Home Search Collections Journals About Contact us My IOPscience

The measurement of viscosity of alloysa review of methods, data and models

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2005 Meas. Sci. Technol. 16 354

(http://iopscience.iop.org/0957-0233/16/2/005)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 129.128.221.47
This content was downloaded on 25/04/2017 at 19:17

Please note that terms and conditions apply.

You may also be interested in:

Measurement and calculation of the viscosity of metals---a review of the current status and
developing trends
J Cheng, J Grbner, N Hort et al.

Viscosities of FeNi, FeCo and NiCo binary melts

Yuzuru Sato, Koji Sugisawa, Daisuke Aoki et al.

An absolute oscillating-cylinder (or cup) viscometer for high temperatures

K Torklep and H A Oye

Viscosity Determination by the Oscillating Vessel Method I: Theoretical Considerations

R Roscoe

Capillary viscometer with a pressure sensor: a subject for student projects

Taha Massalha and Rafael M Digilov

The Viscosity of Liquid Indium and Liquid Tin

M F Culpin

The viscosity of eutectic Pd-Si alloys

I Egry

Rotational viscometersa subject for student projects

Yaakov Kraftmakher

Viscometers having damped torsional oscillation

B S Ashwin, T Hagyard, I C B Saunders et al.
INSTITUTE OF PHYSICS PUBLISHING MEASUREMENT SCIENCE AND TECHNOLOGY
Meas. Sci. Technol. 16 (2005) 354362 doi:10.1088/0957-0233/16/2/005

The measurement of viscosity of alloysa

review of methods, data and models
R F Brooks, A T Dinsdale and P N Quested
NPL Materials Centre, National Physical Laboratory, Teddington, TW11 0LW, UK

Received 9 August 2004, in final form 8 November 2004

Published 19 January 2005
Online at stacks.iop.org/MST/16/354

Abstract
Values of the viscosities of liquid metals are important in the prediction of
fluid flow in many metallurgical manufacturing processes. This paper
describes a number of methods used to measure the viscosity of liquid
metals, including capillary, oscillating vessel, rotational bob or crucible,
oscillating plate, draining vessel, levitation using the damping of surface
oscillations and acoustic methods. A number of models used to estimate
viscosity for elements, the temperature dependence of viscosity, and
viscosity of multicomponent systems are also given, including the Andrade
equation, Arrhenius equation, Hildebrands free volume theory, Chhabra
models, Moelwyn-Hughes model and thermodynamic models. The scatter
of data available in the literature are highlighted by comparing two reviews
of data for elements.
Keywords: viscosity, liquid metals, elements, alloys, models, estimation of
viscosity, measurement methods
(Some figures in this article are in colour only in the electronic version)

1. Measurement of viscosity of liquid metals modified Poiseuille equation or HagenPoiseuille equation as

Values of the viscosities of liquid metals are important in the
prediction of fluid flow in many metallurgical manufacturing
processes. There are a large number of methods to measure
the viscosity of liquids, but those suitable for liquid metals
are limited by the low viscosities of metals (of the order of
1 mPa s), their chemical reactivity and generally high melting
points. Proposed methods include: capillary, oscillating
vessel, rotational bob or crucible, oscillating plate, draining
vessel, levitation using the damping of surface oscillations and
acoustic methods. However, despite the number of possible
methods, there are few data available for pure metals, and even
less for alloys, and literature values can show a high degree of
scatter.
1.1. Capillary method
The capillary rheometer is generally thought to be the best
method for the measurement of the viscosity of liquids [1]
and is based upon the time for a finite volume of liquid to
flow through a narrow bore tube under a given pressure. The
relation between viscosity and efflux time is given by the
follows:
r 4 ght mV
= (1)
8V (l + nr) 8(l + nr)t
where r and l are the radius and length of the capillary,
respectively, h is the effective height of the column of liquid,
is the liquid density, V is the volume discharged in time t, and m
and n are constants which can be determined experimentally.
gh may be replaced by P, the pressure drop along the
capillary, and nr is called the end correction, and corrects
for surface tension effects as the liquid is expelled from
the capillary. For liquid metals, with their relatively high
densities, the second term, which corrects for kinetic energy,
is particularly important. An example of a capillary viscometer
is given in figure 1.
This technique is often used as a relative, rather than
absolute, method, as the experimental procedures are simple,
and any errors incidental to the measurement of dimensions
are thereby avoided. For a viscometer in which r, l, h and
V are fixed, equation (1) reduces to
C2
= C1 t (2)
t

0957-0233/05/020354+09$30.00 2005 IOP Publishing Ltd Printed in the UK 354

 The measurement of viscosity of alloys

Positional
Adjuster
sensor wires

Gas Inlet
(O2 free)
known volume of liquid
Oscillation
Initiator
alumina seal
capillary tube Coupling

Torsion Wire
Supsension
3 zone furnace
cone to stop surface Temperature
tension effects Controlled
Jacket
catch pot (30C)

Mirror
water cooled sealed end piece
Window

Figure 1. Diagram of a capillary viscometer.

Laser Beam

where the values of Cl and C2 are easily evaluated using Diode Array
viscosity standard reference samples, but are constants equal
to Thermocouple

r 4 gh mV Water Cooled
C1 = , C2 = . (3) Section (Cold)
8V (l + nr) 8(l + nr)
In determining the viscosities of metallic liquids by the
capillary method, an especially fine and long-bore tube (in
general, r < 0.150.2 mm, l > 7080 mm) is needed so as to Suspension Rod
satisfy the condition of a low Reynolds number for ensuring
laminar flow. This in turn requires a furnace with a similarly Alumina Enclosure
long and uniform hot zone. Blockage of the capillary by
bubbles or oxide inclusions is a common problem, particularly Container
with aluminium alloys, and due to materials problems a
temperature limit of about 1200 C is often imposed, but metals Furnace
such bismuth have been successfully measured [1].

1.2. Oscillating vessel viscometer

Most measurements of the viscosity of metals use some form of
oscillating vessel viscometer. A liquid contained in a vessel,
normally a cylinder, is set in motion about a vertical axis
and the motion is damped by frictional energy absorption and
dissipation within the liquid. The viscosity is determined
from the decrement and time period of the motion. The
main advantages of the method are that the time period and
decrement are easily measured and the amount of liquid is
relatively small, which allows stable temperature profiles to
be attained. One of the major difficulties is relating the
measured parameters to the viscosity through the second-order
differential equation for the motion of an oscillating system
and there are a number of mathematical treatments appearing
to yield different results [26] with the same experimental
data. Since this is the most commonly used method it will be
discussed in more detail below.
Figure 2 shows the experimental arrangement for an
oscillating vessel viscometer and details of this particular
design are given in [7]. The sample is contained within an
Figure 2. Diagram of the NPL oscillating viscometer.
alumina crucible (105 mm long by 14 mm internal diameter),
which is screwed into a molybdenum lid and suspension rod
and suspended on a torsion wire. A rotary solenoid is used to
impart oscillatory motion to the crucible, and an optical pointer
with a diode array is used to measure the time constant and
decrement of the system. The sample is heated by a two-zone
furnace. These authors claim an uncertainty of measurement
of 9% within a 95% confidence limit. A major contribution
to the uncertainty is the extrapolation of the dimensions of the
crucible and the height of liquid at high temperatures.
For a right circular cylinder that is infinitely long
containing a fluid the equation of motion of the damped
cylinder is
I0 (d2 /dt 2 ) + L(d/dt) + f = 0. (4)
I0 is the moment of inertia of empty cup and suspension,
t is the time, f the force constant of the torsion wire, is
the angle of displacement of any small segment of the fluid

355
R F Brooks et al

from its equilibrium position. L is a function of the density

and viscosity of the fluid; the internal radius of crucible and
height of liquid expressions for L are determined by solving the
NavierStokes equations for the motion of the liquid within the
vessel (neglecting nonlinear terms). The working formulae for
the oscillating cup viscometer reviewed by Iida and Guthrie
[1] are from Knappworst [2], Shvidkovskiy [3] and Roscoe
[4]. The majority of measurements made by this method have
used the analysis by Roscoe, mainly for reasons of simplicity
of presentation of the working formulae of that reference and
following other workers in the liquid metals field. A more
complete model of oscillating cup viscometers was described
by Kestin and Newell [5] and Beckwith and Newell [8]. The
resulting working equations were not stated explicitly in the
early references and required some further analysis for their
derivation. These have been adopted by chemical engineers
and it has been suggested [9, 10] that the BeckwithNewell
model is more accurate and comprehensive than that of Roscoe
and should therefore be adopted for future analysis.

1.2.1. The Roscoe analysis. Roscoe [4] describes the Figure 3. Reproduced from Beckwith and Newell [8], showing
derivation of the necessary formulae for determining viscosity region of interest 8 < 0 < 23, 68 < 0 < 200 to be within the large
from oscillatory spherical and cylindrical viscometers. The cup regime.
decay of the oscillations was determined from

    these values for the NPL design correspond to the large cup
R3H 1 1R 3 4 R 1 regime.
I = (t) 2 1+ a0 +
2 4H 2 H p The corresponding operating equation due to Beckwith,
    
3 9R a2 63 45 R a4 in the form of Brockner et al [6], is
+ + (5)

8 4 H 2p 2 128 64 H 4p 4 H R 4 1 B Cq
A(p q) 2 + 3 = 2( o /) (9)
where 2I x x x
  12   12 where
(1 + 2 ) + 1 1 + 2 1 R 16 R 3 9R
a0 = (1 ) (1 + ) A=4+ , B =6+ , C= + ,
2 2 H H 2 H
3 3 1 to  1

1
= 1  2  3 (6) = , p = 1 2  + (1 + 2 ) 2 2 ,
2 8 16 t
  12   12 1 1
1 + 2 + 1 1 + 2 1 q= , x = R(2/(t)) 2
a2 = + 2p
2 2
where to is the oscillation period and 2o is the logarithmic
 1 2 1 decrement for the empty cup. The other quantities are as
=1+ +  3 (7)
2 8 16 defined in the previous section.
2 12  3 5
a4 = a2 /(1 +  ) = 1 + 2 3 (8) 1.2.3. Numerical comparisons. Comparing the treatments
2 8 16
with a typical set of NPL measurement data for iron, the
and corrected Roscoe equation (2.5) gives 6.047 mPa, while the
1
p = (/t) 2 R. corresponding value calculated using the BeckwithNewell
analysis (3.2) yields 6.044 mPa. For the aluminium data, the
In (11), R is the cylinder radius, H the height of the liquid viscosity results for the Roscoe and BeckwithNewell analyses
within it, I the moment of inertia of the suspension, 2 are 0.949 mPa s and 0.954 mPa s, respectively. The difference
the logarithmic decrement between consecutive swings, the between the results for aluminium is about 0.5% and that for
fluid density, the viscosity and t is the period of oscillation. steel very much less.
In his review of the equations Ferriss [9] found that (12)
was printed incorrectly in the original paper, and should be 1.2.4. Possible errors. Comparison of the oscillating
read as (3/2), as given above. viscometer with the capillary viscometer reveals some
differences. Iida and Guthrie [1] suggest the end effect is
1.2.2. The Beckwith, Kestin and Newell analysis. The inadequately weighted in the Roscoe treatment and should
analysis due to Kestin and Newell [5] and Beckwith and also include the effect of the liquid meniscus on the height
Newell [8] identifies the small, intermediate and large cup of liquid, suggesting that a correction factor, , is introduced
regimes according to the magnitude of certain dimensionless into the (uncorrected) formula of 1 0.04. Various studies
parameters (figure 3). Ferriss et al [9] have shown that [1, 1114] have suggested that wetting of the crucible may also

356
 The measurement of viscosity of alloys

Viscous Resistance = 3 DV(1+3/16Re) sine wave

power amplifier generator
balance
h = viscosity
oscillator
D = diameter of sphere
V = velocity of movement
Re = VD / leaf spring
target

platinum suspension wire laser

displacement voltmeter
furnace
transducer
alumina tube

alumina cap
computer

platinum sphere computer

platinum crucible

thermocouple
oscillating plate

Figure 4. Diagram of a counterbalanced sphere viscometer.

be important, and that if the metal does not wet the crucible
it may slip during the oscillation and thus provide greater
damping. The aspect ratio of the sample may be important in
both these assessments, and further work is required to justify
modification of the (Roscoe) equation.
1.3. Falling body or counterbalanced sphere viscometer
In this method a body (normally a sphere) is either allowed
to drop due to gravity or dragged upwards through the liquid.
This can be achieved with the apparatus shown in figure 4.
The viscosity is calculated using Stokes law:
2gr (k liq )
2
= (10)
9S
where g is the gravitational constant, r is the radius of the
sphere, k and liq are the density of the sphere and the liquid
respectively and S is the velocity of descent or ascent of the
sphere. It is normal to correct for the expansion of the ball [15]
and the diameter of the ball should be less than 1/10 of the
diameter of the crucible [16]. This is not an absolute method
requiring an apparatus constant to be derived from calibration
curves with liquids of known viscosity. The method was
successfully demonstrated by Endell et al [17, 18] for slag
and glasses by the time taken to pull a ball from a melt at a
constant measured force.
Hara et al [19] used the counterbalance sphere of Pt
20% Rh to determine the viscosity of an oxide system in air,
but for melts with a low viscosity (Reynolds number Re <
0.1) Stokes law does not hold and a modified equation must
be used [19]. Although the counterbalanced sphere reduces
the experimental difficulties associated with the falling body
method in a dense, opaque liquid metal, the authors have found
no reference on its use for metals.
1.4. Oscillating plate method
Figure 5. Diagram of an oscillating plate viscometer.
derived [23, 24]:
 2
RM 2 fa E a
= 1 = Ko (11)
f A2 fE
where is the density, Ea and E are the amplitudes of
oscillation in air and the liquid respectively, fa and f are the
frequency of the oscillation in air and the liquid, A is the area
of the plate and RM is the real component of the mechanical
impedance. The assumptions of the method are:
1. The liquid is a Newtonian fluid.
2. Turbulent flow does not occur.
3. There is no slippage between the plate and the fluid.
4. The size of the oscillating plate must be larger than one
wavelength of the oscillation wave and is planar.
5. The end effects of the plate on damping of the amplitude
are negligible.
6. The size of the vessel is large enough that the effects of
the waves reflected from the walls are negligible.
It has been shown that the resonant frequency in air and
in liquid can be considered to be the same for all practical
purposes so that
 2
RM 2 Ea
= 1 = K. (12)
f A2 E
2
= f RM
A2
is the damping factor and K = (Ea /E) 1
is determined experimentally using viscosity reference
materials. Recent comparisons [25] of this technique with the
rotating cylinder method, for slag systems, have shown good
agreement, typically within 7% [20]. Although in theory
a simple method, for low viscosity metals a thin oscillating
plate of large area must be vibrated slowly within the liquid.
As far as the authors are aware, this method has not been used
to measure the viscosity of liquid metals.

This method [2022], shown in figure 5, relies on vibrating

a flat plate in a liquid with a constant driving force. The
amplitude of motion will be dependent on the viscosity of the
liquid. The amplitude of the plate oscillations is measured in
air and in the liquid and the following relationships have been
1.5. Rotating cylinder
For the rotating cylinder technique the torque on a cylinder
rotated in a liquid is related to the viscosity of the fluid.
Viscometers of this type consist of two concentric cylinders,

357
R F Brooks et al
Viscometer head
articulated
Mo spindle
sample
Mo bob
furnace
Mo crucible
thermocouple
Figure 6. Diagram of a rotating bob viscometer.
i.e. a bob and a crucible. The viscosity is determined
from measurements of the torque generated on the rotor arm
of the rotating cylinder. When rotating the cylinder at a
constant speed the viscosity can be obtained from the following
equation:
 
1 1 M is claimed that because the principles and mechanism of this
= 2 (13)
r12 r0 8 4 nh
where M is the torque, n is the number of revolutions per
second, r1 is the radius of the bob, r0 is the radius of the
crucible, and h is the height of the bob. The theory is applicable
to infinitely long cylinders and it is normal to calibrate the
system with reference materials using the equation
= G = S/n (14)
where S is the scale deflection and G is the apparatus constant.
This is the most common method for the measurement
of slag viscosity and for several practical reasons it is more
customary to rotate the bob rather than the crucible. A typical
rotating bob apparatus is shown in figure 6. Probably the most
important reason is that the rotating bob viscometer is based
upon readily available and cheap commercial instruments. It
is also easier to centre the bob, but the viscosity range is less
than for a rotating crucible instrument. It is critical that the bob
rotates axisimetrically and concentrically within the sample.
Any instability due to rotation speed or alignment will increase
the apparent viscosity. Nakashima et al [26] show a modern
example of a rotating crucible method.
In order to obtain the necessary sensitivity to measure
the low viscosity of liquid metals the clearance between the
stationary and rotating parts has to be made very small, and
it is difficult to maintain the system coaxially. In spite of the
experimental difficulties the rotating bob technique has been
used to measure the viscosity of aluminium and its alloys [27].
An adaptation of this method has been developed as part
of a spiked parallel plate creep/rotational viscometer [28],
shown in figure 7. This elegant design to measure a wide range
of viscosities from about 1012 to 101 Pa s has been successfully
demonstrated with various glasses to a temperature of 1200 C.
The instrument combines an indentation method for the highest
viscosities, a creep parallel plate method for the intermediate
358
displacement
transducer coil
core holder
load
load cell
pully
x-y stage
reference rod
(for expansion)
spiked disk
sample
rotary disk
furnace
servomotor lifting mechanism
Figure 7. Spiked parallel plate creep/rotational viscometer.
range, 107 to 104 Pa s, and a rotating plate method for the lower
range. The plate acts in the same way as the rotating cylinder,
with a large surface area to obtain the necessary sensitivity. It
viscometer are simple, the absolute values of sample viscosity
can be determined, at least to the right order, without any
calibration using standard reference samples. There are no
reports of it being used on liquid metals.
1.6. Industrial viscometers
Several simple viscometers have been developed for on
plant determination of slag viscosities and quality assurance
exercises, but these have not been used for liquid metals.
Examples are the Herty viscometer [29] where the viscosity
is determined by the length of the slag ribbon formed, and
the Krabiell immersion viscometer [30, 31]. In recent years
the inclined plane method has been investigated in more
detail [32]. In this method a molten slag is poured from a
crucible onto an inclined steel plate plane with a V section.
The viscosity can be related to the length of the slag ribbon
obtained. Mills et al [32] claim that measurement uncertainties
of 15%, for slags with viscosities in the range of 0.1 to
1 Pa s, are possible. The oscillating plate method has also
been proposed as an in situ measurement.
1.7. Draining vessel method
It is common for comparative measurements of viscosity to be
made for oils and slurries in industrial applications by use of
a flow cup, where the time taken for a volume of sample to
flow through a small orifice in the bottom of a cup is measured
and viscosity is derived from look-up tables. Roach et al [33]
have derived equations to adapt this method for liquid metals,
and to provide values of viscosity, surface tension and density.
The apparatus is shown in figure 8. The values for aluminium
are lower than usually quoted, but the method is robust and
experiments simple to perform.

thermocouple
induction coil
h 570
Q r0 stopper
load cell
Figure 8. Diagram of a draining vessel viscometer.
1.8. Oscillating (levitated) drop method
The oscillating drop method is widely used for surface tension
measurements of liquid, levitated samples. It is based on the
fact that the frequencies of the surface oscillations of a liquid
drop are related to the surface tension by Rayleighs formula
[34]; for earthbound levitation, his formula must be corrected
as proposed by Cummings and Blackburn [35]. For a viscous
drop, these oscillations are damped due to the viscosity of the
liquid. Therefore, it is, in principle, also possible to determine
the viscosity from the damping. The damping constant  is
given by
20 R
= (15)
3 m
where R and m are radius and mass of the droplet, respectively,
and is the viscosity. In order to apply the above formula, the
following conditions must be satisfied:
The liquid drop must be spherical.
The oscillations must persist undisturbed for a time,
t = 1/ .
There must be no additional damping mechanism present.
It is often thought that none of these conditions is met
in terrestrial experiments; the sample is deformed, and
there is an overlap of self-excited oscillations, making the
decay of one single excitation hard to detect. Also, the
electromagnetic fields necessary for earthbound levitation
inevitably induce turbulent fluid flows inside the sample which
lead to additional damping, although Rhim et al [36] and
Ishikawa et al [37] have published measurements of refractory
metals made terrestrially using this method and electrostatic
levitation. Under microgravity conditions, the fields are
much weaker, and a laminar fluid flow can be expected, at
least for sufficiently viscous materials. In microgravity an
electromagnetic positioning field is used, with a separate
heating coil. A pulse of power initiates the oscillations
which are monitored with a video camera. The temperature
is recorded with a pyrometer. Image analysis is used to
obtain the frequency and decay of the oscillations as shown in
figure 9. Microgravity experiments utilizing the oscillating
drop technique were carried out on liquid gold in 1994. This
The measurement of viscosity of alloys
600
PdCuSi shape oscillations at 950 C
590
580
amplitude [a.u.]
560
550
540
0 1 2 3 4 5 6 7 8
time [s]
Figure 9. Damping of shape oscillations at 950 C.
material has a low viscosity and unphysically high results
for the viscosity were obtained, indicating turbulent flows in
the sample. During the MSL-1 Spacelab mission STS-83,
the same experiment was carried out on the highly viscous
eutectic system Pd78Cu6Si16 [38]. The values agree well with
one set of previously published data for the viscosity of this
alloy near the melting point, although there is debate as to the
true viscosity of this system. More recently Wunderlecht et al
[39] have carried out measurements on parabolic and sounding
rocket flights, on the nickel based superalloy CMSX-4 which
show reasonable agreement with terrestrial measurements of
the same alloy using an oscillating cylinder method.
1.9. Damping of an acoustic wave
The viscosity of a liquid can be measured by measuring the
damping of an acoustic wave. There is one example for the
measurement of aluminium [40] and some alloys, using hole
theory to interpret the results.
2. Viscosity data for elements
There are several compendia of viscosity of elements in the
literature, and each shows scatter in measured values. It is
likely that as well as differences in experimental technique,
values are dependent on the purity of the sample. To illustrate
the expected values for common elements found in commercial
alloys table 1 lists data from two compendia [41, 42], listing the
viscosity at the melting point and the temperature dependence
in the Arrhenius form
(T ) = o exp(E/RT ). (16)
E is the activation energy for viscous flow and o is the pre-
exponential viscosity, which are both constants. T is the
temperature in K, R is the gas constant = 8.3144 J mol1 K1.
Table 1 lists the values for the two compendia for some
relevant elements. Clearly there are discrepancies between
the values and this again reflects the difficulty of obtaining
accurate data. In the Iida and Shiraishi [41] compendium
some elements are not quoted whereas Battezati and Greer
[42] make clear that the values for some refractory elements
(e.g. Cr) are uncertain and in other cases large discrepancies
in the values exist in the literature (e.g. Mn).
359
R F Brooks et al

Table 1. Data for the viscosity of relevant elements drawn from two compendia [41, 42].
Melting point (Tm) o E (Tm) [2]
Element Tm ( C) (mPa s) (mPa s) (kJ mol1) (kg m3)

Al 660 1.38 0.257 13.08 2380

1.30 0.149 16.5
4.38 0.529 23.85
Cu 1083 8000
4.0 0.301 30.5
Fe 1535 6.93 0.315 46.5 7030
5.5 0.370 41.4
Mg 650 1590
1.25 0.025 30.5

Mn 1244 5760
5 0.121.02 2046.5
Si 1414 2530
0.8 0.060.12 2736
Zn 419 3.50 0.527 10.91 6580
3.85 0.413 12.7
4.27 0.589 20.34
Ag 960 9300
3.88 0.453 22.2
Cr 1905 6290
5.7 1.7 1014 185

Ni 1455 7900
4.9 0.166 50.2
1.87 0.467 5.83
Sn 231 6980
1.85 0.538 5.4
2.2
Ti 1725 2.2 0.034 68 4130
5.2

3. Models for viscosity of metals
The following section presents some of the models that
are used to describe the viscosity of the elements and
multicomponent systems. The list is not comprehensive
and we have not included molecular models or relationships
between liquid diffusion coefficients and viscosity such as the
StokesEinstein equation.
3.1. Elements
3.1.1. Andrade equation for viscosity at melting point. This
model [43] is based upon the supposition that the characteristic
vibration frequencies in the liquid (L ) and solid (S ) near the
melting point are equal using the fact that the specific heats of
liquid and solid metals at their melting point are similar.
(Tm ) = CA {Tm A}1/2 /V 2/3 . (17)
Andrade showed that CA is approximately a constant for
pure metal melts and estimated a value of 1.65 107
(J K1 mol1/3 )1/2 with A = atomic weight and V = molar
volume at Tm.
Battezati and Greer [42] have tested Andrades equation
and note several exceptions to the predicted value of CA due
to: the reliability of available data for some elements; the
formation of associates in elements such as selenium; and the
anomalous behaviour of semiconductors in the solid and in
the liquid.
3.1.2. Temperature dependence of viscosity of the elements.
There are a large number of equations that have been derived
to describe the viscosity as a function of temperature for
pure liquids. (See Chhabra and Seth [44] and Beyer and
Ring [45].) These frequently involve two or three adjustable
parameters. Chhabra investigated three common models:
Arrhenius, Andrade and the Hildebrand fluidity equation, and
compared them against experimental data.
3.1.3. Arrhenius equation. The Arrhenius equation is the
one most frequently used to represent data on viscosity:
= A exp(E/RT ). (18)
T is the temperature in K, A is the pre-exponential factor, R
is the gas constant and E is the activation energy for viscous
flow.
Chhabra et al [44] derive values for A and E from various
experimental data for Ag, Al, Au, Bi, Ca, Cd, Co, Ca, Cu, Fe,
Ga, Ge, Hg, In, K, Li, Mg, Na, Ni, Pb, Pu, Rb, Sb, Sn, Tl
and Zn.
3.1.4. Andrade treatment. The equation by Andrade [43]
described above underestimates the temperature dependence
of viscosity, and so Andrade suggested that the viscosity was
lowered by the loss of local molecular orientational order but
there is a discussion whether this is a tenable argument [41].
Andrade gave an equation in the form
(T ) = C1 / 1/3 exp(C2 /T ) (19)
where C1 and C2 are constants and is the specific volume.
This is similar to the Arrhenius equation [42, 44].
Hirai [46] modified the Andrade equation based upon
characteristic frequency of vibration to give
(Tm ) = 1.7 107 2/3 Tm 1/2 M 1/6 (Pa s) (20)
where is the density, M is the atomic weight. Using
a compendium of data [47] he derived the following

360

relationships for the viscosity of the elements:
= A exp(E/RT ) (21)
where
A = {1.7 107 2/3 Tm 1/2 M 1/6 }/A exp(E/RTm ) attempts to model viscosity using thermodynamic concepts
and
E = 2.65 Tm 1.27 .
Although the expressions describe the viscosity of the elements
well and have formed the basis of the unary data for at least
one model [48] extension of these equations to alloys has not
met with success.
3.1.5. Hildebrands free volume theory. The principle of this
model [49] is that fluid flow is governed by the extent of the
free space available:
1/ =  = B{(V V0 )/V0 }. (22)
V is the atomic volume, V0 is the intrinsic volume where flow
is stopped, B is a characteristic constant. Both B and V0 are
considered constants independent of temperature.
3.1.6. Chhabra model. Mehrota [50] developed a model
based on the viscosity values of organic liquids following
modifications of equations by Walther [51]:
log( + 0.8) = 10 b1T b2 . (23)
A strong correlation was shown between b1 and b2 for light and
medium organic liquids. Chhabra and Tripathi [52] extended
the model to liquid metals and derived an equation:
log( + 1) = 10 b1T b2 . (24)
They were unable to demonstrate a strong correlation between
b1 and b2 but claim that the equation gives improved
description of viscosity values over the Arrhenius equation.
3.2. Multicomponent systems
3.2.1. Moelwyn-Hughes model [53]. The viscosity of a
binary system is given by
The measurement of viscosity of alloys
at two temperatures there was a maximum difference of 8%
between the predicted and measured values.
3.2.3. Thermodynamic models [4648]. There have been
[48, 55, 56]. Kucharski [55, 56] developed two models: one
based upon rate theory to describe multicomponent mixtures
where all the parameters have physical meaning and another
model predicting multicomponent mixtures from knowledge
of thermodynamic properties, densities and viscosities for the
limiting binaries. She tested this for a AgCuAu system. Du
Sichen et al [48] used an absolute reaction rate theory and were
able to reproduce the viscosity of some binary compositions
well.
Acknowledgments
This work formed part of the Measurements for
Processability Programme of the UKs Department of Trade
and Industry.
References
[1] Iida T and Guthrie R I L 1988 The Physical Properties of
Liquid Metals (Oxford: Clarendon) (ISBN 0-19-856394-9)
[2] Knappworst A 1952 A new method of high temperature
viscometry by the method of oscillating hollow bodies
Z. Phys. Chem. 200 81
[3] Shvidkovskiy Ye G 1962/1955 Certain Problems Related to
the Viscosity of Fused Metals (NASA Technical Translation
F-88) translation of Netotoryye voprosy vyazkosti
rasplavlennykh metallow (Moscow: State Publishing House
for Technical and Theoretical Literature)
[4] Roscoe R 1958 Viscosity determination by the oscillating
vessel method: 1. Theoretical considerations Proc. Phys.
Soc. 72 57684
[5] Kestin J and Newell G F 1957 Theory of oscillation type
viscometers. The oscillating cup Z. Angew. Math. Phys. 8
433
[6] Brockner W, Torklep K and Oye H A 1979 Viscosity of
aluminium chloride and acidic sodium chloroaluminate
melts Ber. Bursenges Phys. Chem. 83 111
[7] Brooks R F, Day A P, Andon R J L, Chapman L A, Mills K C
= (xA A + xB B )(1 2xA xB /RT ), (25) and Quested P N 2001 Measurement of the viscosities of
where A and B are viscosities of elements, xA and xA the metals and alloys with an oscillating viscometer High Temp.
mole fractions,  the regular solution interaction parameter.
Equation (25) predicts a negative deviation of the viscosity
for systems having a positive heat of mixing, Hmix, in the
liquid. It appears to be successful for some metal systems
but does not predict the lower viscosities expected at eutectic
compositions. Iida and Guthrie [1] modified the model to
include the effect of the size of the atoms.
3.2.2. Chhabra model. Chhabra [54] modified his model to
extend it from elements to mixtures and modified the b1 and
b2 coefficients:
 
b1,alloy = xi b1,i , b2,alloy = xi b2,i (26)
log( + 1) = 10b1,alloy T b2,alloy . (27)
He applied the model to 29 binary systems and one ternary
system. For most binaries it predicted the viscosity within
10% although there were some exceptions. The worst case
was SnMg with a 90% difference. For the PbSnSb ternary
High Press. 33 73
[8] Beckwith D A and Newell G F 1957 Theory of oscillation type
viscometers. The oscillating cup: II Z. Angew. Math. Phys. 8
450
[9] Ferriss D H, Quested P N, Chapman L A and Day A P 2002
The choice of equations for the measurement of viscosity
by the oscillating cylinder method Presented at ECTP
(London)
[10] Wang D and Overfelt R A 2002 Oscillating cup viscosity
measuremnts of aluminium alloys: A201, A319 and A356
Int. J. Thermophys. 23 106376
[11] Kimura S et al 1995 Proc. 4th Asian Thermophys. Prof. Conf.
(Toyko, September) Paper A1a1
[12] Kimura S and Terashima K J 1995 Density measurement of
molten silicon by an improved Archimedean method Cryst.
Growth 139 225
[13] Sasaki H, Tozizaki E, Huang X, Terashima K and Kimura S
1995 Temperature dependence of the viscosity of molten
silicon measured by the oscillating cup method Japan. J.
Appl. Phys. 34 3432
[14] Sato Y et al 2003 Viscosity of molten silicon and the factors
affecting measurement J. Cryst. Growth 249 404
[15] Ladenberg R 1907 Viscosity of liquids Ann. Phys. 22 287
361
R F Brooks et al
[16] Bacon L R 1936 Absolute viscosity by the falling sphere
method Franklin Inst. J. 221 251
[17] Endell K, Tielsch A and Wens C 1934 Uber die
Temperatur-viskosisitatsbeiziehungen einiger Bleischlacken
Metall. U. Erz 31 353
[18] Endell K, Kley D and Stahl U 1939 Ueber die Abhangikeit der
Temperatur-Zahigkeit-Beziehungen saurer
Hochofenschlacken von der Chemischen
Zusammensetzung Eisen 59 677
[19] Hara S, Inoue K and Tanaka T 1997 Viscosity of molten slags
containing iron oxide Molten Slags, Fluxes and Salts 97
Conference (Sydney, Australia, 58 Jan. 1997) (Warrendale,
PA: Iron and Steel Society) p 515 (ISBN 1-886362-14-9)
[20] Tokuda M, Nakajima K, Yamamoto T, Nakajima H and
Kawaguchi K 1990 Development and utilization of a new
oscillating-plate type viscometer for casting powder
evaluation The Sumitomo Search no 44 p 368
[21] Iida T, Tanaka T and Koike M 1993 Effect of reflected wave
from the wall of a vessel and end effect of the plate in the
oscillating-plate method ISIJ 33 218
[22] Sasahara S, Inaba S and Tomita M 1992 New oscillating plate
viscometer for instantaneous measurement of viscosity of
molten slag 4th Int. Conf. on Molten Slags and Fluxes
(Sendai, Japan, 811 June 1992) (Japan: Iron and Steel
Institute) p 456
[23] Morita Z, Iida T, Kawamoto M and Mohri A 1984 A study on
an oscillating-plate viscometer for instantaneous and
simultaneous measurements of the viscosity of melts and
the noise with their chemical reactions Tetsu-to-Hagane 70
1242
[24] Iida T, Kawamoto M, Fujimoto S and Morita Z 1985 An
experimental investigation on characteristics of the new
oscillating-plate viscometer Tetsu-to-Hagane 71 1490
[25] Mills K C, Chapman L A, Fox A B and Sridhar D 2001 Round
robin project on estimation of slags viscosities Scand. J.
Metall. 30 396403
[26] Nakashima K, Kawagoe T, Ookado T and Mori K 1997
Viscosity of binary borate and ternary borosilicate Molten
Slags, Fluxes and Salts 97 Conf. (Sydney, Australia,
58 Jan. 1997) (Warrendale, PA: Iron and Steel Society)
p 215 (ISBN 1- 886362-14-9)
[27] Jones W R D and Bartlett W L 1952/1953 The viscosity of
aluminium and binary aluminium alloys J. Inst. Met. 81
14552
[28] Shiraishi Y, Nagasaki S and Yamashiro M 1997 Spiked
parallel plate creep/rotation viscometer and its
characteristics ISIJ 37 383
[29] Herty C H 1934 Neerungen im Siemens-Martin Betrieb
Stahl u. Eisen 54 609
[30] Herty C H 1936 Die physikalische chemie der Stahlerzeugung
Stahl u. Eisen 56 165
[31] Krabiell H J 1944 Entkohlungsgeschwindigkeit und
Sauerstoffgehalt des Stahles im Basichen
Siemens-Martin-Ofen Stahl u. Eisen 64 399
[32] Mills K C, Halali M, Lorz H P, Kinder A, Pomfret R and
Walker B 1997 A simple test for the measurements of slag
viscosity Molten Slags, Fluxes and Salts 97 Conf. (Sydney,
Australia, 58 Jan. 1997) (Iron and Steel Society) p 535
(ISBN 1-886362-14-9)
362
[33] Roach S J, Henein H and Owens D C 2001 A new technique to
measure dynamically the surface tension, viscosity and
density of molten metals Light Metals 2001 ed J L Aujier
(Warrendale: TMS) pp 128591
[34] Rayleigh L 1879 On the capilliary nature of jets Proc. R. Soc.
A 29 71
[35] Cummings D L and Blackburn D A 1991 Oscillations of
magnetically aspherical droplets J. Fluid Mech. 224 395
[36] Rhim W-K, Ohsaka K, Paradis P-F and Spjut R E 1999 Non
contact technique for measuring surface tension and
viscosity of molten materials using high temperature
electrostatic levitation Rev. Sci. Instrum. 70 2796801
[37] Ishikawa T, Paradis P-F, Itami T and Yoda S 2005
Non-contact thermophysical property measurements of
refractory metals using an electrostatic levitator Meas. Sci.
Technol. 16 44351
[38] Egry I, Lohofer G, Seyhan I, Schneider S and Feuerbacher B
1998 Viscosity of eutectic Pd78Cu6Si16 measured by the
oscillating drop technique in microgravity Appl. Phys. Lett.
73 4623
[39] Wunderlecht R et al 2004 Presented at 2nd Int. Symp. on
Physical Sciences in Space (Toronto, Canada, 2327 May
2004) Microgravity Sci. Technol. J. submitted
[40] Nozdrev V F, Stremousov V I and Takuchev V V 1979 Russ. J.
Phys. Chem. 53 6779
[41] Iida T and Shiraishi Y 1988 Chapter 4 viscosity Handbook
of Physico-Chemical Properties at High Temperatures
ed Y Kawai and Y Shirashi (Japan: Iron and Steel Institute)
[42] Battezati L and Greer A L 1989 The viscosity of liquid metals
and alloys Acta Metall. 37 1791802
[43] da Andrade E N 1934 The theory of the viscosity of liquids: I
Phil. Mag. 17 497
[44] Chhabra R P and Seth D K 1990 Viscosity of molten metals
and its temperature dependence Z. Metallkde 81 26471
[45] Beyer R T and Ring E M 1972 Liquid Metals: Chemistry and
Physics (New York: Marcel Dekker)
[46] Hirai M 2002 Estimation of viscosities of liquid alloys ISIJ 33
2815
[47] Mills K C 2002 Recommended Values of Thermophysical
Properties for Selected Commercial Alloys (Cambridge:
Woodhead Publishing Ltd) (ISBN 1 85573 569 5)
[48] Sichen Du, Bygen J and Seetharaman S 1994 A model for
estimation of viscosities of complex metallic and ionic
melts Metall. Trans. B 25 519
[49] Hildebrand J H 1977 Viscosity and Diffusivity: A Predictive
Treatment (New York: Wiley)
[50] Mehrota A K 1991 A generalised viscosity equation for pure
heavy hydrocarbons Ind. Eng. Chem. Res. 30 1367
[51] Walther C 1931 Erdol Teer 7 382
[52] Chhabra R P and Tripathi A 1993 A correlation for the
viscosity of liquid metals High Temp. High Press. 25 713
[53] Moelwyn-Hughes E A 1961 Physical Chemistry 2nd edn
(Oxford: Pergamon)
[54] Chhabra R P J 1995 A simple method for estimating the
viscosity of molten metallic alloys Alloys Comp. 221 L13
[55] Kucharski M 1986 The viscosity of multicomponent systems
Z. Metallkde. 77 3936
[56] Kucharski M 1988 A model for predicting the viscosity of
multicomponent solutions Z. Metallkde. 79 2646