American Mineralogist, Volume 74, pages 333-338, 1989
Effect of fluorine on the viscosity of diopsideliquid
DoNar.o B. DrNcwnr,r,
BayerischesGeoinstitut, Universitat Bayreuth, Postfach 10 12 51, 8580 Bayneuth,Federal Republic of Germany
Ansrnlcr
The etrectof F on the shear viscosity of diopside liquid has been investigatedat I atm
and 1300 to 1500'C using the concentric cylinder method. To obtain glassstarting materials, 100-9 batches of compositions along the CaMgSi,O6-F2O-'join-CaMgSi,Ou,
CaMgSirO,,rrForr, CaMgSirOrrrFor, and CaMgSirOrrF (denotedDi, DiF0.25, DiF0.5, and
DiFl, respectively)-were melted in air. F volatilization was significant during batch melting but not during subsequentviscometry determinations. Electron-microprobe analyses
of theseglassespermit comparison of viscosities as a function of F content.
The exchangeofF for oxygen in these liquids reducesviscosity over the range ofcompositions investigated in this study. The viscosity reduction as a function of F/(F + O)
+ O)' > 0), as is the case for albite
exhibits a positive curvature (i.e., d'z(log,o4)/d(F/(F
melt (Dingwell, 1987a),but the viscosity-reducingeffect of F on diopside melt is much
lessthan that observedfor albite melt (Dingwell et al., 1985).
Raman spectraof quenchedmelts (glasses)on the CaMgSirOu-FrO-'join (Luth, 1988a)
and liquidus phaseequilibria in depolymerized systems(Foley et al., 1986; Luth, 1988b)
have indicated that the addition of F (as FrO-,) increasesthe activity of SiO' in these
liquids, and thus Luth (1988a, 1988b)has hypothesizedthat F (as RO-') might increase
depolymerized melt viscosities (due to the increasein SiO, activity). Instead, the present
study shows that the exchangeof F for oxygen, the F O r component, reducesthe melt
viscosity of diopside melt.
The polymerization of diopside melt may be quantified using the NBO/T calculation
(seeMysen, 1988) at 2.0, whereasanhydrous chemical analysesof the most basic, depolymerized, natural melts yield NBO/T values of 1.4 (Mysen, I 987). Thus the presentstudy
shows that the viscosity-reducing effect of F on silicate melts persists to compositions
more depolymerized than any naturally occurring silicate melt. It is therefore proposed
that F will reduce the viscosity of all low-pressure(HrO-poor) silicate melts of geologic
interest.
INrnooucrroN
recently by Foley et al. (1986), who noted that the effect
of F on liquidus phaserelations in the KAlSiOo-MgrSiOoSiO, system can be explained in terms of an increasein
the activity of SiO, in liquids in this system. Similarly,
Luth (1988b) has studied the effect ofF on the liquidus
phaserelations in the NaAlSiOo-CaMgSirOu-SiO2
system,
concluding that the substitution ofF for oxygen (via the
FrO-, exchangeoperator) results in an increase in the
activity of SiO, in theseliquids. Additionally, Luth (1988a)
presentedRaman spectraof quenched melts (glasses)in
the CaO-CaFr-SiO,systemin support of F complexation
with Ca and concomitant increasein the activity of SiO,
in these glasses.If one defines the polymerization of the
melt simply as a measureof the concentrationof bridging
oxygens,rather than a ratio of nonbridging oxygens to
tetrahedrally coordinated cations (as is done for the calculation of NBO/T), then one can describe the effect of
F substitution for oxygen (resulting in an increasein the
activity of SiOr) as a polymerization of the melt. This
wasdoneby Luth (1988a,1988b)and Foleyet al. (1986)
and resultedin the hypothesisof Luth (1988a)that the
The viscosity-reducingeffect of F in silicate melts has
been known and utilized for centuries in glassmaking
(Agricola, 1529). Indeed the element fluorine owes its
name to this property. Fluorspar (CaFr) or "flow-stone"
(from the Latinfluere meaning "flow" or "flux") was the
sourceof elementalfluorine that was first isolatedin 1886.
Recently, the effect of F in reducing the viscosity of
fully polymerized silicate melts has been determined
(Dingwell et al., 1985;Dingwell, 1987a),and the significanceof such viscosity reduction for the petrogenesisof
various F-rich silicic igneous rocks has been recognized
(e.g.,Dingwell, 1985, 1987b; Christiansenet al., 1986;
Clemenset al., 1986;Hannah and Stein, 19861Stern et
al., 1986; Pichavant et al., 1987; Kortemeier and Burt,
I 988).
In addition, certain, very basic igneousrocks (e.g.,melilitites, leucitites, minettes) also contain up to 2 wto/oF
(Aoki et al., l98l). The importance of F in influencing
the petrogenesisof mantle-derivedmelts has beenstressed
0003-004x/89/0304-o3
333
33$02.00
)5+
DINGWELL: EFFECT OF F ON VISCOSITY OF DIOPSIDE LIQUID
"polymerizing" effect of FrO , on depolymerized melts
might result in an increasein melt viscosity.
This last point highlights two different usesof the term
"polymerization" that must be distinguished.The structure and properties of melts along binary metal oxidesilica joins due to the presenceof different metal cations
are often discussedin terms of differencesin "polymerization" at a constant mole fraction of SiO, (e.g., Hess,
1980).This usageof the term "polymerization" derives
from the literature dealing with polymer theory applied
to basicsilicatemelts(e.g.,Masson,1972).Theterm "polymerization" in this sensedeals with the fact that the
bridging-oxygencontent, and, therefore,the SiO, activity
(Hess, 1977),varies with the identity of the metal cation
in melts with equal molar content of SiOr. A melt with a
higher concentration of bridging oxygens is said to be
more polymerized.
The seconduse of the term "polymerization" is to describe the stoichiometrically constrained ratio of nonbridging oxygensto tetrahedrally coordinated cations or
NBO/T (e.g.,Mysen et al., 1985).The coniept of NBO/T
as a structural parameterof silicate melts is derived from
the glass-scienceconcept of a stoichiometrically defined
"oxygen-bridgedensity" (e.g.,Carron, 1969).It is useful
for illustrating the stoichiometric constraint of the number ofnonbridging oxygenspresent in the structure, [assuming that the concentrationof(free) oxygensunbonded
to tetrahedrally coordinated cations is minimall. It is not
yet clear which use of the term "polymerization" is best
suited to structural modeling of the process of viscous
flow in liquid silicates.
Many previous studies of the effect of fluorides on depolymerized melts already exist (e.g., Schwerin, 1934;
Hellbriiggeand Endell, 1941,Owens-IllinoisGlassCompany, 1944; Kozakevitch, 1954; Bills, 1963; Hirayama
and Camp, 1969)and invariably indicate that the addition of fluorides to basic, depolymerized slags reduces
liquid viscosity (seeFig. 5, Dingwell et al., 1985). The
complication arising from the simultaneousadditions of
F and an alkali or alkaline earth (as a fluoride) however,
precludesthe isolation ofthe effectofFrO , on the liquid
viscosity from these previous studies. I am aware of no
previous study where the effect ofFrO , on the viscosity
of depolymerized, Al-free silicate melts has been investigated. Accordingly, the presentstudy was performed to
test the hypothesisof Luth (1988a)that FrO , might insreasethe viscosity of a depolymerized melt.
Diopside was chosenfor the present investigation because (l) the phase diagram for CaMgSirOu-FrO, has
beeninvestigatedby Luth (1988b),(2) the Raman spectra
of quenched melts on this join are consistent with increasingSiO, activity with increasing F content, (3) the
l-atm stability ofF in Ca-bearingsilicate liquids is substantially higher than in alkali-silicate liquids, and (4) the
viscosity of diopside is well known at I atm (Licko and
Danek, 1986;Scarfeet al., 1983)and liquid viscositiesin
the CaO-MgO-SiO, systemare well investigated(Machin
etal.,1952: Machin and Yee, 1954).
The viscosities were determined at l-atm pressureusing the concentric-cylindermethod. Although l-atm viscometry has the drawback of possibleF volatilization, in
practice such volatilization can be ruled out during viscometry becauseno drift in viscosity is recorded during
a closed cycle of viscosity-temperature determinations.
The alternative method of high-pressure synthesis and
measurementof viscosity using the falling-spheremethod
(e.g.,Dingwell, 1987a)was ruled out becausethe errors
associatedwith this technique,at high pressure,are much
larger than for concentric-cylinder viscometry at I atm,
and the effect of F on depolymerized melt viscosity was
anticipated to be smaller than that observedfor the more
polymerizedmelts studiedby Dingwell et al. (1985).
ExpnnrnrnNTAl METHoD
Starting materials
The compositions of this investigation were chosento
lie on the compositionaljoin CaMgSirOu-FrO-,.Four
compositions, corresponding to progressivesubstitution
of F for oxygen in this system, were chosen at molar
valuesof 0, 0.25, 0.5, and 1.0 mol of F per four-cation, a M g S i , O rr , r F or , ,
b a s i s d i o p s i d e( i . e . , C a M g S i r O uC
CaMgSirOrrrFor, and CaMgSi2O5
5F,denotedDi, DiF0.25,
DiF0.5, and DiFl, respectively).The startingglassesfor
the viscometry experimentswere synthesizedfrom CaCO,
pure, -325
(ultrapureprecipitate-Merck), MgO (99.5o/o
mesh-Ventron), SiO, (99.9o/opure, -325 mesh-Ventron), and CaF, (geftillt rein-Merck). CaCO., CaFr, and
MgO powders were dried at 120 "C for at least 24 h, SiO,
was fired aI 1250 "C for l-2 h, and all were stored at 120
'C prior to use. The powders were weighed in stoichiometric proportions into 100-9 (decarbonated weight)
batchesin a 250-mL plastic beaker,transferredto a plastic bottle, and mixed by agitation for - l5 min. The powder mixes were fused directly at 1500 "C by stepwiseaddition of powder mixes to a 50-cmr thin-walled Pt
crucible. The melts were held at 1500 .C for I h in a
MoSi, box furnace.The fused samplesthus obtained were
poured onto a stainless-steelplate for cooling.
Glass analysis
Chips of thesestarting glasseswere analyzedby microprobe for Ca, Mg, Si, and F, and the results of these
analysesare presentedin Table l. The instrument was a
JEoL JSM-35with Krisel control system. The standards
were synthetic diopside-jadeite glassfor Ca, Mg, and Si
and synthetic F-bearing albite glass for F. Precautions
were taken to avoid F volatilization during analysis.Further details of microprobe analysesare given in Table l.
All batchesprobably lost F by volatilization, and the loss
is noticeablein DiFl, which is low in F. The Ca, Mg, and
Si contents of the glassesare within microprobe errors of
the nominal composition, although the Mg values are
slightly low, perhaps owing to F loss as MgF, (cf. Dingwell and Scarfe.1985).
DINGWELL: EFFECT OF F ON VISCOSITY OF DIOPSIDE LIQUID
335
TABLE1, Glass analyses
Di
Sample
CaO
Mgo
Jtu2
CaMgsiro6
(nomF
25 90
1 86 1
55.49
26 61
19.29
54 11
nar)
F
Total
Mg
DI
o
T
FI(F ! n l
DiF0.25
98.21
wtTooxides.
26 39
1 77 5
ss.96
2.39
97 42
Four-cationmolar basis
1.024
1. 0 1 9
1. 0 0 0
1 033
0.957
1.000
1944
2.024
2.000
6.000
5 944
5.883
0.284
0.0461
25.42
17.55
57.03
4.65
99.11
26.01
17.75
56.24
12 65
98.55
o
o
0 986
0 948
2.066
5.792
0.548
0.0864
1. 0 0 8
0.958
2.034
5 230
1.549
0.2285
Note: Glassanalyseswere performedby wavelength-dispersiveanalysis
instrumentand Kriselcontrolsystem Operatingconusinga JEoLJsM-35
voltage,100-nAbeam currenton C,30-s
ditionswere 15-kVaccelerating
maximumcount times usinga 10 x 10 pm rasterand movingthe sample
continuouslyunderthe beam.Standardswere a syntheticdiopside-jadeite
glassfor Ca, Mg, and Si and AB1F glass(seeDingwellet al , 1985)for F.
Analysesot F-bearingglassesare the meansof five analysispoints.
- Weight percentanalyseshave been normalizedon the basis of CaO
+ MgO + SiO, : 100% to facilitatecomparisonwith nominaldiopside.
o
5'6
t't
.,o,ooo,r,*,
relationships
of four melts
Fig. I . Theviscosity-temperature
24.SeeTable
Errorbarsrepresent
on thejoin CaMgSirOu-FrO-,.
I for compositions.
Viscometry
The viscosities were determined with the concentriccylinder method using a modified version of the apparatusdescribedby Dingwell (1986).Improvementsin the
design of the present viscometer include a precise positioning system for the immersion depth and the x-y
centering of the spindle, using micropositioning stage
componentspreciseto +50 pfl, orrd a computer-based
time-averagingof the output signal. The viscometer has
been standardized with National Bureau of Standards
StandardReferencematerial SRM 7l l, lead-silicaglass.
The torque was recorded over a range of spindle speeds
at a seriesof temperaturesfor SRM 711. Plots of the
millivolt signal (from the rotary variable displacement
transducer)versusviscosity (from the SRM 7 I I certificate
ofviscosity values) were constructed for each speedand
used as calibration curves for viscosity determinations.
The precision of this apparatus,determined by replicate
measurementsof viscosity is +2.5o/oat 2o. The accuracy
of the method is taken as the sum of the precision of
The spindle
standardand sampledeterminationsat + 50/0.
used in this study is the same as described by Dingwell
and Virgo (1987). Each viscosity determination was made
at three different rotation speeds(20, 50, and 100 rpm).
The calculated viscosity was independent of shear rate;
i.e., Newtonian behavior was observed for all compositions. The melts were sampled by dipping a Pt wire loop
into the samplecrucible after eachviscosity determination
to ensurethat the samplewas a single-phasesilicateliquid.
No liquation, crystallization, or vesiculation was observed. The viscometry runs were begun at the highest
temperatureand then at successivelylower temperatures
constrainedby the phasediagram of the CaMgSirO6-FrO,
system (Luth, 1988b) to lie above the liquidus (where
known). The final determination of the run was a reoc-
cupation of the highest-temperaturedetermination. No
difference was observed between initial and final determinations at the high-temperaturepoint, indicating that
no instrumental or compositional drift took place during
the viscometry experiments.
Rrsur-rs AND DIScussIoN
The results of the viscosity determinations on Di,
DiF0.25, DiF0.5, and DiFl are presentedin Table 2 and
illustrated in Figure l. The viscosities decreasewith increasingF content at all temperatures.The temperature
dependenceof the viscosities are Arrhenian within error
and are fit to equations ofthe form
* E^/2.303RT,
logro4: log,o4o
(l)
where a is the viscosity at temperature T, R is the gas
constant, and E^ and 46are termed the activation energy
and pre-exponentialfactor, respectively.
In order to quantify the effect of FrO-, on the viscosity
of CaO-MgO-SiO, melts, we have plotted the viscosity
join as-afunction
of thesemelts on the CaMgSirOu-FrO-,
of the anionic mole fraction of F, F/(F + O), in Figure 2.
The dependenceof viscosity on composition exhibits a
smooth function with positive curvature when plotted
againstF/(F + O).
Comparisonwith polymerizedmelts
The results of the determination of diopside viscosity
are within error of thoseof Licko and Danek (1986)using
oscillation viscometry and those of Scarfeet al. (1983)
using concentric-cylinderviscometry. This agreementil-
t36
DINGWELL: EFFECT OF F ON VISCOSITY OF DIOPSIDE LIQUID
TABLE2. Melt-viscositydata
Log,oviscosity (poise)
DiF0.25
1 4 9 5 . 0" C
1 4 4 58 . C
1421.2"C
1396.6.C
1372.0"C
1347.5"C
1322.9"C
o
o
o
o
1303.2
rc
jadeite
log tto
E" (kcal/mol)
o
o
,.
perarK
0 640
0.805
0.974
-5.024
45.83
0.688
0.763
0.840
0.930
101
4.662
42.07
DiF0.5
0.749
0.811
0.895
-4.078
36.85
0.664
0.728
0.839
-5.496
45.61
nepheline
ot
o'
o3
,nr*o,
Fig. 2. Comparisonof the effecton melt viscosityof substituting F for oxygenin severalsilicatemeltsat 1400'C. Data
from Riebling(1966),Dingwellet al. (1985),and this study.
"Peralk"is a peralkalinecompositionwith 75 molo/o
silica.
lustrates that the results of these two methods on melts
in this viscosity range give consistentresults.
The positive curvature in the viscosity vs. F content
[F(F + O)] plot of Figure 2 is similar to thar obtained
by Dingwell (1987a) for the NaAISi3OB-F2O_,
system at
7.5 kbar. Figure 2 compares the efects ofF on reducing
the viscosity of albite, jadeite, nepheline, and diopside
melts at I atm. The viscosity-reducingeffect of F on fully
polymerized melts has already been shown to be a strong
positive function of SiO, content [or the Si/(Al + Si) valuel. Similarly, Figure 2 illustrates that at constant mole
fraction SiO, (equalto 0.50 for jadeite and diopsidemelrs),
the viscosity of the fully polymerized jadeite melt is lowered much more strongly than the viscosity of depolymerizeddiopsidemelt. This observationis consistentwith
the previous comparison (at 75 molo/oSiOr; Dingwell et
al., I 985) showingthat F hasa largereffecton the viscosity
of fully polymerized albite melt than on peralkaline and
peraluminous melts that are depolymerized by excesses
of Na and Al, respectively.
Melt structure
In the past few yearsthere have been a seriesofpapers
in the geologicalliterature outlining the possiblesolution
mechanismsof F in silicate liquids; largely in the context
ofglass spectroscopyand experimental liquidus phaserelations (e.g., Manning et al., 1980; Mysen and Virgo,
1985a,1985b;Foley et al., 1986;Luth, 1988a,1988b).
In fully polymerized silicate melts suchas albite and jadeite, it has been recognizedthat F substitution for oxygen
must depolymerize the melt structure, as oxygens involved in bridging bonds to two adjacent tetrahedral cations are stoichiometrically replaced by F. The simplest
schemefor F substitution might involve the formation of
Si-F bonds to replacethe bridging oxygensby what could
be termed "nonbridging" F atoms. Spectroscopicevidence for this substitution mechanism in highly polymerized glassesis equivocal (Mysen and Virgo, 1985b;
Hayashi et al., 1987).
Mysen and Virgo (1985b) realized that the effect of F
on the properties of fully polymerized melts could be explained by an alternative solution mechanismfor F where
an unequal number of Na and Al atoms are withdrawn
from their network-stabilizingand network-forming roles,
respectively,to form charge-balancedfluoride complexes
no longerdirectly participating in the aluminosilicatemelt
structure. Such a mechanism would depolymerize the
"residual" aluminosilicateor oxide proportion of the melt
becausethe unequal remaining Na and A1 contents of the
oxide portion would require a network-modifying role for
the Na (or Al) in excessof that required to chargebalance
the Al (or Na). Mysen and Virgo (1985b) went on to
propose that, to a first approximation, the viscosity vs.
temperaturerelationshipsof the F-bearing melts of Dingwell et al. (1985)could be modeled in terms of the bulk
composition of the "residual" aluminosilicate oxide portion of those melts. The proportions of Na and Al coordinated as fluoride complexeswere estimated from Raman spectra, and the "residual" melt composition was
calculated.The data of Riebling (1966) was then used as
a "template" for estimating the bulk F-bearing melt viscosities. Central to this approach was the idea that the
presenceof neutral fluoride complexes has a negligible
direct effect on the viscosity of F-bearing silicate melts.
In depolymerized melts, such as diopside, the solution
of F might proceed by the formation of Si-F bonds in
replacementofbridging or nonbridging oxygensor by the
formation of calcium and magnesiumfluoride complexes
resulting, in this Al-free composition, in an increasein
the proportion of oxygen that forms bridging bonds. Raman spectroscopicobservations of glassesand liquidus
phaseequilibria favor the latter interpretation as both are
consistentwith an increasein the activity of SiO, in the
liquid (Foleyet al., 1986;Luth, 1988a,1988b).Using the
DINGWELL: EFFECT OF F ON VISCOSITY OF DIOPSIDE LIQUID
53 t
samereasoningas outlined by Mysen and Virgo (1985b), then F is predicted to decreasethe viscosity of all such
Luth (1988b)hypothesizedthat F might increasethe vis- low-pressure,relatively dry, basic melts.
cosity of diopside melt owing to such an increasein the
activity of SiO., which presumably reflectsan increasein
AcrNowr,oocMENTS
the proportion of oxygen bridging bonds. If the contriI wish to thank Bob Luth for analysesofthe glassesand Georg Herrbution of neutral fluoride complexes can be neglected, mannsdiirfer for technical assistancewith adaptationsto the viscometer
then this structural parameter,the activity of SiOr, should and Mark Brearley for his comments F. J. Ryerson and J. Stebbinsare
thanked for constructivereviews.
dominate the effect of F addition on melt viscosity.
The present study shows that the effect of F is to significantly reducethe viscosity of diopside melt. Thus we
RnrBnBNcBscrrnn
concludethat if, as is likely, the F in diopside-FrO-, melts Agricola, G. (1529)De Re Metallica (not seen;extractedfrom Hammond,
is presentas neutral fluoride complexes,then the presence
C.R, The elements.In R.C. Weast, Ed., Handbook of physicsand
of such fluoride complexes is the dominant control on
chemistry (53rd edition). The Chemical Rubber Company, Cleveland,
Ohio.1973).
melt viscosity. This conclusicrnleadsto the generalization
that it is not appropriate to model (quantitatively or qual- Aoki, K, Ishikawa,K., and Kanisawa,S. (1981)Fluorinegeochemistry
ofbasaltic rocks from continental and oceanicregionsand pelrogenetic
itatively) viscous flow in mixed fluoride-oxide (silicate)
application. Contributions to Mineralogy and Petrology,76,53-59
melts entirely in terms of the oxide (silicate or alumino- Bills, P M (1963)Viscositiesin silicateslagsystems.Journalofthe Iron
and SteelInstitute.201. 133-140.
silicate)proportion of the melt structure,to the exclusion
Carron, J.-P. (1969) Vue d'ensemblesur la rh6ologiedes magmassilicat6s
of the consideration of fluoride complexes.
Geologicalimplications
The recognition that many depolymerized,silica-poor,
mantle-derived magmas contain significant concentrations of F (Aoki et al., l98l) has led to interest in the
possibleeffectsofF on physical and chemical properties
of depolymerizedsilicatemelts (e.g.,Foley et al., 1986).
In this context, diopside is a useful model compound for
discussingbasaltic and more basic melts, just as albite is
often used as a model for granitic melts. On the basis of
a qualitative comparison of the effectof F on polymerized
and depolymerized silicate melts, I conclude that F will
be less effective in reducing the viscosities of more basic
melts than is likely the casefor granitic and other felsic
melts. Several factors combine to reduce the predicted
effect of F on more basic melts. First, the viscosity-reducing effectis smaller per wt0/oof F (Fig. 2). Second,the
liquidus temperaturesof more basicmelts are much higher
and thus the divergence of viscosity-temperature relationships that results from decreasingactivation energy
with F addition cannot be exploited to expand the effect
of F on viscosity in basic systemsas is possible in felsic
systems.
Surtlvr.l,ny
The viscosity of diopside melt is significantly reduced
by the substitution of F for oxygen (via the exchange
operator F,O ,) at least up to F/(F + O) : 0.23. The
effect of F on melt viscosity is much less than that observedfor polymerized melts such as albite and jadeite.
The substitution of F into diopside melt has been proposed to be accompaniedby an increasein the activity
of SiOr. The present study shows that, if previous structural conclusions regarding the substitution of F into
diopside melt are correct, then predictive models for the
viscosity of mixed fluoride-oxide (silicate) melts must
considerthe effectofthe presenceofneutral fluoride complexeson the viscous flow process.
If diopsideis a model for naturally occurringbasicmelts,
naturels. Bulletin de la Soci6t6Frangaisede Min6ralogie et Cristallographie,92, 435446.
Christiansen,E.H., Sheridan,M F., and Burt, D.M. (1986)The geology
and geochemistryofCenozoic topaz rhyolites from the westernUnited
States.GeologicalSocietyofAmerica SpecialPaper 205,82 p.
Clemens,J.D., Holloway, JR., and White, A.JR. (1986) Origin of
A-type granite: Experimental constraints.American Mineralogist, 71,
3t7-324
Dingwell, D.B. (1985)The structureand propertiesoffluorine-rich silicate
melts: Implications for granite petrogenesis(abs.).In R.P. Taylor and
D F Strong,Eds , Granite-relatedmineral deposits.Geology,paragenesis and tectonic setting, p. 72-8 I . Canadian Institute of Mining and
Metallurgy, Montreal.
( I 986) Viscosity-temperature
relationshipsin the systemNarSirOrNaoAlrOr.Geochimica et CosmochimicaActa, 50, 126l-1265
-(1987a)
Melt viscositiesin the syslemNaAlSirOr-HrO-FrO-,.In
B.O Mysen, Ed., Magmatic processes:Physicochemicalprinciples, p.
423431 . GeochemicalSocietySpecialPublication 1. University Park,
Pennsylvania.
-(1987b)
The structure and propefties of fluorine-rich magmas:A
review ofexperimental studies.In R.P. Taylor and D.F. Strong, Eds.,
Recent advancesin the geology of granite-related mineral deposits.
Canadian Institute of Mining and Metallurgy Special Volume 39, 112.
Dingwell, D B., and Scarfe,C.M. (1985) Chemical diffusion of fluorine in
melts in the systemNarO-AlrO,-SiOt Earth and PlanetarySciencektters.73. 377-384.
Dingwell, D.B., and Virgo, D. (1987) The effectof oxidation stateon the
viscosity of melts in the systemNarO-FeO-Fe,O,-SiOr.Geochimica et
CosmochimicaActa. 51. 195-205.
Dingwell,D.8., Scarfe,C.M., and Cronin, D (1985)The effectof fluorine
on viscositiesin the systemNarO-AlrOlSiOr: Implications for phonolites, trachytes,and rhyolites American Mineralogist, 70, 80-87.
Foley,S.F.,Taylor, W.R., and Green,D.H. (1986)The effectof fluorine
on phaserelationshipsin the systemKAlSiOo-MgrSiO4-SiO,at 28 kbar
and the solution mechanismoffluorine in silicatemelts. Contributions
to Mineralogyand Petrology,93,46-55.
Hannah, J.L., and Stein, H.J (1986) Oxygen isotope compositionsof
selectedLaramide-Tertiary granitoid stocks in the Colorado Mineral
Belt and and their bearingon the origin ofClimax-type granite-molybdenum systems.Contributions to Mineralogy and Petrology, 93,347358.
Hayashi,S., Kirkpatrick, R.J., and Dingwell, DB (1987)MASS NMR
study of F-containing NarO-AlOr-SiO, glass(abs.).EOS, 68, 436.
Hellbriigge, H., and Endell, K. (1941) Zusammenhinge zwischenchemischer Zusammensetzungund Fliisigskeitsgradvon Hiittenschlacken
sowieihre technischeBedeutung.Archiv fiir das Eisenhiittenwesen,14,
307-3I 5
338
DINGWELL: EFFECT OF F ON VISCOSITY OF DIOPSIDE LIQUID
Hess,P C (1977)Structureofsilicate melts CanadianMineralogist,15,
t62-178
( 1980)Polymerizationmodel for silicatemells In R B Hargraves,
p 3-48. PrincetonUniversity Press,
Ed, Physicsofmagmatic processes,
Princeton, New Jersey.
Hirayama,C , and Camp, F.E. (1969)The effectoffluorine and chlorine
substitution on the viscosity and fining of soda-lime and potassiumbarium silicate glass Glass Technology, 10, 123-127.
Kortemeier, W T , and Burt, D M. (1988) Ongonite and topazite dikes in
the Flying W ranch area, Tonto Basin, Arizona American Mineralogist,13, 507-523
Kozakevitch, P (1954) Sur la viscosite des laitiers de hauts fourneaux.
Revuede Metallurgie,8, 569-584
Licko, T., and Danek, V (1986)Viscosityand structureofmelts in the
systemCaO-MgO-SiOr.Physicsand Chemistryof Glasses,27, 22-26
Luth, R.W. (1988a)Raman spectroscopic
study of the solubilitymechanisms of F in glassesin the systemCaO-CaFlSiOr. American Mineralogist,73,297-305.
-(1988b)
Efects of phaseequilibria and liquid structure in the system NaAlSiOn-CaMgSlOu-SiO, American Mineralogist, 73, 306-312.
Machin,J S.,and Yee,T B (1954)Viscositystudiesof systemCaO-MgOAl,O,-SiOr: IV. 60 and 650/oSiO: Journal of the American Ceramic
Society,37, 177-186.
Machin, J.S, Yee, T.B., and Hanna, DL. (1952) Viscosity studiesof
systemCaO-MgO-Al,Or-SiO,:III. 35, 45, and 500/oSiO' Journal of
the AmericanCeramicSociety,35, 322-325
Manning, D.A C , Hamilton, D L., Henderson,C.M.B., and Dempsey,
MJ. (1980) The probableoccurrenceof interstitial Al in hydrous,
F-bearingand F-free aluminosilicate melts Contributions to Mineralogy and Petrology,15, 257-262
Masson,C R (1972)Thermodynamicsand constitutionof silicateslags.
Joumal ofthe Iron and SteelInstitute,210, 3-l l.
Mysen,B O. (1987)Magmaticsilicatemelts:Relationsbetweenbulk composition, structureand properties In B.O. Mysen, Ed., Magmatic propnnciples,p. 375-399. GeochemicalSociety
Physicochemical
cesses:
SpecialPublication l. University Park, Pennsylvania.
-(1988)
Structureand propertiesof silicatemelts, 354 p. Elsevier,
Amsterdam.
Mysen, B.O, and Virgo, D. (1985a)Interaction betweenfluorine and
silica in quenchedmelts on the joins SiO'-AlFi and SiOrNaF determined by Raman spectroscopy.Physics and Chemistry of Minerals,
12,77-85
-(l9E5b)
Structure and properties of fluorine-bearingaluminosilicate melts: The system NarO-Al'Or-SiO2 at I atm Contnbutions to
Mineralogy and Petrology,9 l, 205-220.
Mysen, B O., Virgo, D, and Seifert,FA. (1985)Relationshipsbetween
properties and structure of aluminosilicate melts. American Mineralogist,70,88-105.
Owens-Illinois GlassCompany, General ResearchLaboratory. Q944) Effect of fluorine and phosphorus pentoxide on properties of soda-dolomile lime-silica glass.Joumal of the American Ceramic Society,27,
369-372.
Pichavant, M., Valencia Herrera, V , Boulmier, S., Briqueu, L., Joron,
J-L., Juteau,M., Marin, L , Michard,A , Sheppard,S M.F, Treuil, M ,
and Vemet, M (1987) The Macusaniglasses,SE Peru: Evidenceof
chemical fractionation in peraluminous magmas In B.O Mysen, Ed.,
Physicochemicalprinciples,p. 359-373. GeoMagmatic processes:
chemrcalSocietySpecialPublication I University Park, Pennsylvania.
Riebling, E F (1966) Structure ofsodium aluminosilicate melts containSiO, at 1500'C Joumal of PhysicalChemistry,
ing at least50 moleo/o
44,2857-2865
Scarfe,C.M., Cronin, D.J, Wenzel,JT, and Kauffman, DA. (1983)
Viscosity-temperaturerelationships at I atm in the system diopsideanorthite AmericanMineralogist,68, 1083- I 088.
Schwerin,L. (1934')The effectoffluorspar on the viscosity ofbasic slags.
Metalsand Alloys, 5, ll7-123
, . H , F o o r d ,E . E . ,S h i g S t e r nL
, A . , B r o w n ,G E , J r , B i r d , D K , J a h n s R
ley, J E, and Spaulding,L 8., Jr. (1986)Mineralogyand geochemical
evolution of the Little Three pegmatite-aplitelayered intrusive, Ramona, California AmericanMineralogist,7 1, 406427
MANUscRrF.r RECETvEDAucusr 29, I 988
MnNuscnrrr AccEprED Nolrr'lsen 23, 1988