American Mineralogist, Volume 72, pages 566-579, 1987
Experimental study of hydrogen-isotopeexchangebetween
aluminous chlorite and water and of hydrogen
diffusion in chlorite
Cor,rN M. Gnq,HA.IvI
Department of Geology, Edinburgh University, Edinburgh, EH9 3JW, Scotland
J.lNnr A. Vrcr,rNo, Russrr-r. S. Hlnwrox
Department of Geological Sciences,Southern Methodist University, Dallas, Texas 75275,U.5.4.
Ansrucr
Hydrogen-isotope exchangebetween Mg-Al chlorite of sheridanite composition and
water has been studied experimentally over the temperature range 300-700'C. Problems
were encounteredwith changesin the bulk D/H ratios of capsulecontentsduring high P-T
experiments. Below 500t, exchangewas so slow that equilibrium fractionation factors
could not be determined. Over the temperature range 500-700'C, no temperature dependenceis observed,with 1000 ln a;n,-"roaround -28. Chlorite-water fractionations are not
simply related to octahedral cation composition; preferential fractionation of the light
isotope into chlorite reflectsthe presenceof hydrogen bonding in the chlorite structure.
A compilation of experimentally and empirically determined chlorite-water hydrogenisotope fractionation factors over a wide rangeof temperaturesindicatesthat theseare not
measurablydependenton Fe/Mg in chlorite.
Hydrogen-isotope exchangereactions proceededby volume diffusion of hydrogen in
chlorite, for which activation energiesin the range 166-172 kJ/mol H are much higher
than in any other hydrous mineral yet studied. Hydrogen-isotopeexchangebetweenchlorite and water in slowly cooling metamorphic rocks may nonetheless proceed to
subgreenschistfacies temperatures.
Hydrogen-isotopeexchangeduring hydrous metamorphism of the oceanic crust in the
temperature range 200-500"C may be dominated by fractionation factors in the range
-30 to -40.
1000ln a"D,,r1*",",:
INrnonuctIoN
In this paper, we report the results and applications of
in
variations
an
experimental study of hydrogen-isotopefractionation
ofhydrogenand
oxygen-isotope
Studies
rocks and minerals provide a powerful method of docu- between aluminous chlorite and water and of rates for
menting fluid-rock interactions in the Earth's crust (Tay- hydrogen diffirsion in chlorite, at high pressures(2-5 kbar)
lor, 1974; Sheppard, 1977). The hydrogen-isotopecom- and over a tange oftemperatures (300-700oC)represenpositions of hydrous minerals are particularly sensitive tative of a spectrumof metamorphic conditions and some
indicators of the isotope composition of the last fluid with hydrothermal environments. Aluminous chlorites are
which they have equilibrated, because(1) most hydrous common and abundant constituents of a wide range of
minerals contain relatively small amounts of hydrogen crustal rock types crystallizing in these hydrous environrelative to infiltrating hydrous fluids unlessfluid/rock ra- ments. Chlorites, with about 12 wto/owater, contain the
tios were very small, (2) waters of different origins in the majority of structural water present in many low-grade
Earth's crust and hydrospherehave distinctive hydrogen- metamorphic and hydrothermal rocks, most importantly
isotope composition, and (3) hydrogen-isotopeexchange in hydrated mafic oceanic crust. Therefore the results of
betweenhydrous minerals and water should be relatively the study should provide new data applicable to the defast during mineral-water interactions and crystallization termination of the stableisotopecompositionsand origins
of hydrous minerals. In order to calculate the isotope of hydrous fluids in a variety of crustal geologicenvironcomposition of a fluid phase that has subsequentlybeen ments.
Preliminary resultsof this experimentalstudy have been
removedfrom a water-rocksystem,it is necessaryto know
the isotope fractionation factors between hydrous min- reported in abstract form by Viglino et al. (1984) and
erals and water as a function of temperatureand mineral Graham et al. (1984b). No other experimental data are
available on hydrogen-isotopeexchangebetweenchlorite
composition.
566
0003-o04x/87l0506--05
66$02.00
GRAHAM ET AL.: CHLORITE.WATER HYDROGEN-ISOTOPE EXCHANGE
TneLe1. Chemicalanalysesof chlorite9451S
12
sio,
Al,03
Tio,
Fe.O"
FeO
Mgo
MnO
KrO
Naro
HrO.
Total
Si
AI
Ti
Fe3*
Fe2*
Mg
Mn
K
Na
OH
29.79
25.54
0.06
0.34
32.29
0.01
12.40
10043
5.51
5.57
0.01
28.81
26.43
0.24
0.40
3121
0.35
0.14
13.89
101.47
s.23
5.66
0.05
8.90
0.03
0.06
8.45
15.31
0.08
0.05
16.84
567
TneLe2. lsotopecompositions
of startingmaterials
6Dt(Va)
Chlorite94515
Water 1
Water 2
Water 3
Water 4
1F(a' -
-33.5
6.4
3.4
-26.5
-50.8
1)
-39.6
-36.8
-7.2
18.2
Note.'aiis the initial(uncorrected)chlorite-waterfrac.
tionation factor.
hydrogen diffusion in other hydrous minerals (Graham,
1981; Graham etal.,1984a) and may be used to calculate
closure temperatures for cessation of hydrogen-isotope
exchange between chlorite and water in slow-cooling geologic environments.
SranrrNc
MATERTALS
Note. Columnsare (1) Jones (1981).X-ray fluorescence and colorimetry analysis.Water content determined by weight loss to 100eC during DrA experiments.(2) Shannonand Wherry(1922).Wet-chemical
analysis.
t HrO content of
starting material,determinedby inductionheatingin this study, is 12.95%.
The chlorite usedin this study is a sheridanite(specimennumber 94515, Smithsonian Institution Museum of Natural History,
Washington,D.C.) whosecomposition is shown in Table l. This
chlorite is comparably aluminous to typical chlorites in lowgrade aluminous metamorphic and hydrothermal rocks (Velde
and Rumble, 1977). Optical, xro, and serr.r
examination showed
the sample to be free of contamination.
The specimenwas preparedby grinding and sieving to obtain
and water, although data on hydrogen-isotopeexchange sized starting materials with nominal grain-sizefractions of <44
betweenserpentineand water (Suzuokiand Epstein, 1976; pm (fine), 44-63 pm (medium), and 63-105 pm (coarse).The
Sakai and Tsutsumi, 1978)and brucite and water (Satake finest material was removed from the fine grain-sizefraction by
and Matsuo, 1984)have been reported.The serpentine elutriation.
The isotope compositions of the chlorite and the waters used
and brucite data are relevant to this study becauseboth
as starting materials in this study are given in Table 2. The
serpentine and chlorite are sheet silicates, with brucite
isotope compositions of the waters were chosen so that hydroIayersincorporated in the sheetstructure, and serpentine
gen-isotopeequilibrium between chlorite and water would be
is chemically similar to Al-free chlorite.
approachedby exchangefrom two directions during experiments
Empirical estimatesof hydrogen-isotopefractionation (cf. Suzuokiand Epstein,1976;Graham et a1.,1980).All minfactors for the chlorite-water systemas a function of tem- eral-water chargeswere sealedinto separatecapsulesas part of
perature and Fe/Mg ratio in chlorite have been made the same batch, in order to circumvent problems arising from
using isotope data for chlorites from pelitic schists(Tay- variation of starting water composition with time and to ensure
lor, 1974; Kuroda et al., 1976),hydrothermal ore depos- internal consistencyofdata (cf. Graham et a1.,1980).
its (Kuroda et al., 1976) and drill-core samples from a
recentgeothermalfield (Marumo et al., 1980).Theseempirical calibrations are usually based on assumptionsregarding the hydrogen-isotopecomposition of the water
with which the minerals have equilibrated, the temperature of crystallization of the chlorites, and the absenceof
retrograde re-equilibration of the chlorite with hydrous
fluid. These assumptions may be poorly founded, especially becausemany hydrous minerals will readity exchangehydrogen with water to temperatureswell below
thoseof primary crystallization during slow cooling (Graham, l98l). Therefore, empirical calibrations of hydrogen-isotopefractionation factors not derived from direct
experimental calibration must be treated with caution.
Experimental measurement of rates of hydrogen-isotope exchangebetween chlorite and water in this study
permits calculation of ratesof hydrogen diffusion in chlorite. Thesediffi.rsionrates may be compared with rates of
ExprnrnrnNTAr,
AND ANALyTTCAL METHoDS
Experimental and analytical methods are comparableto those
describedby Graham et d. (1980, 1984a).Exchangemns were
made in Au capsulesof 5 mm O.D. x 0.5 mm wall x 20 mm
(approx.) length. Each run typically contained 125-mg chlorite
and 16-mg water, capsulesand mineral powders having been
stored and dried at 110'C prior to loading in order to minimize
adsorption of atmospheric moisture. Runs were made in Nimonic-105 cold-sealpressurevessels(2-kbar runs) and internally
heated gas-pressurevessels(5-kbar runs) using an Ar pressure
medium. After quenching, capsuleswere weighed to check for
leakage,heated at l20t overnight under vacuum to remove
adsorbed moisture, and then pierced under vacuum to release
free water that was collected by freezing into a liquid-N, cold
trap while heating capsulesat 120'C for up to 4 h to ensure
removal of all free water. Hydrogen gas formed by passageof
this water over hot IJ was measuredmanometrically to further
ensurecomplete removal and collection of free water.
PcE
oo@b
o? a? a? d?
@@(o(o
('N:t
NNO
ttl
tl
(.rNst
@@CO
NNNN
lttl
rf('(961
orJ)*rf
*t<ft
A'
o
|ll
| |
||l
| |
||l
| |
||l
rS
||l
(oF.NitNN@
O(DFF-F.@@
NFst
66O
d-\i99"t
I l99.o
tttttlttl
tttl
\t
ll
OO("rNF:ftr)
FF(')@@OO
cddgii\"d
I lqii@
ltttllllll
rorr)o6
6
a;ddddctd
lltlllllll
ootor+
P;E
OOc?NFtS
NNO*$NO
ll*F..N
OoCON
O€NCO
nd?qq
II
<+++
rrlttttllllll
OONO
OOFF
rO(DF
@t\o)
NNAI@c)Nt
NNNO)OrF-@
tO@
Nr(o
FCOGOF.O('TO
N(oo)FN@O
? oo ? s l \ \ o ? a a \ \
rl a
6b
roro6@@(o(o
oooroto
;;dctct
F-F.t\N
cicioo
S9\aaa?a?
oooctctcto
l9
ci
\qaa':a?N
FNFAI(O(9O
+t+++t+l+l+l
@oooo@N
oN@oroN@
dNoddNci
NNNNNNOI
tlltl
c
a
(oOOrO
t\AlsfF.
FrO(')(J@
grC.)*Nrf)
ctdcjcj cjo+F-+
ONO
|
rrrrttllll
(DFON
oNNO
O@ON
@NCdrt
(9
o
I
rrrlttl
oN
-5
@ON
NFO
I ctdct
N(')
(9r
N@
@;
|
b(DO
OFO
cjild
|
t
tF
q\
qqo?
@E
o
COOrJ
r)o
.qo?o?o?
a\qqq
nqu?qqnoq
oqq c\o?
$@
OO
no$
OO(7)
NtrJtr)NtN$
Ns
Fr)S
!al
d6i tri
NoN@
@o6@F
NoNO
NFOS*
tttlllllllllllllllll
F:cdd+
rrtlttl
(OFO
OrOOrtrr(ortF
O+(t(Y)+(9$
^^Ntt
er9d'd'5
OooOb
vd'=6'tr6'd'
O@@OO@ts
cicioi Rid
trN
I
qu?u?q qqrq\
@dc'jrci6i ;d
S
I
tOFSF
\nu?\o?ra
6'd
d'6'i(
oa
6O$
oF
$F-
o?r q\a
()rf,$
ON(9N|onF
F<t
NN
(O(OONNOO)
ION
hONNNF.N
rOO
go
qqq
bOrf,
(99o
C{NNNNNOI
NN
NNN
ooooo
ooooo
@@@@@
ooooooo
ooooooo
Sr)obboro
o9
aa
\t*
000
ooo
c)tr)r)
>>OO
|!L>>O
Ll!r!>>OO
u-() Lur
c)Nrro
c)NN@t
(oro-o@N(o
orco roNc)
NOFN
d)Or(o6@
@<g--
€@6OO
OO--
rtqq?u?
OONAl
r+
O6b6
NNNNN
oooo
oooo
NtsNN
++++
o?o?qq
: F
ONON('
@NgJONOF
dctddctN+
]F
O
FAIF.ONO\f
+
tNtt
F$FO
i-x
No)cO
stO@
6id +rto;
s-
c?a?qqqcq
u?aot\q
,-^^^
d
|
Qq)
Qo
a
;dd+++oi
(DFNO
(DAlNcO
Fo)FF(o
(")
F
t\ftr,
rttllllllt
qu?qq
d'6'vv
FNFN
o
<lNN@Orr
CO$FOFNn
o
+1r
+ I
l+ I I r I + I
ololqq?qlI
6t
oNNoo
q\
| | |
()Ob
c
a<!
*o
x
F-O
dH
E
l
t
N N € J ( ' )
F F A I
F F F N N
N
- O r
GRAHAM ET AL.: CHLORITE-WATERHYDROGEN-ISOTOPE
EXCHANGE
Chlorite run products were washed with distilled water and
examined optically and by xno and snv to check for absenceof
mineral alteration and thermal decomposition.For isotope analysis, 20-30 mg ofchlorite were loaded into previously outgassed
Mo (at Southern Methodist University) or Pt (at Scottish Universities Researchand Reactor Centre) crucibles heated under
vacuum at 120'C ovemight to remove adsorbedmoisture. Structural water was releasedby induction heating for about I h, and
hydrogen yields after conversion by passageover hot U were
measuredmanometrically to checkwater contentsof chlorite for
consistencywith initial values and to ensure complete removal
of structural water.
D/H ratios in waters and chlorites were measuredon a FinneganMAr 251 mass spectrometer(at SouthernMethodist University) and a VG-Micromass 602B mass spectrometer(at Scottish Universities Researchand Reactor Centre). Values for the
relative concentrationare given relative to standardmean ocean
water (SMOW) in the familiar 6 notation (in 7o) in Tables 2 and
3 to a precision of about + l7o where
/ t^'--'
\
- 'I
dD*-on: (#:f..t.
l) x lo,
1tD/H)*o'o"o
(l)
and the equilibrium fractionation factor, a", is related to 6D by
(D/H).hr.d,"
_
o6nr_u2o:
(D/Ht*
I + (dD"hr"d,"/1000)
I + (dD_.,",/1000)
INrEnpolarroN
(2)
oF de
569
ing uncertainty of interpolation of a" at low extents of
isotopeexchangehas been noted by Graham et al. (l 980).
Uncertainties are discussedin detail below.
Graham et al. (1980) suggestedthat the interpolation
method may not necessarilybe applicable to minerals
such as chlorite and serpentinethat contain hydrogen in
more than one site, dependingupon the rate-determining
mechanism in the exchangereaction. This method was
usedby Sakaiand Tsutsumi (1978)in their experimental
study of hydrogen-isotopeexchangebetween serpentine
and water, but none of their experiments attained isotopic equilibrium so that independent verification that
are correct, or that the
the interpolated values of a!",o-1rr6
interpolation method is valid, is not possible.In this study
we find good agreementbetweenfractionation factors calculated from partial-exchangeexperiments at 600"C using the interpolation method and determined by isotope
equilibrium in longer runs at the same temperature (Table 3). This observationis consistentwith-but not conclusive proof of-the validity of the interpolation method
for the chlorites, and by inference possibly also for the
serpentlnes.
Applicability of the interpolation method to the chlorites with hydrogen in different structural sites may indicate that the rate-determining step in the process of
isotope exchangeis related to the transport of hydrogen
through the mineral structure(e.g.,Graham, 1981),rather than to the breaking and reforming of O-H bonds.
Rnsur,rs
Complete isotope exchangeat any temperatureis demonstrated by attainment of the same final fractionation
factor (ar) either in experimentsrun at the same temperature with starting waters of different isotope composition, or in experiments run with waters of the same isotope composition run for different times (a, + a,, where
a, is the initial mineral-water isotope fractionation factor). For most experiments, hydrogen-isotopeexchange
betweenchlorite and water was so slow that equilibrium
was not attained, and the interpolation method of Northrop and Clayton(1966),modified by Suzuokiand Epstein
(1976), was used to calculatea" from the relationship
(o, - l) : (a" - l) - A(a, - a1),
(3)
Results of valid experiments on hydrogen-isotopeexchangebetween chlorite and water are given in Table 3,
together with final fractionation factors (ar) and interpolated equilibrium fractionation factors (a"). Discussion of
theseresults necessitates(l) a considerationofthe experimental problems of hydrogen diffusion through capsule
walls and hydrogen-isotope exchange between capsule
contents and the external experimental environment and
(2) a consideration ofthe uncertainties arising from this
sourceand from analytical uncertainties,in order that the
validity of any experiment may be assessedand realistic
constraints placed upon the uncertainty in a".
(where a, and ai are the final and initial fractionation
factors andAis a constant related to the extent ofisotope
exchange)for two or more experiments of the same duration run with different starting waters. In practice, ar
may be obtaired graphically, algebraically, or by leastsquaresanalysisfrom the intercept 10,(d. - l), in plots
of 103(a,- l) vs. 103(a,* a,), whereI is the slopeof the
line fitted through the data for any group of experiments
with different starting waters (a,) at constant Z and time.
The theoretical kinetic basis of this approach is well
founded (Northrop and Clayton, 1966), and the internal
consistencyof experimental hydrogen-isotope exchange
data observed using this partial-exchangeinterpolation
method has been noted in several studies (Suzuoki and
Epstein,1976;Graham et al., 1980, 1984a).The increas-
Hydrogen-isotopeexchangebetweencapsulesand
pressuremedium
Hydrogen diffusion through capsulewalls and D-H exchangebetweencapsulesand pressuremedium during high
P-Zgas-media experiments may causesignificant shifts
in the bulk dD of experimental charges.Thesechangesin
DD in experiments where complete exchangewas not attained may seriouslylimit the precision with which equiIibrium fractionation factors may be constrained by the
interpolation method, and necessitatea consideration of
the mechanisms and relative rates of hydrogen-isotope
exchangebetween mineral and water and between the
mineral-water system and the external experimental environment (pressure medium, pressure vessel, furnace,
etc.).
570
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
l04
aa
a
o
aa
103
aa
o
E
aoa
102
aa
o
1 0-20
-10
aa
a
+10
+20
+30
A (massbalancevalue)
Fig. l. Plot ofmass-balancevalue (A) againstrun time (hours) to illustrate shifts in bulk 6D ofchlorite-water hydrogen-isotope
exchangeexperiments. A values calculated using Eq. 5. Filled circles indicate runs listed in Table 3. Open circles indicate runs
rejectedon basis ofanomalous hydrogen yields or inconsistent kinetics. Seediscussionin text.
did not inhibit D-H exchangebetween charge and pressure medium. In addition, unlike previous experimental
hydrogen-isotopestudies in the high-pressurelaboratories at Edinburgh, significant changesin bulk DD were in
the direction of both D enrichment and D depletion in
experimentsin both cold-sealand internally heatedpressure vessels(Fig. l). In addition to hydrogen diffirsion
through capsulewalls during high P-Texperiments, shifts
in bulk 6D of chargesmay result from gain of atmospheric
moisture or loss of water during loading and welding of
capsules,analytical error, and incomplete removal of free
water from quenchedcapsulesor ofstructural water from
chlorite. These several alternative sourcesof error may
often be identified by yields ofhydrogen that are inconsistent with starting water or chlorite contents' Indeed,
someruns showinglarge DDshifts in Figure 1 gaveanomalous hydrogen yields. However, numerous runs grvrng
satisfactoryhydrogen yields showed shifts in DD that exceedreasonableisotopic analytical error.
The related problems of controlling the 6D of charges
(4)
xDD;n,+ /DDhro : xDDln,* ydDfrro,
during high P-Z hydrogen-isotopeexchangeexperiments
where x is the number of moles of H in chlorite and y is and of interpreting and understanding the mechanisms
the number of moles of H in the water and i and f refer and processescausing DD shifts have been discussedby
to hydrogen-isotope compositions before and after ex- Graham et al. (1980, 1984a)and recentlyby Richet et al'
periments, respectively.From Equation 4, mass-balance (1986).Whateverthe sourcesof hydrogenin the Ar presvalues were calculated as
sure medium in these studies (bomb walls, filler rods,
moisture from furnace cement or atmospheric sources,
(5)
a: (xdDLr + /6Dl,o) - (xDD;n,+ y6D;,").
impurities in bottle Ar, etc.), tentative explanations to
A values for all experiments are plotted againstrun time date have focused on the existenceoflarge/", gradients
in Figure L Clearly, numerous runs suffered significant acrosscapsulewalls (Graham et al., 1980; Richet et al.,
and ofthe
changesin 6D down to temperatures as low as 400"C, l936). A detaileddiscussionofthese processes
despite the absenceof significant Fe in the chlorite and conclusionsofRichet et al. (1986)will be presentedelsedespitethe use of thick-walled Au capsules,which clearly where. However, here we make the following observa-
Graham et al. (1980) suggestedthat changesin dD of
experimental charges could be minimized at temperaturesbelow about 650"Cby useof thick-walled(0.5 mm)
Au capsules.However, the results of hydrogen-isotope
exchange experiments between amphibole and water
(Graham et al., 1984a)indicated that Fe-rich amphibole
chargesunderwent significant shifts in 6D at temperatures
lessthan 650"C owing to oxy-hornblende reactions driven by the buffering effect of bomb walls on the charges,
while Fe-poor charges showed little or no significant
changein DD in experiments up to 850"C. Also, experiments in internally heated gas vesselswere found to be
much less susceptibleto changein bulk 6D than those in
vessels(Graham et al., 1984a,
Nimonic cold-seal.pressure
Table 3). In all thesestudies(seealsoGraham et a1.,1980,
App. I), changesin bulk 6D of chargeswere always in the
direction of D enrichment.
Mass-balancecalculations for isotope exchangeexperiments in this study were calculatedfrom the relationship
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
tions and conclusions regarding DD shifts from experimental data of this and previous studies (cited above):
(l) The occurrenceof both D enrichment and D depletion
during chlorite-water exchangeexperiments in both internally heated and cold-seal pressurevessels(Table 3)
negatesthe suggestionof Richet et al. (1986) that the
pressuremedium in the internally heatedpressurevessels
in their D-H exchangeon melt-vapor systemswas much
more reducing than in our exchangeexperiments.(2) We
note that A in Figure I shows no time dependence.(3)
We note that the data on diffusivity of protium and deuterium through Pt (Ebisuzakiet al., 1968),if applicable
to other metalsof comparablestructuresuchas Au, would
causethe rapid elimination of initial, transient/", gradients betweencapsulesand pressuremedia well within the
time scaleof our shortestexchangeexperiments.This latter conclusion is quite consistent with the observed absenceof a time dependencefor A in our runs. Therefore
we believe that our chlorite + HrO chargesapproach a
situation of hydrogen-isotopeexchangeequilibrium with
hydrogen-bearingcontaminants of widely variable D/H
in the pressuremedium well within the time scaleof our
shortest experiments. Apparent absenceof DD shifts in
any experiment may be due to coincidental similarity of
6D in chargeand pressuremedium.
The rate of this capsule-pressuremedium exchange
processis much faster than the rate of exchangeof hydrogen isotopesbetweenchlorite and water in this study,
as is evident from an analysisofthe kineticsofthe latter
exchange(Table 3; see discussion below). Therefore, a
first-order correction to the experimental data for the effects of changing bulk DD during high P-T experiments
may be made by assuming that the entire shift in 6D,
calculatedas A (the mass-balancevaluesderived from Eq.
5 and listed in Table 3), may be attributed to a shift in
6Di{ro(the DD of the starting water) beforesignificant exchange between chlorite and water has been achieved.
exchangebeAccordingly, in experimentsin which 1000/o
tween chlorite and water has not been attained, modified
values of 6D;r" (DD';r" : DDi,ro- A) have been used to
calculatethe effective initial chlorite-water fractionation
factor (oi) used in interpolating the equilibrium fractionation factors (Table 3). It is interesting to note that the
equilibrium factors thus interpolated are, with one exception (runs 26 and 27), within 5-6Tmof the interpolated
values of l03ln a" calculatedusing uncorrectedvalues of
a, from Table 2.
Partial-exchangeexperimentslisted in Table 3 and used
to interpolate equilibrium fractionation factors have massbalance values (A) up to over l0 (positive or negative).
Many experiments with A values greater than + I l, and
a few with smaller A values, gave anomalous hydrogen
yields for water or chlorite and were rejected. Beyond
this, internal consistencyof results is taken as a primary
criterion of data validity; kinetic analysis of the experimental data in Table 3 provides a useful approach to
demonstratingsuchinternal consistency(Graham, 1981)
and is consideredbelow (Fig. 5).
571
Considerationof errors and uncertainties
A rigorous quantitative treatment of uncertainties in
the experimental and analytical data is problematical, as
is calculation ofrealistic uncertaintiesin the equilibrium
fractionation factors. Several steps are involved in the
derivation of eachvalid datum, eachwith some attached
uncertainty or error. Some sourcesof uncertainty (e.g.,
analytical error in hydrogen-isotope analysis) are normally distributed, whereasothers (shifts in dD of charges
by exchangewith the pressuremedium) are not. Correction of the experimental data for the latter effecthas been
considered in detail above. The analytical uncertainty,
estimatedto be + l7-, is the most tractable,and probably
the largest, source of normally distributed uncertainty.
This + l7m uncertainty gives rise to an uncertainty of
+l.4lvn in l03ln a, (= dDln,- DDf'r")for any chloritewater pair. This will be the uncertainty in l03ln a" only
when complete exchangeis attained (e.g., runs 2 and 3,
Table 3). However, it is evident from a consideration of
the graphical representationof the interpolation method
for deriving equilibrium fractionation factors from partial-exchangeexperiments (e.9., Northrop and Clayton,
1966;Suzuokiand Epstein,1976)that the uncertaintyin
l03ln a" will be greaterthan + I .41 and will increasewith
decreasingextent of exchange(/in Table 3). Taking limiting values of DD for chlorite and water consistent with
analytical uncertainty in eachrun used to interpolate values ofa", uncertainties in interpolated values ofequilibrium fractionation factors (l03ln a") were calculated(Table 3), ranging from + 7.7 at 800/oexchange(runs 22 and
23) to +3.7 at 38o/oexchange(runs 26 and' 27).
Relationship betweenfractionation factor
(a") and temperature
Exchangerates (given by the extent of exchange,/in
Table 3) in the systemchlorite-Hro were sufficiently slow
that complete exchange(f : l) was attained in only one
pair of experiments,run at 600'C for 46 d with the finegrained starting material; although both charges(runs 2
and 3 in Table 3) gained D, attainment of l00o/oexchange,as demonstratedby the samevalue of ar, obviates
the need for interpolation of a'. Runs 16 and 17 with the
medium-grained starting material at 600"C for only 7l h,
attained only 52o/oexchange,but the interpolation method gave a value of d' comparable within experimental
error. Theseresults would seemtojustify the application
of the interpolation method to the chlorites, as discussed
above.
Owing to slow exchangeand decreasingextent of exchange (/), interpolated values of a" are less well constrained at lower temperature,despitethe long run times.
Thus at 500'C, runs attained only 38-47o/oexchange(2643 d). At 300 and 400"C, exchangewas insufficient to
constrain interpolated values of a" (Table 3).
Equilibrium fractionation factors a1700,600,and 500"C
are given in Table 3 and plotted in Figure 2. Al 700 and
600"C, where there is 50-1000/oexchange,equilibrium
572
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
T"c
and octahedralcation composition (Mg:Fe:Al) applicable
: -22.4(106/
to sheetsilicates,in which 1000ln di,i...ur-szo
68X.") + 28.2, where X' is the
T2) + (2X^t 4Xve
S e r p e n t i n e- E x p
fraction of cation M in octahedral sites, but premole
o
dicted that chlorite might be deuterium depleted relative
@
C h l o r i t e- M o d e l
to fractionation factors predicted by the model because
I - 10
of hydrogen bonding in chlorite. Comparison of model
6
o
.4
versusexperimentally measuredfractionation factors for
Chlorite
o E-20 -Exp
the chlorite used in this study shows this to be the case
{\
8
(Fig. 2), the experimentally measured fractionation facB r u ci t e - E x p
-3c
-30
tors being 15 to 30V* more negative than predicted by
(
3
)
o
the model ofSuzuoki and Epstein(1976).
Serpenline-Emp
The efect ofhydrogen bonding on D-H fractionation
-40
in mineral-water systemswas studied in detail by Graham et al. (1980),who confirmed the prediction of SuChlorite-EmP
-5
zuoki and Epstein (1976) that hydrogen-bondedminerals
should strongly concentrateprotium relative to non-hyminerals. A qualitative model relating
drogen-bonded
1234567
hydrogen-bond structure and geometry to fractionation
106/T2(K)
factor and exchangerate (Graham et al., 1980) was subFig. 2. Relationships between hydrogen-isotope fractionsequentlyshown not to be supportedby more recent neuation factor (1000 ln ai,n*ur_*u,".)
and temperature[plotted as 106/ tron-diffraction studies of hydrogen in epidote.
P (K)l for minerals in the system MgO-FeO-AlO3-SiOr-HrO.
Recent studies of the chlorite structure show that hyChlorite-Exp.: this study; Chlorite-Model: calculated for clinodroxyls
occupy two structurally different sitesin the brucchlorecompositionusingthe model ofSuzuoki and Epstein(1976)
layers. The structure permits hydrogen bondite
and
talc
(seetext). Chlorite-Emp: empirical curve of Taylor (1974); Brucite-Exp.: experimental data of Satakeand Matsuo (1984); Ser- ing between hydroxyls of the brucite layer and oxygens
pentine-Exp.: experimental data of Sakai and Tsutsumi (1978); of the talc layer, whereashydrogenswithin the talc layer
Serpentine-Emp.:
empiricalcurve of Wennerand Taylor (1973). are not hydrogenbonded (Phillips et al., 1980;Joswiget
Tentative extrapolations (dashedlines) of experimental data (this
al., 1980).Long hydrogen bonds betweenthe brucite and
study) for chlorite-HrO based on fractionation factors given in
talc layers stabilize the chlorite structure and contribute
Table 4 and derived from (1) Kuroda et at. (1976); (2) Heaton significantlyto the interlayer bond energy(Bish and Giese,
and Sheppard(1977); (3) Marumo et al. (1980). Symbols for
l98l). Therefore,we might expectthe hydrogen-bonded
varying Fe/(Fe + Mg) in chlorite: filled circle : 0 to 0.09; cross :
hydroxyls
associatedwith the brucite layer to concentrate
0.20 to 0.29; diamond : 0.30 to 0.39; triangle : 0.40 to 0.49;
protium relative to the talc layer; that is, hydrogen iso:
square 0.50 to 0.69. Seealso Marumo et al. (1980,Fig. 8).
topeswill fractionate differently betweenthe brucite layer
and water than between the talc layer and water. Dehydration of chlorites occurs in two stages,with dehydrafractionation factors derived from different pairs of runs
tion of the brucite layer occurring at lower temperatures
at each temperature are in reasonable agreement. At
than dehydration of the talc layer, and isotopic analysis
500"C, equilibrium fractionation factors calculated from
of water releasedduring step-heatingof natural chlorite
runs using the fine- and medium-grained starting mateshows that water in the brucite layer is lower in AD than
< 0.5) are in close agreement (1031na' *
rials (0.4 <
water in the talc layer (Suzuoki and Epstein, 1976), as
"f
-28), but are significantly more negative than those calwould be expectedif hydrogenassociatedwith the brucite
culated from runs using the coarse-grained starting malayer in chlorite is hydrogen bonded.
terial (l03ln a' x -21;f :0.38). We place more reliance
The experimentally determined chlorite-water fracon the runs with the higher extents ofexchange and contionations (Fig. 2) are sigrrificantly more negative than
sider -28 to be the best estimate of the chlorite-water
fractionations in the systems brucite-water (Satake and
fractionation at 500'C. There is no evidence in all these Matsuo, 1984) and serpentine-water(Sakai and Tsutsudata for any measurable temperature dependence for the
m| 1978),reflectingthe hydrogenbonding in chlorite that
chlorite-water hydrogen-isotope fractionation over the
is absentin brucite and serpentine(Suzuoki and Epstein,
temperature range 500-700t,
the fractionation factor re1976; Satake and Matsuo, 1984). Comparison of the
maining close to l03ln a' : -28.
chlorite-water and brucite-water curves (Fig. 2) is also
complicated by the distribution of Al in chlorites, which
Drscussrou
is strongly partitioned into the brucite layer (Phillips et
Controls on fractionation factors
al., 1980; Joswig et al., 1980), such that in the chlorite
Suzuoki and Epstein (1976) proposeda generalrela- used in this study, betweenone-third and one-halfofthe
tionship between hydrogen-isotope fractionation factor brucite-layer octahedraare occupied by Al.
700 500
300
100
573
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
fractionationfactors
TneLe4. Chlorite-water
estimatedfrom chloriteD/Hanalvses
f
fC)
103In c&1,",*
o roo-ra9"c
r tso-t99oc
o 200-2490C
.250-299oc
x goo-aoooc
g soo-zoooc
Chlorite
Fel(Fe + Mg)
FromKurodaet al.(1976)
-42.1
200-300
0.06
-34.7
200-300
0.07
(1977)
FromHeatonandSheppard
-31.7
400
0.51
-31.7
400
o.42
-36.9
0.48
400
-32.8
0.40
370
-34.8
370
0.63
-33.8
0.28
370
FromMarumoet al.(1980)
_13.3
130
0.28
_ 16.6
0.23
140
_
1
6
.
6
160
0.36
-35.8
170
0.35
-42.2
0.43
170
-28.2
180
0.43
-39.1
200
0.43
-33.4
220
0.43
-40.3
0.43
230
-36.0
0.44
230
_40.9
240
0.51
-37.5
250
0.36
-36.4
0.39
250
-37.7
0.36
250
-31.7
0.44
250
-35.3
0.36
250
-33.5
U.JC
250
-46.7
250
0.52
_47.8
0.48
250
-37.4
250
0.34
-39.1
0.36
250
20
;
'
Exoerimental data
I
tt'
a
I
c -30
x
o
I
x
o
-.!!
!l
!
o.2
xl
-x
o
9o
a
0.4
F%"*ue
06
fractionationfactor(1000ln
Fig.3. Plot ofhydrogen-isotope
vs. Fe/(Fe+ Mg) (chlorite).Data from this study(exaEn,-*,*J
perimentaldata),and Kurodaet al. (1976),Heatonand Sheppard(1977),and Marumoet al. (1980).Empiricaldatalistedin
Table4. Dataof Marumoet al. (1980)areunderlined.
becauseof uncertainty regarding crystallization temperatures, 6D of coexisting water, and continued equilibraComparisonwith empirical chlorite-water
tion of chlorites with water to temperatureswell below
fractionations and effect of FelMg
those of crystallization of chlorite during slow cooling
Taylor (1974) calculated an empirical curve for D-H (Graham, l98l), the data in Table 4 representlow-temfractionation in the system chlorite-water (Fig. 2) using perature chlorites from geologicsituations where cooling
data for chlorites from pelitic schists.The compositions rates are relatively fast by geologic standardsand DD(HrO)
and origins of thesechlorites are not reported, but it may and temperature are quite well constrained, particularly
be surmised that they were comparably or slightly less the geothermaldrill-core data of Marumo et al. (1980) in
aluminous than the chlorite used in this study (Velde and which Z and DD (H'O) may be measureddirectly.
Rumble, 1977)but with extensive and variable replaceMarumo et al. (1980) have assertedthat there is a corand Fe/Mg (chlorite), and that
ment of Mg by Fe2+.The extrapolation of this curve to relation between a"n1-*1r6
higher temperaturelies closeto the experimentally deter- a"n,."ro"does not depend on the temperature explicitly."
mined fractionation factors (Fig. 2). On the other hand, However, the numerically smallest and largest fractiona more securelybased empirical calibration of chlorite- ations (1000 ln a) in their data set, on which their supwater hydrogen-isotope fractionation discussed below posed Fe/Mg effect is based, are from their lowest- and
suggestsdiferent fractionation factors from Taylor's cal- highest-temperaturesamples, respectively (Fig. 3). The
ibration while remaining consistentwith the experimen- effectsof temperature and Fe/Mg (chlorite) on chloritetal data.
water fractionation cannot be unambiguously separated
A limited amount of data on the D/H of natural chlo- in the data of Marumo et al. (1980), and, taking the data
rites from metamorphic (ocean floor), geothermal, and of Table 4 together with the experimental data (Figs. 2,
hydrothermal environments in which temperature,6D of 3), the above assertionof Marumo et al. is not supported.
coexisting water, and Fe/Mg (chlorite) are well-known The data in Table 4 do not include information on the
(Kuroda et al.,1976; Heaton and Sheppard,1977;Mar- octahedral Al contents of the chlorites, which could also
umo et al., 1980)allow us to assessthe effectsof Fe/Mg influence chlorite-water D-H fractionations. However,
and temperature on chlorite-water fractionations over a from thesedata we may construct a tentative and empirrangeof temperaturesdown to 100'C (Table 4). Although ical calibration of chlorite-water D-H fractionations for
the empirical calibrations must be treated with caution Z < 400'C that is broadly consistentwith the 500-700"C
574
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
experimental data for Fe-poor aluminous chlorite. These
data (Fig. 2) are only sufficient to constrain 1000 ln
a:hr-H2o
to +5Vmat best at any temperature, but allow us
to make the following tentative conclusions:(l) Any Fe/
Mg (chlorite) dependenceis within the scatterof the data
for 0 < FelMg < 0.6 and must await experimental verification. (2) The experimental data for Fe-poor sheridanite indicate that 1000 ln o;n,_rro
is about -28 over the
temperature range 500-700'C, with no temperature dependence. (3) Over the temperature range 500-200t,
fractionations show only a small temperaturedependence
at most, with 1000 ln a;n,.rrobecoming more negative
with decreasingtemperature, but lying in the range -30
to -40. (4) Below 200t, fractionationsmay becomemore
positive with decreasingtemperature,as found in the experimentally studied systemsserpentine-HrO (Sakai and
Tsutsumi, 1978),epidote-H,O,and boehmite-HrO(Graham et al., 1980).(5) The empirical calibrationin Figure
2 is shifted by 5 to l}Vm to more positive fractionations
than those deducedby Taylor (1974) using chlorites from
pelitic schists.
Satakeand Matsuo (1984)choseto assumethat hydrogen-isotopeexchangebetweenbrucite and water followed
first-order kinetics, although they were unable to demonstratethis unequivocally(cf. Graham, l98l). Secondorder kinetics give an equally good linear fit of the brucite
data to the Arrhenius relationship as do first-order kinetics.Nonetheless,it is quite likely that the brucite-water
experimental data fit a dffirent rate law than other experimental OH mineral-water exchangedata, becauseSatake and Matsuo (1984) observedthat dissolution-reprecipitation was very likely the major process of D-H
exchangerather than volume diffusion ofhydrogen in the
solid (cf. Graham, 1981).
Minerals may changein grain size during exchangeexperiments owing to mechanical breakdown or dissolution-reprecipitation;ifthis occurs,exchangerate may vary
with time. Chlorite-water exchange experiments used
small amounts of water (cf. Graham et al., 1980)in order
to minimize grain dissolution-reprecipitation.Chlorite run
products of flve exchangeexperimentswere examined by
seu in order to assessthe extent of either of these processes.Comparison of run products with starting mateIrnplications for experimental actinolite-HrO
rials (Fig. 4) indicates that no significant changesin grain
D-H fractionations
size or morphology have occurred and that chlorite D-H
Graham et al. (1984a) experimentally determined the exchangemust proceed dominantly by volume diffusion
D-H fractionation for actinolite-waterat 400"Cto be 1000 ofhydrogen in the chlorite.
: -29, using an actinolite from a metaln a;.,6o,,,.-"ro
An Arrhenius plot of second-orderconstants (log Kr)
morphosed basaltic rock. In these experiments, 55-600/o vs. 103/Z for the chlorite-water exchangedata (Fig. 5)
exchangewas attained in 59 d. However, the actinolite gives good and consistentlinear fits for the medium and
starting material contained a small amount of chlorite coarsegrain-size fractions and is a useful test and demthat could not be separated,and therefore the measured onstration of the internal consistencyof the experimental
fractionation is only approximate. In this study, much data set. Activation energiesfor hydrogen transport in
longer experiments(105 and 157 d) producednegligible chlorite are around 150 kJ (36 kcal)/mol H for the meD-H exchange between chlorite and water at 400"C. dium and coarsegrain sizes(Table 5). Rate constantsfor
Therefore, unless the chlorite intergrown with actinolite the coarsegrain size are displaced to more negative valhas a particularly small ffictive grain size, this chlorite ues of log K, than those for the medium and fine grain
in all probability underwent very limited exchangeduring sizes(Fig. 5), consistent with the conclusion that the efthe actinolite-waterexchangeexperiments.Therefore,the fective chlorite grain size has not varied during exchange
measuredactinolite-waterD-H fractionation at 400"Cmay experiments,that exchangehas proceededby volume difbe a good approximation to the equilibrium fractionation fusion of hydrogen in chlorite, and that the measured
factor.
grain sizesare the effective grain sizesfor diffusion. Calculated activation energiesare much higher than for hyKrNnrrcs oF D-H ExcHANGE
drogen exchangebetween other hydrous minerals and
The kinetics of hydrogen-isotope exchange between water(cf. Gt'aham,l98l; Graham et al., 1984a),as would
chlorite and water may be quantified by application of be predicted from the very slow exchangerates at Z s
reaction-kinetics theory (Graham, 198l). Here we as- 400"C.
sume,followingGraham(1981)andGrahametal. (1984a),
DrrrusroN oF HYDRocEN IN cHLoRrrE:
that exchangereactions may be approximated (empiriCLOSURETEMPERATURES
cally) by second-orderkinetics, so that
Calculation of diffusion coefficients
f/(r - J): K,t,
(6)
where/is the fractional approach to equilibrium (Table
3), / is the time of exchange,and K. is the second-order
rate constant. Activation energies(O) for the rate-determining step may be derived from the Arrhenius relationship
log K,:
log a - Q/2'303RT,
where c is a constant (the pre-exponentialfactor).
(7)
Closuretemperaturesfor cessationof hydrogen-isotope
exchangebetween chlorite and hydrous fluid in cooling
metamorphic and hydrothermal systems may be calculated from estimated diffusion parameters for hydrogen
diffusion in chlorite. Methods of calculating hydrogendifrrsion coefficientsfrom experirnental data on hydrogen-isotopeexchangeare describedin detail by Graham
(1981), and their application to the chlorites is briefly
outlined here.
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
575
Fig. 4. Scanningelectron microscope (snv) photographsof fine-grainedchlorite in starting material (b) and in run product (a)
of exchangeexperiment at 600t for 217 h. Scalebar is in micrometers. Photographsindicate absenceof solution-reprecipitation
or mechanicalbreakdown during exchangeexperiments.
The calculation of diffusivities involves a knowledgeof
grain geometriesin order to account for diffusional anisotropies, especially for minerals with well-developed
cleavagessuch as the sheet silicates. The mean effective
grain diameters, plate thicknesses,and aspect ratios for
the three chlorite grain-size fractions were measuredoptically and are given in Table 6. sru observationsshowed
that measured grain thicknesseswere greater than true
thicknessesowing to the difficulty of viewing the platy
grains edge-on in optical mounts, and grain thicknesses
have been adjusted accordingly to half the optically estimated thicknessesto take account of this problem and
to assessthe effect on calculated diffusion coefficientsof
varying the grain thickness.Two geometrical models are
applicableto hydrogen diffusion in platy minerals: (l) the
infinite cylinder model in which hydrogen diffuses parallel to the layersand (2) the infinite plate model in which
hydrogendiffirsesperpendicularto the layers.For a stirred
solution of limited volume, Equations5.36 and 4.43 respectively of Crank (1975) are applicable (see Graham,
1981). Correspondingvalues of log D calculatedusing
theseequationsare given in Table 3. For the plate model,
both measuredand adjusted plate thicknesseswere used;
the effect ofreducing plate thicknessby halfis to reduce
D by about half an order of magnitude.
The correct difusion model should yield similar diffusion coefrcients for the different grain sizesat any tem-
perature. However, although chlorite is a platy mineral,
the aspect ratios of the starting materials are not large
enough to allow a clear-cut choice to be made between
the two models (cf. muscovite; Giletti and Anderson,
1975;Graham, 1981),becausefor both models all data,
irrespective of grain size, may be fitted to a single linear
regressionon an Arrhenius plot (Fig. 5), with a correlation coefficient(r'z)better than 0.97 (Table 5). Although
the data may be interpreted to mean that hydrogen diffusion occursat comparablerates parallel and perpendicular to the layers, it is more likely, by analogywith muscovite (Graham, l98l) that hydrogen diffusion will occur
faster parallel to the layers(cylinder model). For this case,
diffusivities are one to two orders of magnitude faster
than for the plate model (dependingon the adopted plate
thickness).
Calculated activation energies for both models are
comparable at 172 kJ (41 kcal)/mol H and 167 kJ (40
kcal)/mol H for cylinder and plate models, respectively
(Table 5). These activation energiesare of considerable
interest, being significantly the largest for any hydrous
mineral studied to date. By comparison the activation
energiesfor hydrogendiffusion in muscovite, hornblende,
tremolite,and zoisiteare l2l (29),84 (20),71 (17),and
103 (24.5) kJ (kcal)/mol H, respectively(cylinder model;
Graham, 198I ; Graham et al., 1984a) (Fie. 5). It is possible that the large interlayer site in muscovite, for ex-
EXCHANGE
GRAHAM ET AL.: CHLORITE-WATERHYDROGEN-ISOTOPE
576
r cc)
700
600
Fig. 5a. Arrhenius plot ofsecond-order rate constants (log Kr) vs. reciprocal temperature for hydrogen-isotope exchangeexperiments between
chlorite and water. Data given in Table 3. Seetext
for discussion.
N
Y
o-
A fine grain size
O m e d i u mg r a i n s i z e
O coarse grain siz€
1Q
1.0
r o3rr(r)
r ("c)
600
Fig. 5b. Arrhenius plot of diffusion coefrcients
vs. reciprocaltemperaturefor diffusion ofhydrogen
in chlorite, using both cylinder and plate models.
Plate (l) refers to optically estimated grain thicknesses;plate (2) refersto adjustedgrain thicknesses
(seeTable 6). Muscovite data (cylinder model) from
Graham(1981)shownfor comparison.Symbolsas
in Fig. 5(a). Data given in Table 3. See text for
discussion.
o
* -rz
o
-18
line grain size
O O medium grain size
O ! coarse grain saze
A
10
1o 3/T(K)
ample, allows faster transport pathways parallel to the
layersthan are available in chlorite, where interlayer sites
are lacking. Diffusion coefficientsfor hydrogen diffusion
in chlorite, adopting the cylinder model for comparison,
are up to 3 to 4 orders of magnitude slower over the range
of temperatures of crystallization and cooling of metamorphic and hydrothermal rocks than diffi.rsion coefficients for hydrogen diffirsion in other hydrous minerals,
excludingepidote(Graham, l98l; Graham et al., 1984a).
The implications of these results for calculated closure
temperaturesfor the cessationof hydrogen diffusion in
chlorites relative to other hydrous minerals are consideredbelow.
Exchange mechanisms
Matthews et al. (1983) conductedone oxygen-isotope
exchangeexperimentusing clinochloreand water at 500t
for 3 d. This experiment achieved only ll0/o exchange.
By comparison, the medium-grained chlorite used in this
study attained 45o/oexchange of hydrogen isotopes at
500"C in 26 d. The grain size of the starting material in
the oxygen-isotope exchange study was not stated by
577
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
400
Tc('C)
300
coolinsRate too ("clmv)
$f
Fig. 6. Plot of calculated closure (blocking) temperature (71) vs. logarithm of cooling rate dT/dt (dee/m.y.) for cessationof
hydrogen-isotopeexchangebetweenchlorite and water, using both plate and cylinder models. Grain sizes(in mm) refer to diameter
of chlorite grain (cylinder model) or thickness of chlorite plate (plate model). Calculation method describedin text.
Matthews et al. (1983), but was most likely very much
finer grained than the chlorite used in this study (Matthews, pers. comm.), indicating that hydrogen diffusion
in chlorites is more rapid than oxygen diffusion. This is
consistent with the observed absenceof solution-reprecipitation in our experiments and with the observations
of Graham (1981) that hydrogendiffusion in muscovite
proceededfastestparallel to the layers and was about frve
orders of magnitude faster than oxygen diffusion, which
occured fastest perpendicular to the layers (Giletti and
Anderson, 1975).Similarly, O'Neil and Kharaka (1976)
observednegligibleoxygen-isotopeexchangebetweenclay
minerals and water in low-temperature (=350"C) exchange experiments in which significant hydrogen-isotope exchangeoccurred. These authors concluded that
hydrogen and oxygen isotopes exchanged by diferent
mechanismsin the absenceof solution-reprecipitation.
TABLE
5. Activationenergies(O)derivedfromthe Arrheniusrelationshio
Closure temperatures
Assuming that the diffusion coefficientsfor hydrogen
diftrsion in chlorite determined at 500-700t in this study
may be extrapolatedto lower temperatures(i.e., that diffusion mechanismsare not temperature dependent),clo-
(kJ/mol) (kcal)
Chlorite-waterhydrogen-isotopeexchange at 500-700'C.
-7.89
Mediumgrain size
3.70
151.1 36.1 0.99
-7.95
Coarse grain size
3.56
152.2 36.4 0.99
Hydrogendiffusion in chlorite at 500-70eC.'
-5.21
-8.97
Cylinder model
171.7 41.0
Plate model
-6.70
-8.72
(measuredthickness)
166.9 39.9
Platemodel
-7.32
(adlustedthickness)
166.5 39 I
8.70
TneLe6. Chloritegraingeometries
0.97
0.98
0.98
Note: b : 412.303R;r, is the correlationcoefficient; f is in kelvins
. O calculated by least-squaresfitting of calculatedsecond-orderrate
constants (Kr) to the equation log K,: a + q103/7). Grain geometriesare
given in Table 6.
'- O calculatedfor both the infinite-cylinderand plate models
by leastsquares fitting of calculateddiffusioncoefficientsto the equation log D:
a + 410317).
Nominalgrain size (pm)
Average grain diameter (pm)
Measured grain thickness (pm)
Adjusted grain thickness (pm)
Asoect ratio
63-105
167.6
20.6
10.3
16 . 3
44-63
109.9
16.7
8.4
13.1
<44
34.5
5.0
2.5
13.8
Note.' Average grain diameters and thicknesses measured optically.
Measuredgrain thicknessesare too large due to difficultyof viewing platy
grains edge-on. Grain thicknesses were adiusted to half the measured
values, and both sets of thicknesses were used in diffusion-coetficient
calculationsto assess the effect of varying grain thickness.
578
GRAHAM ET AL.: CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
sure temperaturesfor the cessationof hydrogen-isotope
exchangebetween chlorite and water during cooling of
metamorphic or hydrothermal rocks may be modeled as
a function of grain size and cooling rates. The closure
temperatures(2.) were calculated using the relationship
amounts during slow cooling (seealso Graham, l98l);
preservationof hydrogen-isotopeequtlibrium, on the other
hand, might indicate loss of such a fluid at an early stage
ofthe cooling process.
Pnrnolocrc APPLICATIoNS
Metamorphism
of basic rocks to chloritic "greenQ/R
(8)
T":
stones" in the oceanic crust is thought to have occurred
in environments dominated by relatively high fluid/rock
ratios, with fast cooling rates being indicated by the very
young agesof many studied samples(e.g.,Stakesand O'after Dodson (1973, Eq. 23), which is readily solved by Neil, 1982). For these circumstances,closure temperaiterating values of T.; Q is the activation energy for dif- tures for cessationofhydrogen-isotope exchangebetween
fusion of hydrogen in chlorite (Table 5), I is the diffu- chlorite and water are in excessof 250'C (Fig. 6). Hydrosional anisotropyparameter(: 27 for cylinder; : 8.65 gen-isotope fractionation factors between chlorite and
plate),R is the gasconstant(: 8.314J.mol '.deg-'), Do water at temperaturesabove 250"C should lie in the ap: -30 to -40vm (Fig.
is the pre-exponential factor in the Arrhenius relation- proximate range 1000 ln a!,,1-*u,".
ship, a is the grain-size term (radius of cylinder, half- 2). Inasmuch as chlorite is a major reservoir of hydrogen
thickness of plate), and dT/dt is the cooling rate (deg. in the oceaniccrust, the above fractionation factor should
s '), where cooling rate is assumedto be linear with l/Z
representa dominant component of the overall seawaternear 2". Calculatedclosure temperatureswill show small hydrated oceaniccrust hydrogen-isotopefractionation for
variations with varying fluid/mineral ratio, as noted by that portion of Earth's history during which sea-floor
Cole et al. (1983)for experimentaloxygenexchangedata spreadinghas operated.
(e.g.,for cylinder model, l0-20 deg variation for two orThere are currently limited data available on the dD of
ders of magnitude variation in water-chlorite ratio), but chlorites from ocean-floor, metamorphic, or hydrotherincorporation ofthis factor into the calculation of 7" re- mal-geothermal environments. Some of these data (Kuquires a more elaborateanalysisthan has been attempted roda et al., 1976:,Marumo et al., 1980;Heaton and Shepto date. Figure 6 shows the calculated relationship be- pard, 1977) have been utilized and discussedabove in
tween I. and cooling rate for various chlorite grain sizes, assessingthe effect of FelMg on chlorite-HrO D-H fracfor both plate and cylinder models.
tionation and the fractionation factors at low temperaThe following generalconclusionsmay be drawn from tures.
Figure 6. First, in slow-coolinggeologicenvironmentssuch
Surprisingly few data are available on the DDofchlorite
as those of regional metamorphism [e.g., I < log(dT/dl) <
in hydrated mafic rocks of the oceancrust and of ophiolit21,even the largestchlorite crystalsin metamorphic rocks ic rocks. Satakeand Matsuda (1979) analyzedthe wholewill be susceptible to continued hydrogen-isotope ex- rock 6D of hydrated metabasalt from the Mid-Atlantic
changewith coexisting hydrous fluid to temperaturesbe- Ridge in which chlorite (typically 24 modal percent)is the
low those of greenschist-faciesmetamorphism. Loss of dominant hydrous mineral, and thereforeDD** t 6D"r,r:
hydrous fluid will of course permit the chlorite to pre- -38 to -30V*. Separatedchlorites from Mid-Atlantic
serve a record of higher-temperature fluid interactions Ridge greenstones(-317m, -34Vm; Stakesand O'Neil,
during slow cooling (e.g.,Graham, l98l).
1982)lie within this rangeof values.At the likely closure
The much faster cooling rates likely to characterizehy- temperaturesfor cessationof D-H exchangeduring cooldrothermal greenschistmetamorphismof the oceaniccrust ing of hydrated ocean crust, water in equilibrium with
or contact metamorphism of mafic rocks may be suffi- theserocks should be closeto dDro : 07- (using fractioncient to lead to closure temperaturesfor cessationofhyation factors estimated from Fig. 2), consistent with a
drogen-isotopeexchangebetweenchlorite and water that seawaterong1n.
are within the greenschistfaciesand not far removed from
Similarly, little data exist on the dD of chlorites and
temperaturesof chlorite crystallization. This conclusion coexisting hydrous minerals in metamorphic rocks. Rye
will, however, be dependenton chlorite grain size and on et al. (1976) studied the stable isotope composition of
the diffusion model adopted (Fig. 6).
coexisting hydrous minerals in high-pressuremetamorFinally, computed closure temperatures for chlorites phic rocks of Naxos (Greece).In a low-grade schist crysare signiflcantly higher than those for other hydrous min- tallized at about 480'C, chlorite of dD : -55V* coexists
erals(cf. Graham, 1981;Graham et al., 1984a),as would with white mica, consistent with available experimental
be expectedfrom the high activation energiesfor hydro- and empirical fractionation data (this study Suzuoki and
gen diffusion in chlorites (Fig. 5). The variations in clo- Epstein, 1976), and the calculated dD'ro for the metamorphic water is in the range -25 to -15Vm.
sure temperaturesand activation energiesamong diferent hydrous minerals will mean that mineral-mineral
Magaritz and Taylor (1976) reported dD.hrand DD**
hydrogen-isotopeequilibrium is unlikely to be preserved for chlorite-rich samplesin a wide rangeof metamorphic
if a hydrogen-bearing fluid is present even in small rocks of the FranciscanFormation of California. Except
GRAHAM ET AL. : CHLORITE-WATER HYDROGEN-ISOTOPE EXCHANGE
579
Marumo, K., Nagasawa,K, and Kuroda, Y. (1980)Mineralogyand hydrogen isotope geochemistryofclay minerals in the Ohnuma geothermal area, northeastemJapan. Earth and PlanetaryScienceLetters, 47,
255-262
Matthews, A , Goldsmith, J R., and Clayton, R N. (1983) Oxygenisotope
fractionation betweenzoisite and water. Geochimica et Cosmochimica
Acta, 47, 645-654
Northrop, D.A , and Clayton, R.N. (1966)Oxygenisotopefractionation
in systemscontainingdolomite.JournalofGeology,74, 174-176
Acrlqowr-rncMENTS
O'Neil, J.R., and Kharaka, Y.K (1976) Hydrogen and oxygen isotope
This work was supportedby NATO Grant 0551/82 (to CM.G and
exchangereactions between clay minerals and water Geochimica et
R.S.H.)and NSF Grant EAR 82-18380(o R.SH). The Experimental
CosmochimicaActa. 40. 241-246.
PetrologyUnit in Edinburgh and isotope geologyfacilities at the Scottish Phillips,T.L., Loveless,J.K., and Bailey,S.W. (1980)Cr-3 coordination
Universities Researchand Reactor Centre are supponed by NERC. Disin chlorites: A structural study of ten chromian chlorites. American
cussionswith Simon Sheppard,Alan Matthews, and Gawen Jenkin and
Mineralogis6
t ,5 , l l 2 - 1 2 2
helpful reviews from John Valley and J. Rice are acknowledged.
Richet, P , Roux, J., and Pineau,F. (l 986) Hydrogenisotopefractionation
in the systemHroJiquid-NaAlSi,Or: New data and comments on D/H
fractionation in hydrothermal experiments Earth and Planetary SciRrrnnnNcns
enceLetters,78, I l5-120.
Bish, D.L., and Giese,R.F. (1981) Interlayerbonding in IIb chlorite.
Rye.R O , Schuiling,R.D., Rye, D M., and Jansen,J.B.H.(1976)Carbon,
AmericanMineralogist,66, 12| 6-1220
hydrogenand oxygenisotopestudiesofthe regionalmetamorphic comCole, DR., Ohmoto, H., and Lasaga,A.C. (1983) Isotopicexchangein
plex at Naxos,Greece.Geochimicaet CosmochimicaActa, 40, l03lmineral-fluid systems I. Theoretical evaluation ofoxygen isotopic exI 049
changeaccompanyingsurfacereactions and diffusion. Geochimica et
Sakai,H., and Tsutsumi, M. (1978) D/H fractionationfactorsbetween
CosmochimicaAcra.47. 168l-1693
serpentineand water at I 00 to 500"c and 2000 bar water pressure,and
Crank,J. (1975)The mathematicsofdiffusion Oxford University Press,
the D/H ratios of natural serpentines.Earth and PlanetaryScienceLetLondon
t e r s .4 0 , 2 3 1 - 2 4 2 .
Dodson, M.H. (1973) Closuretemperaturein cooling geochronological
Satake,H., and Matsuda, J. (1979) Strontium and hydrogen isotope geopetrological
and
systems.Contributions to Mineralogy and Petrology,
chemistryof freshand metabasaltdredgedfrom the Mid-Atlantic Ridge.
40.259-2'74.
Conrributionsto Mineralogyand Petrology,70, 153-157.
Ebisuzaki,Y, Kass,W J., and O'Keeffe,M. (1968)Solubilityand diffuSatake,H., and Matsuo, S. (1984) Hydrogen isotopic fractionation factor
sion of hydrogen and deuterium in platinum Journal of Chemical
betweenbrucite and water in the temperaturerangefrom 100'to 500'C'
Physics,49, 3329-3332.
Contributions to Mineralogy and Petrology, 86, 19-24.
Giletti, B.J.,and Anderson,T.F. (1975)Studiesin difusion, II: Oxygen
Shannon,E.V.,and Wherry,E.T. (1922)Noteson white chlorites.Journal
in phlogopiticmica. Earth and PlanetaryScienceLetters,28,225-233.
of the Washington Academy of Sciences,12, 239-241.
Graham,C.M (1981)ExperimentalhydrogenisotopestudiesIII: DiffuSheppard,S M.F. (1977)Identificationofthe ongin ofore-forming solusion of hydrogen in hydrous minerals, and stable isotope exhangein
tions by the use ofstable isotopes.In Volcanic processesin ore genesis,
metamorphic rocks. Contributions to Mineralogy and Petrology, 76,
p 25-41.Institution of Mining and Metallurgy and GeologicalSociety,
216-228.
London.
Graham, C.M., Sheppard,S.M.F., and Heaton, THE. (1980) Experi- -(1980)
Isotopic evidencefor the origins of water during metamormental hydrogen isotope fractionation factors in the systemsepidotephic processesin oceaniccrust and ophiolite complexes.In Association
H,O, zoisite-HrO and AIO(OH)-H,O Geochimica et Cosmochimica
mafiques ultra-mafiques dans les orogenes.Colloques Internationaux
{cta,44,353-364
du C.N R 5..272.135-147.
Graham,C.M., Harmon, R.S.,and Sheppard,S.M F (1984a)ExperimenStakes,D.S., and O'Neil, J R. (1982)Mineralogyand stableisotopegeotal hydrogenisotope studies:Hydrogen isotope exchangebetweenamchemistry ofhydrothermally alteredoceanicrocks. Earth and Planetary
phiboleand water.AmericanMineralogist,69, 128-138.
ScienceLetters. 57. 285-304.
Graham,C.M., Atkinson,J., and Harmon,R.S.(1984b)Hydrogenisotope Suzuoki,T, and Epstein,S. (1976)Hydrogenisotopefractionationbefractionation in the system chlorite-water. Progressin Expenmental
tween OH-bearing minerals and water Geochimica et Cosmochimica
Petrology,6,139-140 Natural EnvironmentResearchCouncilofU.K.
Acta,40, 1229-1240
PublicationSeriesD, 25, London.
Taylor,H P. (1974)The applicationofoxygenand hydrogenisotopestudHeaton,T.H.E., and Sheppard,S M.F (1977)Hydrogenand oxygenisoies to problems of hydrothermal alteration and ore deposition. Ecotope evidencefor seawater-hydrothermalalteration and ore deposition,
nomic Geology,60, 843-883
Troodos Complex, Cyprus In Volcanic processesin ore genesis,p. 42Velde, B , and Rumble, D. (1917)Alumina contentof chlorite in mus57 Institution of Mining and Metallurgy and GeologicalSociety,LonCarnegieInstitution of Washington Year
covite-beanngassemblages.
don.
Book 76, 621-623.
Jones,A.A. ( I 98 I ) Chargeson the surfacesof the chlorites Clay Minerals,
Viglino, J.A, Graham, CM., and Harmon, R.S. (1984) Experimental
t6,347-359.
hydrogenisotopefractionation factorsin the systemchlonte-Hro. GeoJoswig,W., Fuess,H., Rothbauer,R., Tak6uchi,Y., and Mason, S.A.
logical Societyof America Abstracts with Programs, 16, 683.
( I 980) A neutron diffraction study of a one-layertriclinic chlorite (penWenner, D.B., and Taylor, H.P (1973) Oxygenand hydrogenisotope
ninite) American Mineralogist, 65, 349-352
studiesofthe serpentinizationofultramafic rocks in oceanicenvironKuroda, Y., Suzuoki,T., Matsuo, S, and Shirozu,H. (1976)A prelimiments and continental ophiolites American Journal of Science,273,
nary study ofD/H ratios ofchlorites. Contributions to Mineralogy and
207-239.
Petrology,57, 223-225.
Magaritz, M., and Taylor, H.P. (1976) Oxygen, hydrogen and carbon
isotope studiesofthe FranciscanFormation, Coast Ranges,California.
NowuseR. 5, 1985
MeNuscnrpr RECETVED
Geochimicaet CosmochimicaActa. 40. 215-234.
MnNuscnryr AccEprEDFennunv 13, 1987
for samples for the vicinity of the San Luis Obispo
ophiolite (seediscussionin Sheppard,1980),chloritesgive
calculatedDD"rofor metamorphic fluids within the range
- 36 to - 37m.Theseestimatesare compatible with DD"ro
of metamorphic pore fluids calculated by Magaritz and
Taylor (1976).