HAL Id: jpa-00246593
https://hal.archives-ouvertes.fr/jpa-00246593
Submitted on 1 Jan 1992
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
X-ray investigation of the segregation in Cu-Ti alloys
H. Gürzing, V. Gerold
To cite this version:
H. Gürzing, V. Gerold. X-ray investigation of the segregation in Cu-Ti alloys. Journal de Physique I,
EDP Sciences, 1992, 2 (6), pp.1129-1144. �10.1051/jp1:1992201�. �jpa-00246593�
J.
Phys.
I France 2 (1992) l129-1144 JUNE 1992, PAGE 1129Classification
Physics
AbstractsX-ray investigation of the segregation in Cu.Ti alloys
H.
Giirzing
and V. GeroldMax Planck Institut fur
Metallforschung, Stuttgart
and Institut fur Metallkunde,University
ofStuttgart, Germany
(Received J8 November J99J, revised and
accepted
26 March J992)AbstracL The
segregation
of Cu-Tialloys
with variouscompositions
wasinvestigated using single crystals
and monochromaticX-rays
(Cu~ and MOK«). Thedeveloping
side bandsalong (h00)
of thereciprocal
lattice wereinvestigated
up to(800).
The results werecompared
with those from asimple
one-dimensionalsegregation
model. The difficulties and limits of interpreta- tion are shown. Noprincipal
difference in thesegregation
behaviour was found forsingle
crystalshaving
Ti concentrations between and 2.9 at.fb.1. Introduction.
In the past 20 years the
alloy
systemcopper-titanium
has attracted many researchers since it seemed to be atypical
system forspinodal decomposition.
Inaddition,
thesteep
increase of theyield
stress in theage-hardened
state is remarkable. Theprecipitation
behaviour of thesealloys
has beeninvestigated
for avariety
ofcompositions using
differentexperimental
methods as
X-ray diffraction,
transmission electronmicroscopy (TEM)
or field ionmicroscopy (FIM). According
to thetheory
the kinetics ofprecipitation depend
on theposition
in thephase diagram
relative to thespinodal.
Thus forhigh temperatures
and low titanium contents(I.e.,
above thespinodal)
a normal nucleation andgrowth
process doesoccur whereas for low
temperatures
andhigh
titanium concentrations(I.e.,
below ofit)
aspinodal decomposition
isexpected.
Thisdecomposition
is characterizedby
thewavelength
Lof the concentration modulation
occurring mainly
in one direction which istypically
(100)
in fccalloys.
It isexpected
that thiswavelength depends
on both thealloy composition
and the
aging
temperature.The
typical X-ray powder
diffraction pattem of aspinodally decomposed alloy
shows so- called side bands(SBS)
close to theDebye-Scherrer (DS)
lines. These side bands are observed ifsegregation
occurs in asupersaturated
solid solution.They develop
because of two reasons.The first one is a variation of the local electron
density.
In the well-known bookby
GuinierIll
this is called substitutional disorder ; it causes also smallangle scattering.
This contributiondoes occur
only
if the difference of the atomicscattering
factors of solvent(f~)
and solute~fB)
islarge enough, I.e.,
if a considerable monotonic Lauescattering (MLS)
of the solid solution does exist which isproportional
to(f~ f~
)~. Theintensity
contribution of the SBSJOURNAL DE PHYSIOUEI -T 2. N'6, JUNE 1992 42
1130 JOURNAL DE PHYSIQUE I N° 6
is taken out of the MLS which is reduced at the same time. For the side bands of the DS lines
(H00)
with H=2, 4,
theintensity
variesproportional
to(f~ f~)~ii(Ah)
where H= 2 ao sin
f§/A,
Ah=
h H
=
2
ao(sin
8 sinf§)/A
= 2 ao cos
f§ A8/A,
ao the latticeconstant, 8 the
Bragg angle
andf§
theBragg angle
of thecorresponding
DS line. Thefunction
ii (Ah )
describes theprofile
of apair
of the side bands and isexpected
to be the same function for all orders H.The second reason for the occurrence of the SBS is a variation of the local lattice constants due to the
segregation (displacement
disorderIll coupled
with substitutionaldisorder).
In that case theintensity
of the SBS increases withincreasing Bragg angle
and readsapproximately h~ f~i~ (Ah ).
Now theintensity
is taken out of the DSlines,
which intensitiesare reduced at the same time
according
to a staticDebye-Waller
factor whichtypically
increases withincreasing aging
time. Forspinodally segregating alloys
as Cu-Ti thiscontribution is the dominant one if the
investigation
is madeby X-rays.
Most
segregation phenomena
can be characterizedby
alength parameter
L which can be evaluated from the distance of a side band from the DS line. Thislength
can beinterpreted
either as the modulation
wavelength
of a one-dimensionalspinodal decomposition
wave, or as an average distance betweenneighbouring particles
in a nucleation andgrowth
process. For acopper
alloy containing
4.7 at.fb TiMiyazaki
et al.[2]
found as anexample
L t "~ from thebeginning
of anaging
treatment at 400 °C without a time range where L was constant. Inaddition,
from the intensities of the side bandsthey
calculated asteadily increasing
concentration
amplitude
withaging
time. As it is discussed in the nextparagraph, however,
there exists
already
asegregation
in thisalloy immediately
afterquenching
from thehomogenization
temperature.Obviously,
the observation of side bands inpolycrystals
is not sensitiveenough
to see thebeginning
of thedecomposition
of thisalloy.
In an
alloy containing
2.9 at. fb Ti Hakkarainen[3]
could showby
TEMinvestigation
that a modulated structure resultedalready
afteraging
for 8s at 500 °C. Withincreasing aging
timesuperstructure
spots of the metastablephase p' (Dla
structureCu4Ti) develop.
The contrast of theimage depended
very much on the contrast conditions. Either a modulated structure ordiscrete
particles
could be seen. Similar results were foundby
Comie et al.[4]
and Datta et al.[5].
Foralloys containing
3.8 or 5.2 at.fb Ti modulated structures couldalready
be seen afterquenching
from 860°C. Thecoarsening
of the modulated structure followed a rule L-- t "~
Recently, experiments
with field ionmicroscopy (FIM)
contributed a lot to theknowledge
of
spinodal decomposition.
With this method the localcomposition
within a very small volume elementcontaining only
tens of atoms could be evaluated.Using
this method Biehl andWagner [6, 7] investigated
theaging
of a 2.7 at.fb Tialloy
at 350 °C and found agradual
increase of the concentration fluctuation. After 50 min a Ti content of 20 fb was reached in the enriched
regions
which is theequilibrium
value for the metastablep' phase Cu4Ti.
In theseexperiments
a continuousgrowth
of thewavelength
L was found from thebeginning according
to L t "~ in accordance with theX-ray
results mentioned above.In contrast to these results Alvensleben and
Wagner [8]
found a nucleation andgrowth
behaviour for the metastable
p' phase
for a 1.9 at.fbTialloy aged
at 350 °C. From these results the authorsjudged
that at 350 °C thespinodal
must lie between both concentrations 1.9 and 2.7 at.fb Ti. Table Igives
a survey of thepositions
of thespinodal
foundby
differentauthors. In
spite
of the differences it can be concluded that at 350 °C theposition
of thespinodal
should lie between 1.3 and 2.7 at.fb Ti.The present
experiments
have been undertaken withX-ray
diffraction fromsingle crystals
of different titanium content. Since the formerX-ray experiments
have beenperformed
withpolycrystals
it is felt that the use ofsingle crystals
may enhance the resolutionespecially
forN° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS l131
Table I. The
position of
thespinodal
in thephase diagram
Cu-Tiaccording
to various authors.Alloy Temperature
Referenceat,fb °C
1.0 350 9
~ l.3 and
~ 2.7 350
6,
7, 82.33 300 10
2.57 400 10
3.06 500 10
3.86 600 10
4.05 700 to 725 11
5.20 400 to 500 5
short
aging
temperatures. Theexperiments
wereperformed
withalloys
on both sides of thespinodal
in thehope
to find characteristic differences in theexperimental
results because of the differenttypes
ofsegregation
processes. Inaddition,
model calculations wereperformed
which were used to find
segregation parameters
from theexperimental
results.2.
Experimental procedure.
Two
single crystal
disks wereprovided by
ProfessorWagner
from GKSSForschungszentrum Geesthacht, Germany,
which contained 2.7 and 2.9 at.fbTi, respectively.
Both had a flat(200)
surface which could be used for the present diffractionexperiments.
Inaddition, single crystals
with a lower titanium content wereproduced by
theBridgeman technique.
Theircompositions
were 1.0 and 1.3 at.fbTi, respectively.
From thecrystals
disks wereprepared
which had also a
(200)
surface.The
homogeneization
treatment occurred at 870 °C for 3 h in a vertical fumace in an argonatmosphere.
To avoid oxidation thecrystals
werewrapped
in a niobium foil andquenched
into
liquid nitrogen
after the heat treatment. For theaging
in a temperature range between 200 and 450 °C thecrystals
wereencapsulated
in aglass
tube filled with argon. Afteraging
fora time t~
they
werequenched
into water where the tubes broke into peaces. For shortaging
times the
crystals
were lowered into a salt bath.The diffraction
experiments
were undertaken in a 8 2 8 diffractometerusing
monochro- matized CuKa and MoKaradiation, respectively.
Thequality
of the monochromator allowedan
intensity
ratio of 30 : between the aj and a~ radiation. With thisequipment
the sequence ofBragg peaks (H00)
with H=
2, 4, 6,
and 8including
their side spots(which
furtheron will be called side bands for convenience as in the case of DSlines)
were measuredalong
the line(h00)
in thereciprocal
lattice. The latter twopeaks (600)
and(800)
could be measuredonly
with the MoKa radiation.
Figure
I shows atypical
sequence of side bandprofiles
at the(200)
peak using
CuKa radiation. The firstsign
of side bands could be observedalready
afteraging
1132 JOURNAL DE
PHYSIQUE
I N° 6normalized I/lm
0.002 CuT12.9at%
fi
T"350°C i j
,/ i
) ~,
'l
o.ooi i,
o
1.86 1.93 2 2-07 2.14
rec. lattice vector h
Fig.
I.Intensity profiles
of side bands (more accurately, side spots) in units of the peakheight
of theBragg peak
(200) for variousaging
times at 350 °C. CuKa radiation.for 20 min at 350 °C. The
experiments
were continued up toaging
times of 100.000 min(about
48days).
The
following
parameters were taken from theexperiments:
atfirst,
the averagewavelength
L(measured
in units ofao)
of the concentration fluctuation could be determinedaccording
to theequation [12, 13]
Htan
(f§)
~
(A8)(H~
+K~
+ L ~)
where
(HKL)
are the indices of theBragg peak, 8~
is thecorresponding Bragg angle
and he is theBragg angle
distance of one of the two side bands from the mainpeak.
Since there isusually
aslight asymmetry
in the distance of the two side bands an average value is taken.Parallel to this method of L determination the same parameter was tried to find from line
broadening
measurements of the side bands. Corrections forgeometric broadening
have to be made in order to get the correct half width B. Theseparation
from strainbroadening
occurredby comparing
the widths of the(200)
and(400)
side bandsusing
CuKa radiation[14, 15].
In order to characterize the
intensity
of the side bands thefollowing
two parameters wereinvestigated
theintensity
ratioR~~m (I~)/(I~~~)
of thepeak heights
of the mainpeak (I~)
and of the side band at thehigh angle
side(I~~~),
and the ratioR~j
m(I~~~)/(Ij~~)
of thepeak heights
of the side bands on thehigh
and lowangle
side of the mainpeak
is taken.3.
Experimental
results.3,I INTRODUCTION. The present paper is restricted to results of the
aging
temperature 350 °C where most of theexperiments
were undertaken. In all cases theas-quenched single crystals
did not show anysign
of side bands whichdeveloped only during aging.
Since therewere no
principal
differences in the results of differentalloy compositions, only
the measurements with the Cu-2.9 at.fb Ticrystal
will bereported
in detail.3.2 THE MODULATION WAVELENGTH L. The variation of the modulation
wavelength
Lwith
aging
time is shown infigure
2. For the evaluation of L an average of all Ah values of thepeaks (200)
and(400)
has been taken. For the shortaging periods
up toN° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS 1133
wavelength L
CuT12.9at%
~ 6h (CuKa)
~~~
+ 6h (MoKa) B (CUKa)
g
i~
m.0.1941
10 loo 1000 10000 IOOOOO
aging
time t[min]
Fig.
2. Thedevelopment
of the modulationwavelength
L withaging
time at 350 °Cshowing
tllree ranges I to III ofwavelength growth.
The results from linebroadening
measurements are included.100min the accuracy was limited to L
= 17±1. For the
higher
order reflections thedetermination of the
peak positions
and therewith the accuracy of L was no more sogood.
The
resulting
curves for both radiations are more or less the same.They
can be devided into three ranges. In range I thewavelength stays nearly
constant ; it lasts up to 150 min(2.5 h).
In range II aslope
of m= 0.19 is observed in the double
logarithmic plot
which increases tom = 0.36 in range III. The border between both ranges is about 900 min
(15 h) aging
time.The latter
slope
value is close to m = 0.33 which is thetypical
value for Ostwaldripening
where either the
particle
radius or the meanparticle
distance is used asparameter.
In
figure
2 the results for L from linebroadening
measurements areplotted,
too.They
show
agreement only
in range II. Thefailing
in range I may be anargument
that the diffraction is causedby segregation
withoutsharp
boundaries. The latter is needed for a correctinterpretation
of linebroadening.
It should be mentioned,however,
that Eckerlebe et al.[16, 17]
could detect discreteparticles already
afterquenching
of thisalloy.
3.3 THE INTENSITY RATios
R~~
ANDR~j.
The nextquestion
is theintensity development
ofthe side bands
compared
to theintensity I~
of theBragg peaks
which isgiven by
theintensity
ratio
R~~.
One has to take intoconsideration, however,
thatI~
itself isdecreasing
withincreasing aging
time and withincreasing
value of h while theintensity
I~~~ of the side band isincreasing. Figure
3 shows the timedependence
of all ratios measured with CuKa(H
=
2 and
4)
and MoKa(H
=
2, 4,
6 and8)
in adouble-logarithmic plot.
The ratiodecreases with time and order of reflection and reaches values of
nearly
one. The hdependence
ofR~~
is shown infigure
4 for variousaging
times. For data from range II(150
to10.000min)
theslope
of the curves in the doublelogarithmic plot
is about the same for different times and amounts to values of 2.5 to 2.7. In range III the curves start to flatten off when theintensity
ratio reaches values close to one. For the ratioR~j
of thepeak heights
of both side bands,only
three results are shown infigure
8b wherethey
areplotted
as a function of the order H.At the late
segregation stage
extraspots (h'00)
with h' around 5.91 and 7.88 could be observed which couldbelong
either to incoherentp' (the
lattice constant in thetetragonal
cdirection is
enlarged by
1.5fb)
or to the stablephase Cu~Ti.
It can besuggested
that the appearance of this incoherentphase
determines range III.l134 JOURNAL DE PHYSIQUE I N° 6
lm/lhsb
+- (2QQ)CU
-- (4oo)~cu
~~- (200)~MO
- 14oo),Mo
- 16001,Mo
+- (wol,Mo
T@350°C
i
io loo loco loooo looooo
aging time [min]
Fig.
3. Thedevelopment
of theintensity
ratioR~
= I~/I~~~ of the
peak
intensities of the mainBragg
spot(J~)
and of thehigh angle
side spot iih~b) as function of theaging
time (at 350 °C) for various orders h of the (h00)Bragg
spots.lm/lhsb
loo ? iso mIn
+ 25O mIn
S 600 mIn
° 1600 mIn
X 6600 mIn
° 16700 mIn
1
6 325O0 mIn
Z 100000 mIn CuT12.9at%
T*350°C
o.
i io
rec. lattice vector h
Fig.
4. The hdependence
of theintensity
ratio R~~ = i~/I~~~ (seeFig. 3)
for variousaging
times at 350 °C.4. Model calculations.
4.I THE ONE-DIMENSIONAL MODEL. In order to understand the
development
of theintensity
data and to try tointerpret
the results in terms ofparameters
of thedeveloping
microstructure the
simple
one-dimensionalsegregation
model shown infigure
5 is used as a first step. In aperiodic region
oflength Lao
asegregation
of asupersaturated
solid solutionwith solute concentration
Co
into twophases (I)
and(2)
with concentrationsCi
andC2
is assumed(Fig. 5a).
As a consequence ofthis,
the average latticeplane
distancedo
=ao/2 changes
to two different valuesdj
=
(I
+ej) do
andd~
=(I
+e~) do.
Thischange
N° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS l135
II
L.ao
.
L2.a2 ~~
al Liar
Co C2 Un
b)
x
Fig.
5.- One-dimensionalsegregation
model where a solid solution with solute concentrationCo
segregates (a) into tworegions
with different concentrations C~resulting
in (b) shifts u~ of latticeplanes
n.can be described
by
a shift u~ of the latticeplane
n from its undistortedposition (Fig. 5b).
Theresulting slopes
define the misfit parameters ej and e~.Then,
thefollowing
relations hold Lj +
L~
=
L
(I)
Cj Li
+C~L~
=
COL (2)
ej
Lj
+e~L~
=
0.
(3)
Since the concentration variations have
only
a small influence on theX-ray
intensities(the
atomic numbers of Cu(29)
and Ti(22)
are close to each other and the maximumsegregation
AC amounts
only
to 20fb)
there remain twoindependent parameters
ej andLj
besides thewavelength
L which have to be determined fromexperiments.
Since the amount of misfitdepends
on the localcomposition
an information on the latter can be deduced from the data.The model leads to the
following
diffractionamplitude [14]
:~
_
~~ sjjjjjji~+~jj iii
+f, ~ii iii) ~iellll~ ~~~~"~~~ ~~~~~~~~~
~~~
in this
equation fj
andf~
are the atomicscattering
factors ofregions (I)
and(2)
which do notdiffer very much from each other
and, therefore,
have beenreplaced by
an averagef.
Parameter P is an
integer
and describes the number ofperiods
L in thesegregation.
Fromequation (4)
theintensity
I amounts to1=
f~[A)+Aj+2AjA~cos (grLh)]I~(P) (5)
where Ai and A
~ are the two fractions inside the brackets
given
inequation (4).
The last factorI~(P )
inequation (5)
is the square of thecorresponding
fraction inequation (4)
and has its mainpeaks
ofequal heights
at the idealpositions
Ah=
v/L from the
Bragg peak
withv =
0, 1, 2,
...,
which are broadened with
decreasing
number P ofperiods
oflengths
L. Thesepeaks
are modulatedby
the other functions in the brackets. The termsAl
andAl
have their mainpeaks
atpositions
shifted from thecorresponding Bragg peak
Hby
Ah
= e~
h~.
If theregion
enrichedby
Ti is called(I)
then ej ~ 0 and A contributesmainly
to the
low-angle
side band. The third term(T)
in the brackets is an interference term whichcan take
positive
andnegative
values.As an
example, figure
6 shows the three terms A), Al
and Tat the(600) peak position
for a1136 JOURNAL DE
PHYSIQUE
I N° 6L~55.5, L1.18.5,E1.0.00155,E2.-O.oo077 Intensity
[arbitrary
unitsl12
8
)
4
o 6
4 Ai A2
2 /
-2 -4
-0.04 -0.02 O O.02 o.04
deviation ah
Fig.
6. Theintensity
contribution of the tllree terms in the brackets ofequation
(5) at the(600) Bragg peak.
Below the tllree termsA), Al
and T, their peaks markedby
arrows. Above the sum curve I~~ of all tllree terms. Model parameters used : L=
55.5
Lj
=18.5 (both in units of the lattice constant) ; ej= 1.55
x10~~
Thepositions
of I~ are marked by vertical lines.typical
set of modelparameters.
Theirpeak positions
are markedby
arrows. Thepeak positions
off~
are
given by
vertical lines. For v= 0
(the Bragg peak position)
all three terms in the bracketsgive
apositive
contribution which results in alarge Bragg peak intensity.
Asshown in the upper part of the
figure
the sum curveI~~
of the bracket terms demonstrates the interference of all terms. Inaddition,
the maxima of this curveI~~
have differentpositions
ascompared
to the maxima ofI~, I-e-,
the last factor inequation (5).
Thepeaks
ofI~~
are determinedby
intemalstrain,
theirpositions
vary with the order H of theBragg peaks
while the
peaks
off~
are determined
by
thesegregation wavelength
L. The latter dominate for theintensity
distribution I ofequation (5).
With
(I) increasing
order H and(it) increasing wavelength
L theoverlapping
of the functions Aj and
A~
decreases because of(I) increasing peak separation
and(it) increasing sharpness
of theintensity profiles. Therefore,
the interference term T decreases inintensity
which results in a decrease of the
corresponding Bragg peak.
This model differs from that usedby Tsujimoto
et al.[18] by
the interference term T whichthey ignored.
For the further discussion it is assumed that the model does
give
also a correctpeak height
for the
Bragg peaks.
This would mean that theexperimentally
measuredintensity
is the sum of the intensities of many domains of dimensions FL as it is the case for the side bands. The fact that there exist three different orientations for the one-dimensionalsegregation
isignored
for the
following
reasons. Atfirst,
there existsonly
one set of side spots atBragg peaks (hoc)
since thecorresponding
intensities for side spots in the k andf
directionsare
negligible.
Secondly,
as it will be discussedlateron,
the matrix haslocally segregated
into threedimensions which will influence the
interpretation
of the parameter set of the one-dimensional model.
Thus,
the wholesegregated
volume of thecrystal
will contribute to the side bands as well as to theBragg peaks.
4.2 THE DEPENDENCE OF INTENSITY RATIOS ON THE MODEL PARAMETERS. in order to
study
the influence of theindependent
model parametersP, L, Lj
and ej on theintensity
ratios
R~~
andR~j
one of the parameters was varied while the other three werekept
constant.N° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS 1137
The first
example
shown infigure
7gives
the influence of the number P ofperiods
on the ratioR~~
as a function of the reflection order h. Thisdiagram
may becompared
with theexperimental
results infigure
4.Only
the curves with P= 10 and 50 are close to one of the
experimental
curves(t~=250min). Therefore,
all models have been calculated with P= 50. There is
only
aslight
difference in theslope
of thecorresponding
curves which is 2 for the model curve and 2.5 for theexperimental
one.The
following figures
represent the influence of the distortionparameter
ej(Fig. 8)
and acombination of ej and
Lj (Fig. 9)
on the ratiosR~~(h)
andR~,(h).
Withincreasing
ej the ratios
R~~
decrease while theslope
of the curves remains the same in the double-logarithmic plot (Fig. 8a).
For the ratioR~,(h
an increase is observed, too, but the curves areno more
straight (Fig. 8b).
In the latterfigure
someexperimental
results are shown also which will be discussed later.In
figure
9 theproduct
of e,Lj
waskept
constant while bothparameters
were variedsimultaneously.
For the ratioR~~ only
a small variation of the curve is observed(Fig. 9a)
while alarge
variation occurs for the other ratioR~j (Fig. 9b).
Thus it can be concluded that the ratioR~~(h
will be useful to determine theproduct
F, Lprovided
theexperimental
datashow a similar h
dependence
than the model data.TheTeafter,
aseparation
of bothparameters e, and
L,
ispossible by comparison
of the ratioR~j. Again,
a similar hdependence
is necessary.
lm/lhsb
io
4~ P.3 L.55.5
n p.5 L1.18.5
° P.10 E 1. 0.00155
P.50 E2.-o.000755
1
10 rec. lattice vector h
Fig.
7. The h dependence of theintensity
ratio R~ on the number P ofperiods
L for the one- dimensional model.5.
Comparison
betweenexperiment
and model.5.I THE ALLOY CONTAINING 2.9 at.fbTi. Before the
comparison
is undertaken a fewremarks should be
given
whatparameters
could beexpected.
The metastablemiscibility
gapII 38 JOURNAL DE
PHYSIQUE
I N° 6lm/lhsb
ei . o.oo i
e i
/
O[O014 /
.~~~~
E1.O.O022 O-O018
~"~~~
L.55.5.c0nst. L1.18.s~const.
10
al
rec. iatt;ce vector hlhsb/llsb
w---_
Experiment iO,COO
~
1.8
t~,
min +--, _~
~
l,200
~ °
1.6 ~ 350 ~
___~ 2.2
~
)_
~l.8
Model *~
1.4
L . 55.5
L~
.18.5~~
~
l
,
2 4 6 8 10
b)
rec. lattice vector hFig.
8. The hdependence
of theintensity
ratios la)R~
and (b) R~j on the distortion parameter ej of the one-dimensional model. Infigure (b)
tllreeexperimental
data sets are included.at 350°C has the
approximate
concentration limitsC~
= 0.3fb and
C~
= 20fb for thedepleted
matrix and theprecipitate
in the laterstage, respectively (all
datagiven
in at.fbTi).
From these concentrations the volume fraction of the
precipitate
isexpected
to bef~
= 13.2 fb. From the
composition dependence
of the lattice constant(ao/nm
=
0.3616 +
0.044.C)
thefollowing
distortionparameters
can besuggested:
e~=19.6x10~~
ande~=
-3.2x10~~
The latter data have to becompared
with the model parameters ej and e~.The
comparison
betweenexperimental
results and model calculations came out to be very difficult. For theproduct
ej Lj it was
possible
to reach anagreement
for mostaging
times with theexception
of verylarge
ones. The results are shown infigure
10 where theexperimental
data
points
fromfigure
3 arecompared
with calculated model curves. For eachaging
time anotherproduct
ejLj
was found in this way(see Fig. 12).
The difficulties of
interpretation
arise for theseparation
of bothparameters
ej andLj, I.e.,
theinterpretation
of theintensity
ratioR~j
of both side bands. There were twoproblems
which could not be solvedreasonably
well. Atfirst,
the Hdependence
ofR~j
did not agree with that of the model curves for thehigher
orders H= 6 and 8. This is
N° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS 1139
lm/lhsb
ioo
Li,ei
L1.23.5, E1 . -O.O012
a)
10rec. lattice vector h
lhsb/llsb
L.55.5.const. ~~
Lie~.
L1~L2,
e1,
E2 . variable10
b)
rec. lattice vector hFig.
9. -The hdependence
of theintensity
ratios (a) R~~ and (b) Rhi on the variation of ej andLi
under the condition ejLj
=
0,0285.
shown in
figure
8b where someexperimental
data aregiven
as a function of H.They
can beapproximated by
the modelonly
ifR~j
issteadily increasing
with H which is foundonly
forshort
aging
times below 1000 min. Thus thecomparison
wasonly
undertaken forH
=
2 and 4. In table II a few model data are
given
which have been evaluated fromdiffraction
experiments
for differentaging
times. The numbers for Lj, ej and e~ are the best
fits found in this way.
The second
problem
is theinterpretation
of theseparameters. They
were derived from a one-dimensional model where the ratioLj/L
shouldcorrespond
to the volume fractionf~
of thesegregated phase
enriched in Ti. Thecorresponding
values in table II are much toohigh
andthey
decrease withincreasing aging
time.Since the
segregation
does occur in three dimensions the obtained data forLj
and ej have to be discussed in this way. As the nextpossibility
the model could beinterpreted
as aone-dimensional
projection
of a three-dimensionalperiodic
array of cubicparticles
of size L j in a cubicprimitive
arrangement as it is sketched infigure
II. Theparticles
are located inregion (I)
where the distortion parameter ej is aweighted
average of the distortions in theprecipitate (e~)
and in the matrix(e~).
In this model e~ is then identical with e~. In that case(Lj/L)~
shouldgive f~
in anapproximate
way, which is now toolarge
1140 JOURNAL DE
PHYSIQUE
I N° 6lm/lhsb
o calculation
% ~
experiment: °
O (200)
° (400)
~'~ ~ (800) MO
z j800) MO o.oi
io ioo 1000 ioo00 iooo00
aging time [mini
Fig.
IO. The timedependence
of the ratio R~~ for the four orders of reflection forexperimental
data(symbols)
and for the model (curves)resulting
in an evaluation of theproduct ejLj
as a function ofaging
time.(Tab. II). Therefore,
one has to increase Lj
considerably
in order to fitf~
which results in the data shown as curves PM infigure12.
As a next
possibility
one has to assume that the cubicparticles
are more numerous inregion (I)
offigure
11. As it has been found from neutron smallangle scattering experiments by
Eckerlebe et al.[16, 17]
theparticles
in the later stages are rod-like and have an aspectratio which increases from 1.5 to 4.5 with
increasing aging
time. If Lj is
supposed
to be the diameter of the rods, then for two of the three orientations(rods parallel
to y andz in
Fig.
IIa
projection
on the x axis would contribute to the distortion in this direction. This would increase the volume fraction and therewith allow a decrease ofLj (resp. Lj/L) compared
tothe cubic
particle
model and would result in the curves « PM + asp.r. » infigure
12.However,
this still would notsatisfy
the model valueR~j
as shown infigure
13 for theBragg positions (a)
H= 2 and
(b)
H=
4.
Table II. The parameters
L, Lj
and ej andfunctions of
them.ta L El L
j
L
I ~l ~2 L
l~~ (L
l~L)~min nm
10~~
nm nm 10~~ 10~~ fb fb
20 5.7 2.66 2.2 1.2 0.8 39 5.8
120 6.4 6.01 2.2 2.7 1.4 34 4.0
000 9.6 16.24 2.4 6.8 2.3 25 1.6
10 000 14.8 26.37 3.7 7.2 2.4 25 1.5
30 000 21.2 38.95 4.4 9.5 2.4 21 0.9
N° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS l141
3-dim. PM
L
hl
Fig.
ii. Tllree-dimensionalparticle
model withregions
(I) and (II) normal to the x direction.L,
L1Inml L1*eljnml
Ll.el
L
PM
PM . asp. r PM
a wra a+
r~
CuT2.9at% Tz350°C
lo loo IOOO IOOOO IOOOOO
al
aging time [ministrain El
,
e2
T=350°C ' ~'"~'
jj
PM+ asp. r.
PM
~~
~
-O.O04
~~~ IOOO
lOOf°
1°°°°~lo
~~;ng time [mini
Fig. 12. Variation of the model parameters with the
aging
time of the Cu-2.9 at.9b Ticrystal
for tllree different models : a) the parameters L, ejL,
andLj
; b) the parameters e, and e~.1142 JOURNAL DE PHYSIQUE I N° 6
lhsb/llsb
experiment: CuT12.9at% T«350°C
° (200) Cu
2.5 ~ (200) Ma
calculation=
2 PM + asp, r.
PM + CIUSt.
~ ~ ~ ~
x
~'~
~
~w
I~
~ ~
l
lo loo 1000 10000 100000
a)
aging time lmin]lhsb/llsb
experiment:
° (40O) Cu
~ (40O) MO ~
o o
o o
calculation; o
o ~ ~
PM + asp. r. '
~
xi
+ ~ o a
x
~ ~o
oo
o °
I
io loo loco loo00 looooo
b)
aging time lmin]Fig,
13. The timedependence
of the ratio R~, forexperimental
data(symbols)
and for two models (curves) for tile side bands (a) at (200) and (b) at (400). The resulting parameters Lj, ej and e~ are shown infigure
12.In order to get a better fit one still has to decrease L
j/L
which would mean that now stillmore
panicles
have to be concentrated inregion (I)
offigure
I I in order to reach theexpected
volume fraction
f~
=
13.2 fb. So to
speak
it is a model in between the one-dimensional one where the whole volume ofregion (I)
is enriched in Ti and that one wherepanicles elongated
in the y and z directions with
large
spaces ofdepleted
matrix in between arefilling
thisregion.
This model is called cluster model and the
corresponding
curves « PM + dust. » are found infigure12
and thecorresponding
values in table II.Only
these data fit the modelparameter R~,
for H=
2 and 4 as shown in
figure
13 but no fit could be obtained for thehigher
orders 6 and 8. Athigher aging
times this fit fails also for the order H=
4 as it is shown in
figure
13b.It must be admitted that in these estimates the different contribution of the third
orientation
(rods
in the xdirection)
has not been taken into account. A more accurate calculation would need a muchhigher
effort but would contain so many not well knownparameters
that it has not beenperformed.
5.2 THE ALLOY CONTAINING 1.0 at, fb Ti. For this
alloy
a nucleation andgrowth
behaviouris
expected
for theprecipitation
kinetics.However,
all diffractionexperiments
with thisalloy
N° 6 X-RAY INVESTIGATION OF THE SEGREGATION IN Cu-Ti ALLOYS l143
L,
LlInml
Ll*eiinml
L
+ asp, I
+ clust.
Ll* El
i
CuTilat% T«350°C
o-i
too 1000 10000 IOOOOO
al
aging time lm;njstrain E2
PM , clust.
CuTilat% T.350°C 0.006
PM . asp- r.
~i
PM
~~
L2
0.00~
~~loo 1000 lO000 100000
b)
aging time lmin]Fig. 14. Variation of tile model parameters with tile
aging
time of the Cu-1.0 at.9b Ticrystal
for three different models : a) tile parameters L, ejLj
andLj
; b) the parameters ej and e~.single crystal
did not show remarkable differences in thedevelopment
of side bandsexcept
aslower
segregation
rate(Fig. 14). Therefore, only
the ranges I(up
to 1000 minaging time)
and II can be seen for thedevelopment
of thesegregation wavelength
L. At the transitionbetween both ranges a kink in the curve
Li
ej is deduced which results fromreproducible
kinks in the
R~~(t)
curves for all orders of reflections.Otherwise,
the same arguments hold for the modeladaptions
as for the formeralloy crystal.
6. Conclusions.
The
present investigations
have shown that there is no marked difference in thedevelopment
of the side band
profiles
for bothalloys
discussed so far as for the otheralloys
studied in the range between I and 2.9 at,fbTi. Thus it cannot beproved
if for thehigher
concentratedalloys
thesegregation
process isgovemed by spinodal decomposition
as it has been foundby
FIMtechniques [6, 7]
for a similaralloy (2.7
at.fbTi)
for the first 50 min ofaging
at 350 °C.l144 JOURNAL DE
PHYSIQUE
I N° 6Obviously,
it was alsoimpossible by
neutron smallangle scattering,
to find any prove for this mechanism[16, 17].
In that case the diffraction curve is causedby changes
of the local Ti concentration and notby
localchanges
of the latticeparameter
as in the presentexperiments (which
makes theinterpretation
of the results even moredifficult). Obviously,
diffractionprofiles
from differentsegregation
processes as nucleation andgrowth
orspinodal decomposi-
tion do not differ very much from each other which makes it
impossible
to differentiate between these both mechanisms.Acknowledgments.
The authors are very much
obliged
to Prof. R.Wagner
of theForschungszentrum
Geesthacht, Germany,
whoprovided
them with slices ofsingle crystals having higher
titaniumcontents
(2.7
and 2.9at,9b).
The critical commentsby
him andby
the referees of thejoumal
have
improved
themanuscript
and aregratefully acknowledged.
References
[Ii GUINIER A., X-ray Diffraction in
Crystals, Imperfect Crystals
andAmorphous
Bodies (Freeman andCompany,
San Francisco, 1963).[2] MIYAzAKI T., YAJIMA E, and SUGA H., Trans. JIM 12 (1971) l19.
[3] HAKKARAINEN T., PhD Thesis,
University
ofTechnology,
Helsinki, Finnland (1971).[4]
CORNIE J. A., DATTA A. and SOFFA W. A., Metall. Trans. 4 (1973) 727.[5] DATTA A. and SOFFA W. A., Acta Metall. 24 (1976) 987.
[6] BIEHL K. E. and WAGNER R., Proc. Int. Conf, on Solid-Solid Phase Transformations, H. J.
Aaronson et al. Eds. (TMS/AIME,
Pittsburgh,
Penn., 1981)p.185.
[7] BIEHL K. E. and WAGNER R., Proc. 27th Int. Field Emission
Symp.,
Y. Yashiro and N.Igata
Eds.,Tokyo, Japan
(1980), p. 267.[8] ALVENSLEBEN L. V. and WAGNER R.,
Decomposition
ofAlloys
theEarly Stages,
P. Haasenet al. Eds., 2nd
Acta-Scripta
Conf. Series(Pergamon
Press, Oxford, 1984) p. 143.[9] SAXLOVA M. and BALIK J., Czech. J.
Phys.
B 31 (1981) 215.[10] DUTKIEWITZ J., Bull. Acad. Polan. Sci. 22 (1974) 323.
[I
Ii
LAUGHLIN D. E. and CAHN J. W., Acta Metall. 23(1975)
329.[12] DANIEL V. and LIPSON H., Proc. Roy. Sac. A181 (1943) 368.
[13]
DANIEL V. and LIPSON H., Proc.Roy.
Sac. A182(1944)
378.[14] DEHLINGER U. and KOCHENDbRFER A., Z. Kristall. A101 (1939) 134.
[15] GURzING H., Doctoral Thesis,
University
of Stuttgart,Germany
(1990).[16] ECKERLEBE H.,
Diploma
Thesis, University of Hamburg-Harburg (1985).[17] ECKERLEBE H., KAMPMANN R. and WAGNER R.,
SANS-Investigation
ofEarly Stage Precipitation
Kinetics in Cu-2.9 at,9b Ti ; AtomicTransport
and Defects in Metalsby
NeutronScattering,
C.Janot, W.Petty,
D. Richter and T.Springer
Eds.(Springer-Varlag,
BerlinHeidelberg,
1986) p. 66.[18] TSUJIMOTO T., HASHIMOTO K. and SAITO K., Acta Metall. 25 (1977) 295.