),
SCHANDL, E.S. & NALDRETT, A.J. (1992): CO2 metasomatism 123-127.
of serpentinites, south of Timmins, Ontario. Can. Mineral.
30, 93-108. WU, J., SHEEN, J., CHEN, S. & FAN, Y. (2001): Feasibility of
CO2 fixation via artificial rock weathering. Ind. Eng. Chem.
________ & W ICKS , F.J. (1991): Two stages of CO 2 Res. 40, 3902-3905.
metasomatism at the Munro mine, Munro Township, On-
tario: evidence from fluid-inclusion, stable-isotope, and ZEVENHOVEN, R. & KOHLMANN, J. (2001): CO2 sequestration
mineralogical studies. Can. J. Earth Sci. 28, 721-728. by magnesium silicate mineral carbonation in Finland. Sec-
ond Nordic Minisymposium on Carbon Dioxide Capture
and Storage (Gotenborg): http://www.entek.chalmers.se/
~anly/symp/symp2001.html.

Received December 8, 2003, revised manuscript accepted

September 25, 2004.

APPENDIX
Bulk geochemical samples were crushed and powdered in a tungsten carbide shatterbox. Powders were sent to
the Geochemical Laboratories at McGill University, Montreal, Quebec, for bulk analysis by X-ray fluorescence.
Results are included in Table A. Samples were examined for major, minor and trace elements, CO2 content by
induction furnace, and total volatile content by loss on ignition (LOI). Standard errors used in calculations were
determined from seven blind replicate analysis of one sample. In Table B, we present the ratio of residuals to one
standard error for data illustrated in Figure 6.
CARBONATED SERPENTINITE (LISTWANITE) AT ATLIN, B.C. 239