You are on page 1of 7


The Strain-rate, Temperature and Pressure
Dependence of yield of Isotropic
Poly(methylmethacrylate) and
Poly(ethylene terephthalate)*
R. A. D U C K E T T , S. R A B I N O W I T Z t , I. M. W A R D
Department of Physics, University of Leeds, Leeds 2, UK

The yield behaviour of poly(methylmethacrylate) (P M M A) has been investigated in tension
and compression over a range of testing temperatures and strain-rates. Both tensile and
compressive yield stresses were found to increase monotonically with increasing strain-
rate and decreasing temperatures. Compressive yield stresses were in general found to be
more dependent on strain-rate.
The results of this investigation have been correlated with previous published data for
the dependence of the torsional yield stress of P M M A on hydrostatic pressure. This was
done by a modification of a theory proposed by Robertson which uses the internal viscosity
approach to yield in glassy polymers. The modified theory clearly explains the temperature
and strain-rate dependence of the yield stress and provides a quantitative explanation of
the differences in behaviour between tension and compression in terms of the dependence
of yield on the hydrostatic component of the applied stress.
The tensile yield behaviour of isotropic amorphous poly(ethylene terephthalate) (PET)
sheets has also been investigated over a wide range of temperatures and strain-rates. No
torsion or compressive yield stresses are available because of the sheet form of the PET,
but the results obtained in tension are shown to be fully consistent with the above theory,
and with other published work.

1. Introduction drawn between a pressure dependent von Mises
Many workers have attempted to extract from yield criterion and the Coulomb criterion used
the complex non-linear mechanical behaviour of widely in soil mechanics [2, 3]. However, a
isotropic glassy polymers simple failure criteria modification of the yon Mises criterion proposed
which can be used to predict yield behaviour by Hill [5] to allow for anisotropy has been used
under the combined stresses of constant strain- successfully to describe the tensile yield of aniso-
rate and temperature e.g. [1-3].These approaches tropic polymers [6-8].
draw on experience obtained in investigations of Other papers have concentrated on the depend-
the yield behaviour of metals and soils. The yon ence of tensile stress on testing temperature and
Mises criterion, successful for describing the strain-rate. Many of the discussions centre
yield of metals [4], has been found to be inade- around the problem of establishing time-
quate for dealing with polymeric materials temperature superposition, e.g. [9-12]. All the
because of the observed difference in yield stress polymers investigated show a monotonic increase
between tension and compression. The latter has of yield stress with increasing strain-rate and
been attributed to a dependence of yield stress some authors assert that there is an approximately
on the hydrostatic component of stress, but the linear relationship between stress and the
limited range of experimental data presently logarithm of strain-rate. The yield stress for most
available does not allow a clear distinction to be glassy polymers decreases linearly with increas-
*Paper presented at Cambridge, April 1970, during conference on "Yield, Deformation and Fracture of Polymers".
~Present address: Ford Motor Co, P.O. Box 2053, Dearborn, Michigan 48121, USA.
9 1970 Chapman and Hall Ltd. 909
ing temperature over a wide range of tempera- sion type load cell, and also using a compression
tures below the glass transition temperature. cage in conjunction with a tensile load cell.
In this investigation an attempt has been made True yield stresses were obtained by relating
to link together the two approaches to yield out- the maximum observed load to current cross-
lined above. The tensile and compressive yield sectional area. A correction to all measurements
stresses of poly(methylmethacrylate) have been was made to allow for deformation of the
measured over a range of temperatures and machine and compression cage.
strain-rates. It was found that the compressive
yield stress was more sensitive to strain-rate than 3, Results
the tensile yield stress. The theory for yielding of The results obtained are shown graphically in
isotropic glassy polymers proposed by Robertson figs. 1 and 2. Fig. 1 represents the variation of
[13] has been modified to allow for the effect of tensile and compressive yield stresses of P M M A
hydrostatic pressure. Using this modified theory with log10 (strain-rate) at various temperatures.
it has been possible to correlate quantitatively In order for easy comparison to be made
the tensile and compression data from this between tension, compression and torsion data
investigation with the data obtained on the same and also with the Robertson theory, the shear
material in torsion tests under an overall hydro- component of the yield stress is plotted as the
static pressure [14]. ordinate and log10 (shear strain-rate) as the
The tensile yield stress of isotropic amorphous abscissa. In fig. 2 the results obtained for the
poly(ethylene terephthalate) sheet has also been tensile yield of P E T at -- 25, 20, 40, 50 and
measured over a wide range of temperatures and 6 0 ~ are plotted as the tensile component of
strain-rates and fitted to the above theory using stress versus log10 (tensile strain-rate).
parameters deduced from the literature. Fig. 3 is taken from Rabinowitz et al [14] and
represents the variation of torsional yield stress
2. Experimental of P M M A at constant (room) temperature and
The poly(methylmethacrylate) ( P M M A ) tensile shear strain-rate (4 • 10-4 sec-1) with super-
and compression specimens were machined out posed hydrostatic pressure. (The material used
of 1 in. (2.54 cm) thick sheets of I C I "Perspex". was identical to that used in the current investiga-
The tensile specimens were approximately4.3 mm tion.)
in diameter with a straight gauge length of 2.5
cm. Compression specimens were circular i ] I I I t l
cylinders with a diameter of 6.4 mm and length YIELD STRESS OF POLYME,rHYLMETHACRYLATE
of 12.8 cm. 1.0 vs LOG~0 (STRAIN-RA,rE!
The poly(ethylene terephthalate) (PET) was
0.9 o COMPRESSION "r=23~
supplied by I C I Plastics Ltd with a thickness of o 9 TENSION T=60*C
approximately 0.64 ram. The optical birefrin-
gence of the material was measured as less than u)
5 • 10-5 and the yield stress in the plane of the 0.7
sheet was found to be isotropic within experi- Q

mental error. The tensile specimens were dumb-
bell shaped with a straight gauge length of 2.5 cm
width 4 mm. ~0.5
All tests were performed on an Instron tensile ~0.~
testing machine, inside an environmental cham-
ber within which the temperature could be con- ~0.3
trolled to within -t- 0.5 ~ C both above and below ul

room temperature. Tensile tests over a wide 0.2
range of strain-rates were performed on P M M A
at 60 and 90 ~ C and on P E T at -- 25, 20, 40, 50 0.1
and 60 ~ C. A minimum of 20 rain was allowed 1 I I I [ I I
for each specimen to reach thermal equilibrium. -6 -5 -& -3 -2 -1 0
Compression tests on P M M A were conducted LOG10
at room temperature, both in direct compression Figure I The variation of tensile and compressive yield
between the machine crosshead and a compres- stresses with strain-rate and temperature, P M M A .

( STRAIN" RATE/ sec) The present results confirm, at least for P M MA,
]SOTROPIC PET what is already well established, namely that the
-25~ yield stress of an isotropic polymer depends on
(1) temperature, (2) strain-rate and (3) hydro-
static pressure. It is believed, however, that they
represent the first set of data to link these three
2o~ 3.8q aspects on the same material. Let us discuss first
~ o.' points (l) and (2) above.
The tensile and compressive yield stresses of
o.6 poly(methylmethacrylate) and poly(ethylene
terephthalate) are observed to increase with

"-., 3.5~ increasing strain-rate and with decreasing temper-
0.4~ ature. It should be noted that in general the plots
of yield stress versus log (strain-rate) have a small
but definite curvature, indicating that the theory
/" of Eyring [15] in terms of a single simply
~./ 0.2 activated flow process is not applicable to the
/ present data. Our data differ from those of
/ iF 0.1 Bauwens-Crowet et al [16] and Holt [12] in this
| I I I
respect, but are in general agreement with the
-5 -4 -3 -2 -| data of Roetling [l 8 ]. The theory to be described
LOGlo (STRAIN-RATE/ sec ) predicts a definite curvature in this type of plot
Figure 2 The variation of the tensile yield stress of iso- which we therefore consider to be of greater
tropic PET with temperature and strain-rate. generality than the linear behaviour.
Owing to the brittle nature of P M M A , in
tension at room temperature, it was not possible
to obtain direct comparison between tension and
compression, but data from other work [1, 2, 14]

PRESSURE POLYM E TH'~tETHACRYLATE: / clearly indicates that compressive yield stresses
1.5 T=25"C "~=4x10 "4 SEC-1
~" would be higher than tensile yield stresses.
/ Further, the data from this paper shows the
/ greater sensitivity of the compressive yield to
/o strain-rate than the tensile yield.
n/ //o/ It will be shown that the difference in behaviour
o / between tension and compression can be
attributed quantitatively to the dependence of
yield on the hydrostatic component of stress. The
theory is based on that of Robertson [13].
a: 1.(: The tensile behaviour of PET also shows a
3: / 9 YIELD slight curvature in the plots of stress versus log
U~ 02
/ FULL LINE BEST FIT strain-rate. Additionally, at the higher tempera-
-~ 0.8 tures, the stress was found to decrease more
rapidly with decreasing strain-rate than at the
" 02 lower temperatures. This was accompanied by
the formation of tensile crazes before yielding at
0.6 the high temperatures and low strain-rates, i.e.
the normal shear mode of failure was preceded
by craze formation.
I [ [ I I I I
0.4~ 1 2 3 4 5 6 7 5. T h e o r y
A phenomenological theory of yielding has been
Figure 3 The variation of torsional yield and fracture stress developed from Robertson's theory [13] which
of P M M A with hydrostatic pressure (from [14]). uses the measured dependence of shear yield
R. A. D U C K E T T , S. R A B I N O W I T Z , I. M. W A R D

stress on hydrostatic pressure to predict the work during the activation event and that the
observed difference in absolute magnitudes of energy difference should be
tensile and compressive yield stresses, quantita-
tively and also their different sensitivities to
AE -- .rv cos 0 + p D .
temperature and strain rate. Thus whenp is positive (as in a compression test)
Following Robertson, it is assumed that the effective barrier height between the two states
structural elements of the polymer can exist in is increased and when p is negative (as in a
either of two energy states with a difference in tension test) the effective barrier height is
energy of say AE (N.B. this effectively assumes reduced, p(2 represents the work done by the
the planar zigzag model for the polymer chain). hydrostatic component of the applied stress, v
In equilibrium it is proposed that the fraction in and g2 having the dimensions of volume.
each state is given by the Boltzmann distribution. From this point onward we follow the argu-
(The glassy polymer is considered as being ment of Robertson exactly. Because an increase in
quenched in the state it occupied at the glass the fraction of elements in the upper state implies
transition temperature Tg.) Thus before the a tendency towards a structure characteristic of a
application of stress the fraction of elements in higher temperature, and therefore more mobile,
the high energy state is transitions from high to low energy will occur
exp(-- AE/kOg) infinitely more slowly than transitions in the
Xi = [1 + exp(-- AE/kOg)] (1) opposite direction. (This point is discussed at
length by Robertson.) It is therefore possible to
where 0g = Tg if the test temperature T < Tg calculate the maximum fraction of elements in
and0g=TifT> Tg. the upper state averaged over all orientations and
In the isotropic polymer there will be a uni- this can be shown to be:
form distribution of orientation of the structural
elements whether they be segments of the polymer Xmax = 2v--~
chain or small aggregates of chains. The applica-
tion of shear stress ~- will tend to increase the ( i n ( 1 + exp[--(AE--'rv + pY2)/kT]l
fraction of elements in the upper energy state for
elements in some orientations, and decrease it
for other orientations. According to Robertson + k--T-]- -fiT + kT ~ (4)
the shear component of stress ~- changes the E/kO,) \.
energy difference between the two states to 1 + exp(-- AE/kOg)j
AE--~'v cos0. Here -rv cos0 represents the work
done by the shear stress ~-in the transition, and 0 With this fraction of elements in the upper state
is some angle defining the orientation of the the polymer is structurally equivalent to that of
element with respect to the shear stress. the melt at a temperature 01, where
Therefore the fraction of elements in the upper exp(-- AE/k01)
state, with orientation 0 is Xmax = [1 + exp(-- AE/kO0]" (5)
exp[-- (AE -- -rv cos 0 )/kT] Hence a characteristic temperature 01 can be
Xt(O) = {1 + exp[-- (AE -- -rv cos 0 )/kr]} (2) calculated for the polymer under the influence of
Clearly the fraction increases for orientations a shear stress ~- and hydrostatic pressure p. We
such that then compute the effective viscosity of the poly-
mer at the temperature 01 using the W L F
A E -- -rv cos 0 AE equation, and so the resulting strain rate can be
-- 9 (3)
kT kTg shown to be
If the two energy states are associated with the '7"
trans and gauche chain conformations, trans
being the lower energy, then an increase in gauche 01
conformations implies a reduction in density due exp--(2"303 [(01 --Tg+C1C2C2) ~ _ C 1 ] } (6)
to less efficient packing. Hence there will be an
interaction with the hydrostatic component of where Ca, C2 are the "universal" W L F para-
the applied stress. We propose, therefore, that meters and ,?g is the universal viscosity of a glass
the hydrostatic component of stress p also does at Tg. Yield is defined as the load when the

displacement rate arising from the viscous strain TABLE I Table of coefficients for P M M A
rate p just matches the crosshead speed in a
Optimised values Values suggested by
conventional tensile or compression test (i.e. the Robertson
elastic strain-rate is zero and the load to produce
this viscous strain-rate corresponds to the c1 11.1 ~ C 17.44 ~ C
55.9~ C 51.6~ C
observed value of m a x i m u m load), c2
AE 0.88 kcal/mole 1.44 kcal/mole
In addition to the parameters AE, v, Tg, C1, C2 Tg 1050 C 105~ C
and r/g required by the conventional Robertson "qg 10~2poise 1018 -- 10~4-6poise
treatment, the model includes the additional v 109 A3 140/~3
parameter s An approximate relationship o/v 0.175 0
between f2 and v may be obtained from the
results of Rabinowitz et al, [14]. As we have seen
d'c/dp exists for amorphous P E T owing to the
their results show that for P M M A the shear
difficulty of making tubular specimens from thin
yield stress, at constant shear strain rate, is an
sheet material, and so we are forced to estimate
approximately linear increasing function of
a value from several sources. Rabinowitz et al
hydrostatic pressure. Reference to equation 6
[14] and Bowden and Jukes [2] observed that
indicates that over the range of shear stresses
the yield stress of P M M A can be described by
observed by Rabinowitz et al at constant ~7
(0.5 kilobar < "c < 1.1 kilobar) 01 must be "r = "co -}- k P
approximately independent of pressure. Exam-
where k = 0.204 from Rabinowitz (data from
ination of equation 4 indicates that if"cv - - pg2 =
torsion under hydrostatic pressure) and k = 0.175
constant then this condition is satisfied. We
from Bowden and Jukes (data from plane strain
therefore suggest that ~2/v -- (dz/dp)~, and from
compression). The latter technique produced a
fig. 3 take (d'c/dp)~ = 0.2.
value of 0.172 for isotropic amorphous P E T
The data obtained from tension and compres-
(private communication), whereas the former
sion at various strain-rates and temperatures and
yielded k = 0.075 for isotropic crystalline P E T .
the data of Rabinowitz et al were analysed
We therefore choose rather arbitrarily f2/v = O. 1
according to this modified Robertson treatment.
for this amorphous P E T sheet.Treating the other
For the tensile and compression tests "c and p
six coefficients as independent variables the
were taken to be ~/2 and :k ~r/3 respectively,
curves shown in fig. 2 were obtained. A list of
where (r is the axial yield stress ~ is assumed to be
coefficients for P E T is shown in table II. No
equal to half the applied axial strain-rate. The
attempt was made to fit the data in the region of
maximum shear stress on the specimen and
applied hydrostatic pressure were used for "c and
TABLE II Table of coefficients for P E T
p in the torsion tests. The values for AE, v, C1,
C2, ~g, Tg suggested by Robertson, together with Optimised values Values suggested by
= 0.2v do not give an acceptable fit to the data. Robertson
Therefore these values were used as starting cx 15.1 ~ C 17.44~ C
points and all the coefficients were adjusted c2 75.7 ~ C 51.6 ~ C
independently by computer to give a "least AE 1.91 kcals/mole 1.38 kcals/mole
squares fit" using a procedure developed by Tg 90 ~ C 70~ C
Murgatroyd [17]. In the spirit of the Robertson v 234/~ 215 ~8
treatment the shear and hydrostatic components o/v 0.10 0
1014"~poise 1013 -- 10146poise
of stress and temperature were treated as ~g
independent variables, which minimised the
deviations between the predicted and actual rapidly changing stress at 60 ~ C where crazes
shear strain-rates for the tension, compression were observed to precede yield. With regard to
and torsion tests, by adjusting the constants in the values of the coefficients obtained the follow-
the equation independently. The following table ing observations seem relevant; (71 and C~ lie well
(table I) shows the final values obtained for the within the range of values obtained from dyn-
coefficients and those predicted by Robertson amic-mechanical experiments for both P M M A &
for P M M A . These calculated coefficients were and P E T (see for example [19, 20]). The values
used to generate the full curves in figs. 1 and 3. for v, Tg and 9g also compare favourably with
N o such direct data to determine a value for published data. However the values of A E for
R. A. D U C K E T T , S. R A B I N O W I T Z . I. M. W A R D

both P M M A and P E T are in some disagree- The slight disagreement between the data for
ment with independent measurements. The only P M M A and the theory may arise in part
comparison for P M M A is with the value because of the use of a crude planar zig-zag
predicted from Tg using the Gibbs-DiMarzio model of the polymer chain and from the
relationship A E = 3.81 Tg cals/mole. It can be approximation used for calculating X maximum.
seen that the value obtained from this investiga- However, we believe that the fit is sufficiently
tion is only about 60 ~ of this prediction. Our good to warrant three tentative conclusions.
value of A E for P M M A should be compared (1) The difference between tensile and com-
with the value of 0.9 kcals/mole for polystyrene, pressive yield stresses for P M M A may be
obtained from ultrasonic measurements [21]. attributed to interaction with the hydrostatic
For P E T we calculate A E = 1.91 kcals/mole. component of stress. N . B . In principle this
This should be compared with the Gibbs- technique should be capable of distinguishing
DiMarzio value of 1.38 kcals/mole. However, between a modified von Mises and a Coulomb
the work of Riveros and Bright Wilson [22] criterion, i.e. between the expressions
using microwave spectroscopy yields a value of A E - - "rv q- P g 2 and d E - - r v - - err.g2' respect-
0.186 4-0.060 kcals/mole for the difference in ively, for the stress-modified energy difference
energy between the rotational isomeric states of between the two states. For the torsion tests
ethyl formate, separated by a barrier height of p = -- er~rand so/2 = g2' to explain the pressure
1.1 :k 0.25 kcals/mole for upward transitions. dependence in torsion correctly. However, for
The value of 0.186 kcals/mole was used by the axial tests erN = zk cr/2 and p ---- 4- or/3, so
Walker and Semlyen [23] for the methyl group d E - - ~-v + p g 2 ~ A E - - -rv - - c~sg2. It is not
rotation, in a calculation which accurately felt, however, that from the present results we can
predicted the measured cyclic trimer concentra- choose unequivocally between these criteria.
tion in the melt of poly(ethylene terephthalate). (2) The general features of the rate dependence
The results would imply that the barriers to of yield stress may be represented in terms of an
conformational changes at yield are higher than effective viscosity, that is pressure, temperature
those measured in the melt. This might be and shear stress dependent.
because the stiffness of the terephthaloyl units is (3) The effective viscosity relates directly to the
more effective in the denser material. low strain relaxation behaviour through the
constants which appear in the W L F equation.
6. C o n c l u s i o n This is at first sight surprising, in view of the fact
A modification of Robertson's molecular theory that yield occurs at much higher levels of stress
of yielding has been presented which allows the and strain than is usual in linear visco-elasticity.
hydrostatic pressure, temperature and rate-
dependence of yield to be discussed within a Acknowledgements
common formalism. Although the details of the One of us (R.A.D.) is grateful for support from
theory are complex the theory is attractive for an S R C Fellowship throughout the course of
several reasons. Firstly because it emphasizes this work, and S.R. was supported by the
that yield is not a unique point on the stress- National Science Foundation.
strain curve, but merely the point at which the
plastic strain-rate produced by the stress just References
matches the machine displacement rate and 1. w. WHITNEYand R. D. ANDREWS,J. Polymer Sei.
secondly, it does not require that a horizontal log C 16 (1967) 2981.
strain-rate shift will produce a yield master curve. 2. P. B. BOWDEN and J. A. JUKES, J. Mater. Sci. 3
Thirdly, it predicts that yield is governed by a (1968) 183.
temperature dependent activation energy (form- 3. S. S T E R N S T E I N , L. O N G C H I N , and A. S I L V E R M A N ,
ally introduced via the W L F equation). It also Applied Polymer Symposia No. 7, p. 175 Interscience
suggests a possible molecular mechanism of yield (1969).
4. yon MISES, See R. Hill "The Mathematical Theory
in polymers. of Plasticity" (Oxford U.P., 1950)p. 20
The accuracy of the fit obtained is very promis- 5. R. HILL, Proc. Roy. Soc. A. 193 (1948) 281.
ing, but clearly it is desirable to find a material 6. N. B R O W N , R. A. D U C K E T T , a n d I. M. W A R D , Phil.
with which it is possible to do tensile and com- Mag. 18 (1968) 483.
pression tests over a wider temperature range 7. c . B R I D L E , A. B U C K L E Y , a n d J . S C A N L A N , J.Mater.
and also torsion tests under hydrostatic pressure. Sci. 3 (1968) 622.

8. J . G . R I D E R a n d E. H A R G R E A V E S , J . Polymer Sci. 18. J. A. ROETLING,Polymer 6 (1965) 311.
A2 7 (1969) 829. 19. s. SAITO, KotloidZ. 189 (1963) 116.
9. Y U . S. L A Z U R K I N , ibid30 (1958) 595. 20. J. o. FERRY, "Viscoelastic Properties of Polymers"
10. R. E. ROBERTSON,Jr. Appl. Polymer Sci. 7 (1963) 443. (J. Wiley and Sons, 1970) p. 316.
l l . C. B A U W E N S - C R O W E T , J. V. B A U W E N S , and G. 21. H. J. B A U E R , H . I-IASSLER, M. I M M E N D O R F E R , and
HOMES, J. Polymer Sci. 7 (1969) 735. W . P E C H O L D, u n p u b l i s h e d p r o c e e d i n g s o f c o n f e r e n c e
12. D. L. HOLT, J. Appl. Polymer Sci. 12 (I968) 1653. on "Polymer Chain Flexibility", University of Essex,
13. R. E. ROBERTSON,J. Chem. Phys. 44 (1966) 3950. Janualy, 1970.
14. s. R A B I N O W I T Z , I. M. W A R D , and J. S. C. PARRY, 22. s. M. R I V E R O S a n d E. B R I G H T W I L S O N , J. Chem,
J. Mater. Sci. 5 (1970) 29. Phys. 46 (1967) 4605.
15. H . E Y R I N G , J. Chem. Phys. 4 (1936) 283. 23. G. R. WALKERand s. A. SEMLYEN(in press).
16. J. c. B A U W E N S , C. B A U W E N S - C R O W E T ,
and G.
HOMES, J. Polymer Sci. 7 (1969) 1745. Received 22 July and accepted 28 August 1970
17. P. N. MURGATROYD, Electronics Letters 5 No, 26

A Thermoelectric Method for Measuring electrode a nickel-plated martensitic sewing
Inhomogeneities in Metallic Solid needle was used, whose tip was ground on a
Solutions precision lathe from the original 0.04 to 0.06 m m
radius down to 0.005 to 0.008 ram. Heat was
In the study of many metallurgical phenomena, supplied by focusing the light of a projection
measurements of fine scale solute variations are bulb with condenser (not shown) by means of a
necessary. Generally, the more common methods short-focus convex lens, transparent to infrared,
o f direct analysis which can be applied to very onto a blackened copper target soldered to the
small volumes suffer f r o m low sensitivity, par- needle together with an output copper wire. The
ticularly when the base concentration level of the sample was shielded (against heating and solder
solute to be analysed is small. Fine scale measure- evaporation) from the light target by a thin
ments of mechanical properties (e.g. micro- Teflon sheet. Provided that the bulb power was
hardness) are not directly related only to solute stabilised and air draughts avoided, a uniform
concentIation. probe temperature between 250 and 350~
In connection with a program of non- could be obtained. To avoid conductive heat
equilibrium grain-boundary segregation studies, losses to the hardness tester, the needle was
the use of microscopic electrical point measure- mounted in a holder of machineable ceramic and
ments was reconsidered. There have been the original mount for the hardness indenter was
attempts to measure the local thermoelectric replaced by an acrylic fitting. The moving lever
e m f between a fine point and the specimen sur- of the tester was equilibrated by the auxiliary
face [1, 2], which, however, were not repro- loads and was loaded in the range of 1 to 10 g
ducible enough to obtain a satisfactory calibra- by means of laboratory weights.
tion of emf versus concentration [2]. We present The instrument stage was moved uniformly
here the design of a comparatively simple experi- with the speed of about 0.1 to 1 m m / h by means
ment based on the thermoelectric emf method of a geared timing m o t o r and the micrometric
which provided quite satisfactory results. shift of the tester (not shown). The output
Similarly as in [1, 2] a modified microhardness thermoelectric emf was recorded by the potentio-
tester ( L E I T Z " D u r i m e t " ) was used, but in this metric recorder with 1 mV full scale sensitivity.
case the thermoelectric probe was heated by This low thermoelectric emf represents the
means' of a focused light source (fig. 1). After difference-signal resulting from the individual
insulating the compound stage, the sample was voltages of different joints in the chain and pro-
mounted on a holder and both were fastened by vides enough sensitivity for following slight
Plasticine to the stage, and provided with the thermoelectric emf changes caused by changing
soldered output copper wire. F o r the heated composition of the specimen in the place of the
9 1970 Chapman and Hall Ltd. 915