American Mineralogist, Volume 6E, pages 1083-1088,l9B3
Viscosity-temperaturerelationshipsat I atm in the systemdiopside-anorthite
C. M. Scnnppt
GeophysicalLaboratory, Carnegie Institution of Washington
2801 Upton Street, N.W., Washington,D. C. 20008
D. J. CnoNrN,J. T. WEr.rzBrnNn D. A. Keuppueu
Centerfor Materials Science
National Bureau of Standards, Washington, D. C. 20234
Abstract
Laboratory measurementsof the viscositiesof nine compositionson the diopsideanorthite join were made with a concentric-cylinder viscometer at temperaturesbetween
approximately 1200and 1600'Cin air. Viscosities were measuredboth above the liquidus
and in the supercooledliquid region.Viscositieswere independentof shearrate, indicating
Newtonian viscousbehaviorfor the melts. Viscositiesdecreasedwith increasingtemperature and, at constant temperature,they decreasedwith increasingamounts of diopside
component in the mixtures. The temperature dependenceof the viscosity above the
liquidus was fitted to an Arrhenius relationship from which activation energiesfor viscous
flow, ranging from 6l kcal/mole for a melt of anorthite composition to 38 kcaVmolefor a
melt of diopsidecomposition,were derived. Viscositiesover the entire rangeof melt and
supercooledliquid temperatureswere fitted by the method of least squaresto the Fulcher
equation.The fit was better than 0.02 in log viscosity.
The present experimental results are systematically lower at any temperature than the
earlier measurementsof Kozu and Kani (1935).Viscosity is strongly dependenton the
amount of alumina in the melt and on temperature.As diopside is addedto anorthite melt,
there is a modification of the three-dimensionalnetwork structure in favor of a more
depolymerized structure. Activation energiesfor viscous flow, however, suggestthat this
change does not take place gradually across the join and that there are fundamental
differencesin the flow properties of liquids of diopside composition compared to aluminabearingmelts along the join.
Introduction
tion, transport and emplacementof igneousrocks (e.g.,
Bowen (1915)selectedthe ternary system diopside- Shaw, 1965;Spera, 1980).Melt viscositiescan also proanorthite-albite to illustrate the crystallization of haplo- vide information about melt structure (e.g., Bockris and
basaltic and haplodioritic magmas. Since that time con- Lowe, 1954; Bottinga and Weill, 1972; Kushiro, 1980;
siderableattention has been given to the phase relation- Mysen et al., 1980). For these reasons, and because
ships (e.9., Osborn, 1942; Schairer and yoder, 1960; viscosities for this system by Kozu and Kani (1935)
Wyllie, 1963;Kushiro and Schairer, 1969;Kushiro , lg73) appearedto be systematicallyhigh (Scarfe et al., 1979),a
and more recently to the thermochemistry(e.g., Weill et study of viscosity-temperature relationships along the
al. 1980;Navrotsky et al., l9E0; Hon et a/., lggl) of this binary joins and in the ternary system was begun in
classicsystem.In contrast,much less attentionhas been conjunction with structural investigations on quenched
focusedon other petrologically-importantproperties such melts (glasses)with Raman and infrared spectroscopy.
This paper describes viscosity results for the join diopas the viscosity and density of melts in the system.
A knowledge of the viscosity of silicate melts and side-anorthite over the temperature range 12fi)-1600'C,
magmasis important to the understandingof the genera- which for all compositions includes both the region above
the liquidus and part ofthe supercooledregion below the
I Present address: Experimental petrology
liquidus. The objectives were to obtain improved data
Laboratory, Department of Geology, University of Alberta, Edmonton, Canada over a wider temperature range than results previously
T6G 283.
published, and to attempt to interpret the results in terms
0003-004x/83/l
I l2-t 083$02.00
l0E3
SCARFEET AL.: VISCOSITY IN THE SYSTEM DIOPSIDE_ANORTHITE
l0E4
design, calibration procedure and operation of the viscometerare being publishedelsewhere'
Measurementswere usually taken at 25 or 50'C interExperimental method
vals after a stabilization time at each point of approximately one hour. Measurements were routinely perStarting mixtures were prepared in 650 g batches from
reagent grade oxides, carbonates and purified quartz formed during cooling from 1600 to 1200"C, but some
sand. Melts free from bubbles and crystals were obtained measurementswere also made during heating' There
by melting at approximately 100'C above the liquidus were no measurabledifferences in the viscosities determined along these two thermal paths. Viscositieswere
temperature of each composition. Melts were made in
independentofthe rotationalspeedofthe inner cylinder,
platinum crucibles in an electrically-heatedfurnace with
indicatingNewtonian viscousbehavior. This conclusion
SiC resistance elements, and were stirred for approxiis in accord with the observation that most silicate melts
mately two hours to assurehomogeneity.The melts were
of geological interest show Newtonian behavior at temx
diameter
poured
cm
10.5
5.5
then
directly into a
crucible, which was used for viscosity measurements. peratures above their liquidi (Shaw, 1969; Murase and
The homogeneity and compositions of the quenched McBirney, 1973;Scarfe, 1973,1977)'
melts (glasses) were verified by electron microprobe
Results
analysis.
The results are shown in Figure I and Table I for nine
Viscosities were measuredwith a concentric-cylinder
viscometer, which uses a platinum-IO7a rhodium inner compositionsalong the join diopside-anorthite.Viscosicylinder 5 cm long and 1.2 cm in diameter,with conical ties were measuredover a range of temperaturesabove
ends. The inner cylinder rotates and the resultant torque the liquidus and in the supercooled liquid region until
is measured and converted to a millivolt signal. The significant crystallization intervenbd and measurements
apparatus was calibrated with NBS standard lead-silica becametime dependent.Viscosities'for all compositions
glass SRM 7ll, for which the viscosity-temperature decreasewith increasingtemperature and the curves are
relationship is accurately known. The viscosities are sub-parallel to each other. For each composition, the
accurate to -+5Vowith a precision of tlVo, and tempera- viscosities at the liquidus temperature and at 1600'Care
tures have uncertainties of tl"C. Further details of the summarizedin Table 2. At constant temperature,there is
of structural changesin the melts as a function of temperature and composition.
4.0
x An-100
*Di-20
oDi-30
*Di.40
oD-50
vDi - 5E
aDi - EO
oD - 90
oD -100
3.5
30
T
'E
r.'
-\\
f'-.,-
F-E',o
-An-(c&U)
.;s
g r's
.2
1.0
0.5
il00
r200
r300
t400
r500
r600
rru
Tempaoture
l'Cl
plot for compositionsin the systemdiopside-anorthite.Dashedlines are previousdeterminations
Fig. l. Viscosity-temperature
by Kozu and Kani (1935),Cukierman and Uhlmann(1973), and Scarfeet at. (1979).Liquidus temperaturesfor each composition are
given in Table 2.
SCARFE ET AL.: VISCOSITY IN THE SYSTEM DI1PSIDEAN1RTHITE
1085
Table l. Results of viscosity experimentsfor the join diopside-anorthite in log4 (poises) and wLVodiopside
Temp('C)
I 625
I 600
I 575
I 550
I 525
I 500
1475
I 450
1425
1400
1375
I 350
t325
I 300
1275
I 250
1225
I 200
I175
Dl
90Dl
0.465
SODI
58Dl
50Dl
40Di
30Dl
20Dl
0.536
0.85t
0.956
1.029
t.;;
1.252
1.434
1.522
1.625
1.738
1.836
0.559
0.648
0.759
0.990
1.074
1. 1 9 4
I .358
0.678
0.800
0.898
I .145
t)io
r.rio
t.;;
0.821
0.957
1.057
I .329
1.432
I .553
t.;;
1.I28
|.237
1.518
t.;;
t.;;
T.883
1.730
I .848
1.980
1.327
1.442
1.721
I.s39
t.;;
1.8s4
t.989
2.132
1.086
,.rio
,.:lu
2.258
2.384
,.u.i.u
2.730
2.457
1.909
2.098
An
I .333
1.433
1.513
1.625
I .7ll
1.841
1.968
2.101
2.246
2.402
2.567
3 . 14 0
2.468
2.U6
t.;;
a systematic decrease in viscosities from anorthite to
diopside. However, because of the temperature effect
along the liquidus, viscosities at the liquidus increase
away from both end-membermelts to a maximum at the
eutectic composition.
Compositionsacross the join have constant moleVo
SiO2 and CaO but varying molar contents of Al2O3 and
MgO. When the results are plotted against moleVoal:umina in the starting compositions (Fig. 2), for any temperature there is a linear relationship between log viscosity
and alumina to l6Vo alumina. Beyond 16% alumina, the
data for anorthite lie somewhatbelow the linear extrapolation. It should also be noted that the slope increasesas
temperature decreases, indicating that alumina has a
greater effect on viscosity as temperatures approach
thoseof the liquidus.
Viscosities above the liquidus were fitted to an Arrhe.
nius relationship of the form
logz:
(1)
logr6 + Bq|Q.SRT)
where 4s is a constant, E4 the activation energy for
viscous flow, R the gas constant, and ? the absolute
temperature.Activation energiesrangefrom 61 kcaUmole
for anorthite to 38 kcal/mole for diopside (Table 3).
Becauseof the notablecuryaturein log4 vs. llT (Fig. 3),
however, the experimental data are better fitted by the
method of least squaresto the Fulcher (1925)equation
Table2. Liquidustemperatures
andviscositiesin the diopsideanorthit€svstem
l { B St
K-l 532
K-l 756
K-I 542
K-1540r
K-l 554
(-l 545
K- l 560
K-l 547
K-l 538
Diopslde,
wt t
t00
90
80
58
50
40
30
20
0
Llguldus,
1 3 9.15
1358
1343
1274
1323
1388
1438
1488
1553
Logn (T"),
polses"
L03
1.25
1.46
2.14
2.00
1.82
1.79
1.67
|.61
t o g n ( 1 G 0 0. C ) ,
polses
0.47
0.52
0.64
0.85
0.96
I .03
3,.
F3't
s
8
.x
'4:'
u00r
GW,W
I .18
1.2s
1.43
t5
*Eutectlc_colposltlon. Llquldus temperaturesfron phase
diagrrm by
o s b o r n( 1 9 4 2 ) ; s e e a l s o Y o d e r( 1 9 7 6 ) ,
Ilrtob plcot
AlzOr
Fig. 2. Plot of log4 vs. moleVo Al2Or. Isotherms at 50"C
intervals between 1250-l6m'C.
SCARFE ET AL.: VISCOSITY IN THE SYSTEM DIOPSIDE_ANORTHITE
1085
The activation energy may be obtained from the instantaneous slope (S1)at any temperature by using
t2
Eq : 2.3RSr
lro
&
Eo..
-[
tS
I
a
5
0.6
0a
60
t'irrr,o.(r) -
s't
5'2
Fig. 3. Plot of log4 vs. l/T for diopsidemelt. Broken line is
Arrhenius fit and solid line is Fulcher fit. Circles are
experimentaldata points. Additional data points for diopside at
l4lO"C(0.942,0.959), 1395"C(
1.003) and 1385'C(1.038)are not
included in Table I and Fig. l.
los"r=A+B(T-ro)
Q)
where Tis the temperaturein degreescentigradeand A, B
and I0 are constants. The Fulcher equation reproduces
the experimental data for each melt to better than 0.02 in
log4 throughout the entire range of melt and supercooled
liquid temperatures. An example of this fit is given in
Figure 3 and the Fulcher and Arrhenius constantsfor all
compositionsare given in Table 3. By using the Fulcher
constants and temperature, activation energies can be
recalculated by differentiating log4 with respect to llT,
which yields the expression
(3)
d(loer,)/d(l/T): B(TI(T - Td)2
Table3. Arrheniusand Fulcherequationconstants
0lopside,
rtl
logno
too
-3.971
90
-4.700
80
-4.614
58
-5.235
En
A
B
To
37.9
-0.729
44.6
-2.il2
3871
44.8
-r.938
2703
54a.2
5I.9
-1.587
2304
656.5
775.7
946.6
374.3
50
-5.143
52.0
-1.394
2093
706.4
40
-5.042
52.O
-1.945
3024
585.4
30
-5.113
53.9
-1.472
2319
7 2 7. 5
20
-5.324
56.4
-1.615
2540
715.5
0
-5.637
60.5
-t.343
2123
833.0
Arrhenius constants calculated fM
(En in kcal/mle).
in table |.
data above the'liquldus in table I
Fulcher constants calcu]ated frm cqnplete data set
(4)
Activation energies for viscous flow calculated in this
way for all compositions are plotted in Figure 4' The
changein the activation energy for diopside as a function
of temperature is significantly different from the activation energy changes for other compositions. A more
pronouncedchangein the structure of diopside melt as a
function of temperature compared with other compositions is therefore inferred. When activation energiesare
plotted againstmoleVoalumina there is a smooth increase
in activation energy between 4-25 moleVoalumina (Fig.
5). Diopsideand the compositionwith2moleVo alumina'
however, do not complete this linear trend, reinforcing
the observation that there is a major break between the
behavior of diopside melt and alumina-bearing melts
along the join.
Discussion
The present data are compared in Figure I with some
previous determinations. For the same compositions the
present results are lower than the measurementsof Kozu
and Kani (1935).Bottinga and Weill (1972),in a discussion of their model for the viscositiesof silicate melts'
noted similar discrepancies between their calculations
and the results of Kozu and Kani (1935). There is,
however, agreement between the present results for
diopside and those of Scarfe et al. (1979). Results for
anorthite are also in agreementwith the work of Cukierman and Uhlmann (1973) and Cranmer and Uhlmann
(1981).
The dependenceofviscosity on the contentofalumina
is striking (Fie. 2). This relationship may be used to
predict viscosities in the system from the moleVo of
alumina in the starting composition. The results indicate
increasedresistanceto flow as the alumina content increases.Similar systematicchangeshave been observed
in the system Na2O-Al2O3-SiO2with replacement of
Na2Oby AlzOr (Riebling,1966;Kou et al', 1978).
Mysen et at. (1980) interpreted Raman spectroscopic
data for diopside to mean that diopside melt consists
primarily of chains (SiO3-) with some monomers (SiOi-)
and sheets(SirO3-). When alumina is added, aluminum
goes into tetrahedral positions in the melt and calcium
and magnesiumprovide charge balance(Mysen er a/.,
l98l). Under these circumstancespolymerizationof the
melt takes places, Ieading to the observed viscosity
increases.In contrast, spectroscopicevidence(Virgo el
al., 1979) suggeststhat anorthite melt is a completely
polymerized three-dimensional structure having two
types of network unit, one containing more Al than the
other. Furthermore, the X-ray data of Taylor and Brown
(1979)suggestthat the structureofanorthite melt consists
of four-membered rings of SiOr and AIO+ tetrahedra.
SCARFE ET AL,: VISCOSITY IN THE SYSTEM DIOPSIDE_ANORTHITE
Consequently, melt structures along the join should be a
mixture of highly polymerized, three-dimensional, network structures and relatively depolymerized chain,
sheetand monomerstructures.
The data plotted in Figures 4 and 5 strongly suggest
fundamentaldifferencesin the flow properties of diopside
vs. alumina-bearing melts along the diopside-anorthite
join. This behavior in turn implies that the structure of
melts along the join changes radically where Al2O3 is
present as a component. It is assumed, of course,
throughout this discussion that the activation energy for
viscous flow is a structure-sensitiveparameter and that
glassesdo, in fact, fully represent the structure of their
corresponding melts (Mysen et al., 1980; Seifert et al.,
l98l). Unfortunately,at presentit is difficult to go beyond
these qualitative statements because the mechanism of
viscousflow in silicatemelts is poorly understood.
In additionto the compositionaldependenceof the melt
structure, it is clear that temperature-inducedstructural
changesalso occur. This latter effectis most pronounced
in the case of diopside where the activation energy
calculatedfrom the Fulcher equation changesdramatically between14fi) and 1600"C(Fie. a). This type of behavior suggestssignificant structural reorganizationand possible polymerization of the melt as temperature
decreases.lt should be noted that similar efects in the
anorthite-rich melts are less pronounced, suggestingthat
the degreeof reorganization in these highly polymerized
melts is small. These observationssuggestcaution regarding the quenching of silicate melts to glass without
concernfor thermalhistory. It may be possibleto quench
significantly different structures dependingon the degree
of superheatingand thermal treatment above the liquidus.
x An -l0O
1Di-20
o D i- 3 0
*D-to
oDi-50
vDi.58
aD-E0
oDi-90
o Di -l0O
?,*
EI
g
l0E7
a
s
!lm
b
E'
g
o
o,
lfob pmnt
Al2Q
Fig. 5. Activationenergyfor viscousflow ys. moleVoAl2O3.
Isothermsat 50"Cintervalsbetween1300-1600"C.
The present results suggestthat these effects will occur
more readily in relatively depolymerized melts such as
diopside.
Conclusions
Viscosity-temperature relationshipsfor the binary join
diopside-anorthite show that the earlier work by Kozu
and Kani (1935)requires revision. We have shown that
viscosity is strongly dependenton the moleVoalumina in
the melt and is also temperaturedependent. Structurally,
the effect of adding diopside to anorthite melt is the
modification or elimination of the three dimensional network structure in favor of a more depolymerizedmelt. As
a function of temperature, implied changesin the structure of diopside melt are much larger than changes in
melts with a feldspathic component. These data, along
with viscosity and structural information presently being
obtained on other parts of the ternary system diopsidealbite-anorthite, will provide a suitable data base for the
understandingof the relationship between melt structure
and properties in this classic petrological system.
Acknowledgments
Scarfeacknowledges
supportfrom the CarnegieInstitutionof
Washington,the University of Alberta and Canadianxsnnc
GrantA8394.This work wascarriedout duringleavefrom the
Universityof Alberta.The hospitalityandaccessto facilitiesat
the Geophysical
Laboratoryand the NationalBureauof Standards,Centerfor MaterialsScience,are appreciated.
T. Fujii,
W. Haller,B. O. Mysenand H. S. YoderJr. are thankedfor
reviewsthat materiallyimprovedthe finalmanuscript.K. Monr
gomeryassistedwith someof the calculations.Experimental
PetrologyLaboratory,Departmentof Geology,Universityof
Alberta,contributionno. 6E.
Prm
-
|lJ
5
I
't
tro
References
0L
r@o
l2OO
I'O0
r50O
l80O
Tanperoture(C')
Fig. 4. Activation energy for viscous flow ys. temperature.
Bockris,J. M. andLowe,D. L. (1954)Viscosityandthe structure of moltensilicates.Transactions
of the FaradaySociety,
51,1734-1748.
Bottinga,Y., andWeill, D. F. (1972)The viscosityof magmatic
l08E
SCARFE ET AL.: VISCOSITY IN THE SYSTEM DIOPSIDE-ANORTHITE
silicate liquids: a model for calculation. American Journal of
Science.272.438-475.
Bowen, N. L. (1915)The crystallizationof haplobasaltic,haplodioritic and related magmas.American Journal of Science,40,
l6l-185.
Cranmer, D. and Uhlmann, D. R. (1981) Viscosities in the
system albite-anorthite. Journal of GeophysicalResearch,E6,
7951-:t956.
Cukierman,J. and Uhlmann, D. R. (1973)Viscosity of liquid
anorthite. Journal of GeophysicalResearch,7E, 49204923.
Fulcher, G. S. (1925) Analysis of recent measurementsof the
viscosity of glasses.Bulletin of the American Ceramic Society, E, 339-355.
Hon, R., Henry, D. J., Navrotsky, A. and Weill, D. F. (1981)A
thermochemicalcalculation of the pyroxene saturationsurface
in the system diopside-albite-anorthite. Geochimica et CosmochimicaActa. 45. 157-161.
Kou, T., Mizoguchi,K. and Suginohara,Y. (1978)The etrectof
Al2O3 on the viscosity of silicate melts. Journal of the Japanese Metallurgical Society, 42, 775-:lEl.
Kozu, S. and Kani, K. (1935)Viscosity measurementsof the
ternary systemdiopside-albite-anorthiteat high temperatures.
Proceedingsof the Imperial Academy of Japan, I I, 383-385.
Kushiro, I. (1973)The system diopside-anorthite-albite: determination of compositions of co-existing phases. Carnegie
Institute of Washington Yearbook 72, 502-507.
Kushiro, I. (1980)Viscosity, density, and structure of silicate
melts at high pressures,and their petrologic applications. In
R. B. Hargraves,Ed., Physicsof MagmaticProcesses,p.93120.PrincetonUniversity Press,Princeton,New Jersey.
Kushiro, I., and Schairer,J. F. (1969)Diopsidesolid solutionsin
the system diopside-anorthite-albite at I atm. and at high
pressures. Carnegie Institute of Washington Yearbook 6E,
222-226.
Mysen,B. O., Virgo, David and Kushiro, I. (1981)The structural role of aluminum in silicate melts----aRaman spectroscopic
study at I atmosphere.American Mineralogist, 66,678-701.
Mysen, B. O., Virgo, David and Scarfe,C. M. (1980)Relations
between the anionic structure and viscosity of silicate meltsa Raman spectroscopic study. American Mineralogist, 65,
690-710.
Murase, T., and McBirney, A. R. (1973)Propertiesof some
common igneous rocks and their melts at high temperatures.
Bulfetin of the Geological Society of America, 84,3563-3592.
Navrotsky, A., Hon R., Weill, D. F. and Henry, D. J. (1980)
Thermochemistry of glasses and liquids in the systems
CaMgSi2O6-CaAl2Si2OrNaAlSi3Or, SiO2-CaAlzSizOrNaAl
Si3Os,and SiOrAlzOr-CaO-Na2O. Geochimicaet Cosmochimica Acta, 44, l4W-1423.
Osborn, E. F. (1942) The system CaSiOr-diopside-anorthite.
American Journal of Science. 240.751:788.
Riebling, E. F. (l%6) Structure of sodium aluminosilicatemelts
contbining at least 50 mole%oSiOz at 1500"C. Journal of
Chemical Physics, 44, 2857-2865.
Scarfe,C. M. (1973)Viscosityof basicmeltsat varyingpressure.
Nature. 241. l0l-102.
Scarfe, C. M. (1977) Viscosity of a pantellerite melt at one
atmosphere.Canadian Mineralogist, 15, lE5-189.
Scarfe,C. M., Mysen, B. O. and Virgo, David (1979)Changesin
viscosity and density of melts of sodium disilicate, sodium
metasilicate,and diopside composition with pressure. Carnegie Institute of Washington Yearbook 78, 547-552.
Schairer, J. F., and Yoder, Jr., H. S. (l!)60) The nature of
residual liquids from crystallization, with data on the system
nepheline-diopside-silica. American Journal of Science,
2584,273-283.
Seifert, F. A., Mysen, B. O. and Virgo, D. (1981)Structural
similarity of glassesand melts relevant to petrologicalprocesses. Geochimicaet CosmochimicaActa, 45. 1879-1E84.
Shaw, H. R. (l%5) Commentson viscosity,crystal settlingand
convection in granitic magmas.American Joumal of
Science, 263, l2O-152.
Shaw, H. R. (1969) Rheology of basalt in the melting range.
Journalof Petrology,10, 510-535.
Spera, F. J. (19E0)Aspects of magma transport. In R. B. Hargraves, Ed., Physics of Magmatic Processes, p. 265-323.
Princeton University Press, Princeton, New Jers€y.
Taylor, M. and Brown, G. E. (1979)Structure of mineral glasses.
I. The feldspar glassesNaAlSi3Or, KAISi3OEand CaAl2Si2O6.
Geochimicaet CosmochimicaActa. 43,61J6.
Virgo, D., Seifert,F. and Mysen, B. O. (1979)Three-dimensional network structures in the system CaAlzOr-SiOz,
NaAlOr-SiOr, NaFeOz-SiOzand NaGaOz-SiO2.CarnegieInstitute of Washington Yearbook 78, 506-51l.
Weill, D. F., Hon, R. and Navrotsky, A. (1980)The igneous
system CaMgSizOe-CaAlzSizOrNaAlSirOe; variations on a
classicthemeby Bowen. In R. B. Hargraves,Ed., Physicsof
Magmatic Processes,p. 49-92. Princeton University Press,
Princeton, New Jersey.
Wyllie, P. J. (l%3) Effects of the change in slope occurring on
the liquidus and solidus paths in the system diopside-anor'
thite-albite. Mineralogical Society of America SpecialPaper |,
204-212.
Yoder, H. S., Jr. (1976)Generationof Basaltic Magma. National
Academyof Sciences,Washington,D. C.
Manuscript received,June 30, l9E2;
acceptedfor publication, March 24, I9E3.