You are on page 1of 8

JOURNAL OF APPLIED PHYSICS VOLUME 33.

NUMBER 12 DECEMBER 1962


X-Ray Diffraction Study of the Effects of Solutes on the Occurrence of
Stacking Faults in Silver-Base Alloys
R. P. r. ADLER AND C. N. J. WAGNER
Hammond Metallurgical Laboratory, Yale Univtrsity, New Haven, Connecticut
(Received June 21, 1962)
The addition of the solutes cadmium, indium, and tin to silver increases the probability of deformation
faults a in filings from 3 X 10-
3
in pure silver to a maximum value of 45 X 10-
3
at the highest concentrations
of solute. In addition, the twin fault probability f3 measured from center of gravity displacements varies
from lOX 10-
3
for pure silver to 30X 10-
3
for the alloys highest in tin or indium concentration. Lattice
parameters ahkl were determined from all available reflections of the cold-worked and annealed specimens
and plotted as a function of cos
2
8/sinO. By relating the large scatter of the individual ahkl to the occurrence
of deformation faults in the deformed material, the true lattice parameter, ao(CW), and the deformation
fault probabilities a of cold -worked materials could be determined. There was an apparent decrease in lattice
parameter of the deformed Ag-Sn alloys which was largest (,,-,0.1%) for the greatest tin concentration
(Ag-9%Sn). Using Fourier analysis of line profiles, the effective particle sizes (D,lhkl and root mean square
strains [(SL2)avJhklt were determined. The measured effective particle sizes were anisotropic [(D')l\l/
(D,bo=1.7J and are primarily a consequence of deformation and twin faulting. The values for the com-
pound fault probability (1.5a+f3) from peak shift and asymmetry and from anisotropic particle sizes, i.e.,
from peak broadening, agreed rather well.
1. INTRODUCTION
S
TACKING faults can exert an important influence
on the structural and mechanical properties of face-
centered cubic (fcc) metals and alloys. Recrystalliza-
tion/
2
texture formation,a,4 and microstructure
5
-
S
can
be related to the stacking fault energy of the material.
Work hardening
5
,6,S-u as well as low temperature creep12
theories also predict a relationship between cross-slip
and stacking fault energy,
Direct observations of stacking faults can be ac-
complished by using electron transmission microscopy
or x-ray diffraction techniques. Using the electron
microscope, the stacking fault energy 'Y can be meas-
ured
5
,13,14 for metals and alloys of low stacking fault
energy provided that the amount of plastic deformation
is small (i.e., low concentration of stacking faults).
Using x-ray diffraction
15
,16 the stacking fault probability
can be obtained in heavily deformed metals. For a given
amount of deformation, the relative change in stacking
1 H. Hu, R. S. Cline, and S. R. Goodman, Trans. AIME 224, 96
(1962).
2 N. Brown, Trans. AIME 221, 236 (1961).
3 H. Hu and R. S. Cline, J. Appl. Phys. 32, 760 (1961).
4 H. Hu, R. S. Cline, and S. R. Goodman, J. Appl. Phys. 32,
1392 (1961).
6 M. J. Whelan, P. B. Hirsch, R. W. Horne, and W. Bollmann,
Proc. Roy. Soc. (London) A240, 524 (1957).
6 N. F. Mott, Trans. AIME 218,962 (1960).
7 C. S. Barrett, Imperfections in Nearly Perfect Crystals Uohn
Wiley & Sons, Inc., New York, 1952), p. 97.
8 P. B. Hirsch, Internal Stresses and Fatigue in Metals (Elsevier
Publishing Company, Inc., New York, 1959), p. 139.
9 A. Seeger, Defects in Crystalline Solids, Bristol Conference (The
Physical Society, London, 1955), p. 328.
10 A. Seeger, Dislocations and Mechanical Properties of Crystals
Uohn Wiley & Sons, Inc., New York, 1956), p. 243.
11 A. Seeger, R. Berner, and H. Wolf, Z. Physik 155, 247 (1959).
12 P. R. Thornton and P. B. Hirsch, Phil. Mag. 3, 738 (1958).
13 M. J. Whelan, Proc. Roy. Soc. (London)IA249, 114 (1959).
14 A. Howie and P. R. Swann, Phi!. Mag. 6, 1215 (1961).
15 M, S. Paterson, J. App!. Phys. 23, 805 (1952).
16 B. E. Warren, Progress in Metal Physics (Pergamon Press,
New York, 1959), Vo!' 8, p. 147.
fault probability with solute concentration in an alloy
can be related to the change in stacking fault energy 'Y.
Values of 'Y can also be measured indirectly from
mechanical properties (stress-strain relationshipslO,l1
and low temperature creepl2) by assuming that the
theories of work hardening involving the coalescing of
extended dislocations are correct. Using twin boundary
energies,17 'Y may also be found if the assumption is made
that the stacking fault energy is twice as large as twin
boundary energy.
It has been shown that solute additions to pure
metals alter the stacking fault energy. Using the electron
microscope, Howie and Swann
l4
found that the stacking
fault energy decreases with increasing solute concentra-
tion. In terminal copper alloy systems
1S
-
24
x-ray investi-
gations have shown that the stacking fault probability
increases with increasing solute concentration. Similar
results
2
3-25 have been found for terminal silver alloys.
Plastic deformation of fcc metals may produce two
types of stacking faults which can be detected by x-ray
diffraction. If the normal stacking sequence
26
,27 of (111)
planes is ABCABC, then a deformation fault
26
,27 is a
break in this sequence A B C ~ B C A B C where the arrow
indicates the fault plane. A reversal in the sequence
ABCACBA represents a twin fault.
26
,27 Deformation
faults produce a shift of the position of the powder
pattern peak, whereas twin faults cause the line profile
17 R. Fullman, J. App!. Phys. 22, 448 (1951).
18 B. E. Warren and E. P. Warekois, Acta Met. 3, 473 (1955).
19 H. Otte, J. App!. Phys. 33, 1436 (1962).
20 J. C. Helion, M. Eng., thesis, Yale University, 1961.
21 C. N. J. Wagner, Acta Met. 5, 427 (1957).
22 R. E. Smallman and K. H. Westmacott, Phil. Mag. 2, 669
(1957).
23 D. E. Mikkola and J. B. Cohen, J. App!. Phys. 33,892 (1962).
2. R. G. Davies and R. W. Cahn, Acta Met. 10, 621 (1962).
2. L. F. VassamiIIet, J. App!. Phys. 32, 778 (1961).
25 C. N. J. Wagner, Acta Met. 5, 477 (1957).
27 A. H. Cottrell, Dislocations and Plastic Flow of Crystals
(Clarendon Press, Oxford, 1953), p. 73.
3451
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3452
R. P. 1. ADLER AND C. N. J. WAGNER
to become asymmetric.
1s
In addition to the broadening
due to faulting, the peaks of cold-worked fcc metals are
broadened by a reduction in the size of the coherently
diffracting domains as well as by the distortion within
each domain.
ls
In this investigation silver was chosen as the base
metal because it has a very low stacking fault energy14
for a pure metal (resulting in a stacking fault density
that is detectable by x-ray diffraction
26
). Cadmium,
indium, and tin have approximately the same scattering
power as silver, consequently these metals were selected
as solutes to minimize any asymmetrical effects on the
powder pattern line shape due to unequal scattering
powers of the (111) faulting planes.
28
The probabilities of stacking faults in silver alloy
filings, cold-worked at room temperature, have been
evaluated using all information available from powder
pattern peaks: (1) Measurements of peak maxima for
both the cold-worked and annealed states yield deforma-
tion fault probabilities a and lattice parameters ahkl of
individual (hkl) reflections; (2) The difference between
the center of gravity and the peak maximum of (111)
and (200) cold-worked reflections (i.e., asymmetric peak
profiles) is a measure of twin fault probability fJ;
(3) Fourier analysis of line profiles of the available
multiple order peaks provide information about the
anisotropic effective particle sizes, the strain distribu-
tions, and the compound fault probability (1.5a+fJ)
Since the lattice parameters from each individual peak,
ahkl, were calculated, it was of interest to investigate if
there was a procedure to evaluate the absolute lattice
parameter of a heavily deformed material, ao(CW),
containing stacking faults.
II. EXPERIMENTAL RESULTS
A. Sample Preparation
The alloys used in this study were prepared by
melting pure silver (99.99%) and the appropriate
amount of solute (Cd, In, Sn) in an evacuated and
sealed quartz tube; after melting, the alloys were
homogenized in the terminal solid solution region. All
specimens were filed at room temperature; the cold-
worked samples were passed through a 150 mesh (0.1
mm) screen and then examined on the same day while
those filings to be annealed were passed through a 325
mesh (0.044 mm) screen, sealed under vacuum in a
quartz tube, and annealed at 500C for several hours.
For examination by x-ray diffraction, the filings were
pressed flat into a plexiglass sample tray with a dilute
solution of Duco cement in acetone as a binder. The
positions and shapes of the Debye-Scherrer lines were
recorded at 292C using a GE XRD-5 diffraction unit
with filtered copper radiation. To obtain absolute angu-
lar values of 2() the unit was aligned on both sides of true
zero in 2() with a beryllium acetate sample.
B. Peak Position Measurements
Measurement of peak position leads to a direct
determination of the deformation fault probability a
and the true lattice parameter ao. The peak positions
of the annealed material, where the Ked and Ka2
components were resolved either due to high angle
resolution or Rachinger correction,29 were determined
by extrapolating the midpoints of lines drawn parallel
to the background at various heights to the peak
maxima; the peak position was then computed by
taking the weighted average of the two components
(Le., 2()Ci= [2 (2(),,1)+ (28,,2)J/3). The estimated accuracy
in 2(J for the Bragg angles was O.Ol" for 2(J less than
90 and 0.02 for 28 of larger values. For the
broadened cold-worked peaks, the average peak posi-
tion was found by midpoint extrapolation to the
peak maximum for all angles of 2(J greater than 60.
For the low angle (111) and (200) peaks it was necessary
to correct for the asymmetrylG due to the factor
j2(1+cos
2
2(J/sin
2
8). The weighted position of the (111)
and (200) peaks was then calculated from the extra-
polated peak maxima of the Rachinger
29
resolved K
and Ka2 components. As a result of broadening the
accuracy in 2() of the peak maxima was O.02 for
peaks less than 2(J=80" and O.OSo for all other peaks.
According to the Pater'3ou theory,15 the deformation
fault probability a can be calculated from the relative
angular shift, of a cold-worked peak from
the corresponding annealed one
16
,21 :
(28hkl)CW- (2()Ohkl)Ann.
= (360/;r)G
hkl
tan8hkl"a, (1)
where the values of the planar parameter G
hk1
are given
in Table VO
To eliminate positioning errors it is more accurate to
measure relative changes between two neighboring
TABLE 1. Planar parameter Ghkl and spacing fault constant JUl (reference 30).
peak, kkl 111 200 220 311 222 400 331 420 422

GMl -0.035 +0.069 -0.035 +0.013 +0.017 -0.035 -0.007 +0.007 0
Jokl +0.209 -0.167 -0.167 +0.038 +0.209 -0.167 -0.167 +0.284 -0.167
28 B. T. M. Willis, Acta Cryst. 12, 683 (1959).
29 W. A. Rachinger, J. Sci. Instr. 25, 254 (1948).
ao C. N. J. Wagner, A. S. Tetelman, and H. M. Otte, J. Appl. Phys. 33,3080 (1962),
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Alloy
(atomic percent)
---
Ag
Ag-2.5 In
Ag-5.0 In
Ag-7.5In
Ag-l0.0 In
Ag-12.5 In
Ag-15.0 In
----
STACKING FAULTS IN SILVER-BASE ALLOYS 3453
TABLE II. Experimental values for silver-indium alloys.
Peak shift Lattice parameter
(aX 10
3
) (aX 10
3
) Annealed Cold-worked
(Lattice parameter (Peak (1.50+{:I)M+A
shift) shift) ({:IX 10') X 10
3
ao(Ann.) au(C.w.)
(A) (A)
5 3 11 16 4.0866 4.0863
6 6 17 26 4.0941 4.0938
14 12 16 35 4.1030 4.1030
19 17 27 53 4.1104 4.1097
25 27 10 51 4.1188 4.11!lO
28 29 30 73 4.1268 4.1264
38 40 31 91 4.1356 4.1350
Peak broadening
Alloy
(atomic percent) (D,)111 (A) (D,hoo (A) T min (A)
[(l:h_50).v J1ll
t
[(Eh_50).v Jtooi
(1.50+{:I)BX 10' X 10
3
X 10'
Ag
Ag-2.5 In
Ag-S.O In
Ag-7.5 In
Ag-10.0 In
Ag-12.5 In
Ag-15.0 In
Alloy
(atomic percent)
Ag
Ag-1.5 Sn
Ag-3.0 Sn
Ag-4.5 Sn
Ag-6.0 Sn
Ag-7.5 Sn
Ag-9.0 Sn
240 135 370 23
215 110 500 32
150 90 200 33
140 80 205 39
120 70 165 43
90 60 100 45
70 45 76 58
TABLE III. Experimental values for silver-tin alloys.
Peak shift
(aX 10
3
) (aX 10')
(Lattice parameter (Peak (l.5a+.B)M+A
shift) shift) ({:IX 10') X 10'
S 3 11 16
7 6 6 15
11 9 14 28
20 18 19 46
24 23 18 52
38 36 21 75
42 45 37 104
Peak broadening
2.0 2.8
3.2 4.5
3.1 4.8
3.1 5.7
4.2 5.0
4.2 5.0
4.5 5.6
Lattice parameter
Annealed Cold-worked
ao(Ann.) ao(C.W.)
(A) (A)
4.0866 4.0863
4.0932 4.0921
4.0998 4.0982
4.1057 4.1042
4.1122 4.1114
4.1191 4.1172
4.1265 4.1222
[( Eh_50).v JI11i Alloy
(atomic percent) (D,)1l1 (A) (D,hoo (A) T min (A)
[( e
2
L_50).v Jl001
(1.5a+i3hX lO' X 10
3
X1O'
Ag
Ag-1.5 Sn
Ag-3.0 Sn
Ag-4.5 Sn
Ag-6.0Sn
Ag-7.5 Sn
Ag-9.0 Sn
240
220
185
140
135
105
71
135
120
110
80
75
65
37
peaks.l8,21 ,31 The deformation fault probabilitiy ex is thus
related to these shifts
I8
,26:
(6.(2(Jhkl)- (6.(2(Jh'k'l' )=Ha, (2)
where H is a parameter defined by Wagner.
31
The defor-
mation fault probability a was calculated from the
following neighboring pairs (111)-(200), (200)-(220),
and (220)-(311) for the three solutes series studied; the
average value is recorded in Tables II, III, and IV. The
continuous variation of the average 0: with solute con-
centration is shown in Fig. 1.
To determine the true lattice parameter of an
annealed sample, ao(Ann.), the lattice parameters
31 C. N. J. Wagner, Z. Metallk. 51, 259 (1960).
370 23 2.0 2.8
390 28 3.5 4.4
260 29 3.7 5.0
220 42 3.4 4.9
220 43 3.8 5.5
125 43 4.1 6.2
150 94 4.6 5.3
calculated from all available (hkl) reflections ahkl are
plotted versus the extrapolation function, cos
2
(J/sin(J
(this function was chosen because it eliminates the
greatest estimated error due to the displacement of the
TABLE IV. values for silver-cadmium alloys.
Alloy (aX 10')
(atomic percent) (Peak shift) (i3X 10')
Ag 3 11
Ag-lO Cd 9 15
Ag-20 Cd 19 26
Ag-30 Cd 23 28
Ag-40 Cd 30 21
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3454 R. P. I. ADLER AND C. N. ]. WAGNER
Q 50
i
40
:B
30
a..
l20

o 5 10 15 20 25 30 35 40
Atomic Percent Solute
FIG. 1. Deformation fault probability a as a function of
solute concentration in silver alloy filings.
flat specimen from the center of the goniometer). The
true lattice parameter ao can then be found by drawing
a smooth line through all available points ahkl and
extrapolating to 0 = 90 (Figs. 2 and 3, Tables II and
III); the chosen extrapolation function cos
2
0/sinO
yielded a straight line in most cases. The estimated
accuracy in measuring the extrapolated annealed lattice
parameter is 0.OOO3 A due to temperature variations
(the thermal expansion of pure silver
32
is of the order of
0.00008 A per degree Celsius) and an inaccuracy of
0.02 in measuring high angle peak maxima. As
a check the lattice parameters of pure annealed
silver filings (ao=4.0862 A) and tungsten powder
=
a inA
Aq-15In
4.14
4.13
=

4.12

-')C.------ _x ____ !
Aq-2.5In
-'----..,2:!:---'---:!3
cos e/sin9
FIG. 2. Determination of extrapolated lattice parameters for
several annealed and cold-worked Ag-In alloys. Deviations of
individual cold-worked lattice parameters, ahkl, from the extra-
polated line are the result of deformation faulting.
32 W. B. Pearson, Handbook of Lattice Spacings and Structure of
Metals and Alloys (Pergamon Press, London, 1958).
(ao= 3.1650 A) were measured and corrected to 25C;
these values were found to be essentially equal to the
published values for silver
32
(ao=4.0R61 A) and tung-
sten
32
(ao= 3.1650 A).
Similar considerations for a cold-worked material
would yield large scatter of the ahkl values plotted
versus cos
2
0/sinO (Figs. 2 and 3); however, this scatter
is quite consistent with the Paterson theory of stacking
faults.15 Since each individual deviation is a result
of a peak shift caused by deformation faulting, Boyd
and Spreadborough
31
,33 suggested a method for deter-
mination of an apparent lattice parameter of a cold-
worked material, ao(CW), as well as the deformation
fault probability a. The following relation
20
connects the
changes in lattice parameter of each reflection Aa with
the deformation fault probability;
(3)
For a given cold-worked material the shifts are pro-
portional to the planar parameter G
hkl
(Table I). We
can reconcile these displacements due to deformation
faulting and find hypothetical points a' hkl independent
of any deformation faulting effects (i.e., Aahkl represents
the difference between the measured cold-worked lattice
parameter ahkl and the hypothetical lattice parameter
a"'kl). The tabulated values of G
hkl
indicate that the
(200) shift should be twice as much and oppositely
directed (e.g., deformation faults shift a200 to larger
values) as either the (111) or (220) shifts (e.g., defor-
mation faults shift aU! or a220 to smaller values),
whereas the (311) value is shifted one-third as much and
in the same direction as the (200). The (420) and (331)
peaks should also be equally but oppositely displaced
from each other and the (422) should not be shifted at
all. Fitting these conditions to cold-worked materials,
agreement is found with predicted changes if we assume
the Paterson theory15 is valid. These hypothetical
(deformation fault independent) points a' hkl then
comply with the standard linear extrapolation procedure
and result in a determination of an "apparent lattice
parameter" of a cold-worked material (Fig. 2 and 3).
The limit of accuracy in determining the extrapolated
lattice parameter of deformed silver alloy filings was
0.OO08 A (due to temperature fluctuations, a 0.05
inaccuracy in measuring the peak maxima at high
angles, and experimental variations in estimating the
extrapolated line). By using this method no significant
or consistent change from the corresponding annealed
lattice parameters was found for indium alloys (Table
II). For the tin series the cold-worked alloys appeared
to have smaller lattice parameters than those of the
annealed material (Table II).
Additionally, using this method we can find a from
(Aa/ a'hkl without relying on the annealed sample as a
reference. The average values a for the (111), (200), and
(220) peaks agree very closely with the corresponding
33 I. H. Boyd and J. Spreadborough, Acta Cryst. 11, 97 (1958).
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
STACKI.\JG FAULTS IN SILVER-BASE ALLOYS 3455
o
oinA
4.13 cold worked
,
4.13
4.12



4
09E I ! I ! _
. 0 I 2 3
coste/sine
FIG. 3. Determination of extrapolated lattice paramet.ers for
several annealed and cold-worked Ag-Sn alloys. DeviatIOns of
individual cold-worked lattice parameters, ahkl, from the extra-
polated line are the result of deformation faulting.
ones calculated from relative peak shift of neighboring
reflections of cold-worked and annealed samples (Tables
II and III).
c. Peak Asymmetry Measurements
Twin faults asymmetrically broaden a diffraction
peak. Since twin faults are the only cause of asymmetry
arising out of cold work for these alloys, any difference
in position of the center of gravity and the peak maxi-
mum of a cold-worked peak would give an indication
of twin faulting. Cohen and Wagner
34
have derived a
relationship between center of gravity shift and twin
fault probability {3. By combining AC.G. (20), which is
the displacement of the center of gravity from the peak
maximum, for (111) and (200) reflections partial
compensation for systematic and instrumental effects
can be achieved. The following equation allows the
determination of twin fault probability {334;
AC.G. (20) 111- AC.G. (28) zoo
{3
11 tan0l11+14.6 tan8z
o
o
(4)
For all three solute series the twin fault probability {3
increases with alloy content (Tables II-IV). To verify
that the combination of the (111) and (200) center of
gravity displacements compensated for the systematic
and instrumental effects, the value CAe.G. (028
111
)
-AC.G.(028
200
)] (or abbreviating, AAC.G.) was also
M J. B. Cohen and C. N. J. Wagner, J. Appl. Phys. 33, 2073
(1962).
measured for three annealed samples (pure Ag, AAC.G.
=0.005; Ag-9 Sn, AAC.G.=0.009; Ag-15 In, AAC.('.
= 0.001 0); these values were negligible compared to
typical values of AAC.G. between 0.10
0
and 0.40 for
the cold-worked materials. Scatter in values of AAC.G.
was due to errors in evaluating peak tails and in
determining the peak maximum.
D. Peak Broadening Measurements
Stacking faults produce peak profile broadening that
is independent of the order of reflection (i.e., an
apparent particle size effect); therefore, any measured
particle size will contain the true domain size and a
measure of the stacking fault probabilities
I6
a and {3. To
determine the effective particle sizes and strain effects
a Fourier analysis was performed on the K"l peak pro-
files using either Beevers-Lipson strips or a Mader-Ott
harmonic analyzer. Initially correcting for instrumental
broadening with an annealed tungsten powder standard
using the Stokes method,25 the separation of the strain
term from the particle size term was then accomplished
by using the Warren-Averbach technique
16
0n the (111),
(222), (200), and (400) reflections. The intercept of the
initial slope of the order independent Fourier coefficient
ALsF with the abscissa L gives the effective particle
size
36
(De)hkl for a given set of planes (hkl). For the
(111) reflection the following relation holds
1 1 1 v'3
--=-+-+-(1.5a+{3) (5)
(D
e
)1l1 D VlT 4a
and for (200)
1 1 1 1
--=-+--+-(1.5a+{3), (6)
(De) 200 D a
where D is the coherent domain size normal to the
reflecting plane, a and {3 are fault densities, a is the
lattice parameter, and T is the domain size in the
faulting plane. The last term in Eq. (5) and (6) con-
taining the compound faulting probability (1.5a+{3) can
be considered as the reciprocal of the stacking fault
particle size in the defined direction (hkl). There is a
definite decrease in effective particle size as the amount
of either solute (In, Sn) increases (Tables II and III,
Figs. 4 and 5). As defined by Warren
36
a minimum value
of T, T min can be found from Eqs. (5) and (6);
T min=0.82 (7)
These values are given in Tables II and III and indicate
a definite decrease in T min with increasing amounts of
solute (Figs. 4 and 5).
The root mean square strain component (rmss) as a
,6 A. R. Stokes, Proc. Phys. Soc. B61, 382 l1948).
.& B. E. Warren, ]. Appl. Phys. 32, 2428 (1961).
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3456
R. P. I. ADLER AND C. :-.J. J. WAGNER
Atomic Percent Indium
FIG. 4. Variation of reciprocal deformation fault probability
a-I, reciprocal twin fault probability p-l, the effective particle
size in the (hkl) direction (D,hkl, and the minimum particle size
in the plane of the stacking fault T m;n, as a function of indium
concentration.
function of distance L have also been measured for the
(100) and (111) directions using the strain dependent
Fourier coefficients.
16
For all specimens the rmss de-
creased asymptotically with increasing distance L from
the reflecting plane. Taking an arbitrary distance
L=50 A, there is a gradual increase in strain with
increasing solute content (Tables II and III).
III. DISCUSSION
The primary effect of alloying cadmium, indium, and
tin with silver is to increase the stacking fault proba-
bility when the material is severely deformed by filing.
In the range of terminal solid solubility the increase in
deformation fault probability a is an almost linear func-
tion of solute concentration (Tables II-IV, Fig. 1). For
any increment of solute added to silver, tin is most
effective in increasing deformation fault
probability, followed by indium while
cadmium causes the least marked effect
in agreement with recent results obtained by Davies and
Cahn.
24
Similar increases in a have been established in
copper-base alloys.18-22 In fact,comparable compositions
of tin in both silver- and copper-base alloys have similar
values.
Corresponding to the increase in a with alloying there
is also a parallel effect for the twin fault probability {3
which also rises as the concentration of any of the three
solutes increases (Tables II-IV). However, due to
experimental limitations in measuring center of gravity
displacements the continuous variations are not as well
defined.
The values of a for silver-zinc alloys, as reported by
Vassamillet,25 are larger than those reported by Davies
and Cahn.
24
This difference in the fault probabilities for
Atomic Percent Tin
FIG. 5. Variation of reciprocal deformation fault probahility
a-I, reciprocal twin fault probability {3-t, the effective particle size
in the (hkl) direction (D,hkl, and minimum particle size in the
plane of the stacking fault T min, as a function of tin concentration.
silver-zinc alloys may partially be due to the fact that
Vassamillet's values of a were obtained from shifts of
peak centroids. According to Cohen and Wagner
14
measurements, using peak centroids will include an
additional contribution due to twin faults if they are
present in the material. Under these circumstances a
compound probability (a+2.1{3) for the (111)-(200)
couple, as derived from Eqs. (2) and (4), would have
been reported.
The alternate method of finding a from relative
lattice parameter shifts as observed from (111), (200),
and (220) peaks is in very good agreement with peak
shift values for the tin and indium solute series [the
cadmium alloy series was not evaluated in this manner
since the (420) and (331) peaks were not measured
during the initial part of this study, Tables II and III].
Since the (420) and (331) peaks are necessary for a
determination of cold-worked lattice parameters, a
problem is raised as to why the relative lattice
parameter changes for these reflections give values of a
that are too large in magnitude. A probable explanation
is the occurrence of a spacing change at the deformation
fault.
30
The effect of a spacing fault is to add a term to
the planar dependant constant G
hkl
so that Eq. (3) is
altered:
(8)
where E is the relative change in spacing at the fault and
J hkl is a spacing fault constant
30
(Table I). With an
arbitrary (2%) expansion at the layer fault the only
combined planar constants that change significantly
are those for the (331) and (420) reflections (e.g.,
G331= -0.007, G
331
+O.02J
33I
= -0.011, G420= +0.007,
G
420
+O.02J
420
= +0.013). The increase in magnitude of
the (331) and (420) combined planar constants causes
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
STACKING FAULTS IN SILVER-BASE ALLOYS 3457
the anomalously large values of ex to become smaller
and approach the values of ex calculated from (111),
(200), and (220) peaks, but does not appreciably affect
the cold-worked lattice parameter extrapolation. Al-
though spacing faults successfully explain the large
shifts of the (420) and (331), the relative shifts of the
(400), (222), and (311) peaks which do not agree in
magnitude with the predicted values from the Paterson
l5
theory must be the result of some other effect.
Concurrently with the determination of deformation
fault probabilities from relative lattice parameter shifts,
the apparent lattice parameter of the heavily deformed
specimens were measured (Tables II and III). The
presence of tin in deformed silver alloys causes a
decrease in lattice parameter; the maximum decrease
found was ~ O . 1 percent of Ag-9 Sn. Alloying indium
with silver has no significant effect on cold-worked
lattice parameters. Similar effects have been observed in
copper-base alloys; in fact for copper-zinc alloys an
expansion has been observed.
19
,20 The fact that the
lattice parameters of deformed tin and indium alloys
show different trends, although they have similar line
broadening characteristics, indicates that this is a real
effect whose magnitude depends on the type and
quantity of solute present in silver. A greater number
of alloy systems and pure metals must be studied before
any firm conclusions can be made; however, the effect
is probably related to the variation of composition
around a stacking fault (e.g., Suzuki effect
37
).
The particle size due to faulting [Eqs. (5) and (6) ] is
inversely proportional to compound faulting probability
(1.5ex+,S). Since the effective particle size decreases as
reciprocal stacking fault probability decreases (i.e.,
with increasing solute concentration) (Figs. 4 and 5),
stacking faults do contribute significantly to the particle
size broadening. Anisotropic effective particle sizes also
yield information about the particle size in the plane
of the stacking fault T. If stacking faults were com-
pletely extended across a coherent domain dimension D,
then the particle size in the faulting plane T and D
would be approximately equal and relatively large
compared to the domain size due to faulting [i.e., the
"1.5ex+f:l" term in Eqs. (5) and (6)]. In this limiting case
the effect of D and T on effective particle size can be
neglected to a first approximation so that the effective
particle size depends essentially only on the fictitious
domain size due to faulting; the ratio for this condition
of completely extended stacking faults can then be
obtained [(De)I1I/(De)200] = 2.3. The average experi-
mental ratio is 1.7. This indicates that D and more
likely T [because T min is slightly larger than (D
e
)111] do
influence the magnitude of the effective particle size. As
seen from Tables II and III and Figs. 4 and 5, this
minimum particle size in the plane of the stacking fault
Tmin decreases with increasing alloy content and might
37 A. H. Cottrell, Rela,tion of Properties to Microstructure
(American Society for Metals, Cleveland, 1954), p. 131.
be the result of a greater number of stacking faults on a
given (111) faulting plane.
The rms strain distributions for columns of unit
cells normal to both (111) and (200) planes were
measured using Fourier analysis. For all deformed
specimens in both measured directions the strains
decrease asymptotically with increasing distance L
normal to the reflecting plane. For both the tin and the
indium series using an arbitrary distance L=50 A, there
was a gradual increase in rmss in the (111) and (100)
directions as solute content was raised (Tables II and
III). In general for a given amount of solute present and
in either direction, the measured value of rmss
[( 8
2
L=50 ;..)nv]hkZ! in the tin series was higher than in the
indium alloy of corresponding concentration. This,
coupled with the fact that for the same alloy composi-
tion range the tin alloy had the higher value of deforma-
tion fault probability, suggests that the higher density
of dislocations contributed this additional amount of
strain. The existence of the measured anisotropic rmss
within the material cannot be rationalized in terms of
an isotropic stress field where the differences in strains
arise from the variation of the elastic constants with
crystallographic direction. For this ideal isotropic stress
case using the elastic constants for silverS the ratio
would be (8
100
/8
111
)= (E
111
/E
200
) =2.7. The average
experimentally observed value (8
100
/8
111
) = 1.4 falls be-
tween the ideal cases of isotropic stresses and isotropic
strains. Thus the presence of dislocations and stacking
faults which do have stress directionality27 probably
influence this experimentally measured strain ratio.
Since the addition of polyvalent solutes to a mono-
valent solvent metal increases the deformation fault
probability, the influence of the average number of
chemical valence electrons per atom on the stacking
fault energy was considered.
14
,23,24 Plotting ex as a func-
tion of electron concentration e/ a for the solutes tin
(Sn+
4
), indium (In+3) , and cadmium (Cd+
2
) in silver-
base alloys, the values for ex of the indium and tin alloys
fell on the same curve whereas those for the cadmium
alloys at higher concentrations deviated below the
common curve for indium and tin alloys (Fig. 6).
Interestingly, if the recent results for copper base-tin
alloys20 are also plotted in Fig. 6, the values of a
fall on the same common curve as that for the silver-tin
and -indium alloys. This phenomenological behavior
indicates the importance of both the solute specie and
the electron concentration on the occurrence of stacking
faults. However, the results for the silver-cadmium
alloys show that other variables (e.g., atomic misfit)
should also be included to adequately describe the
variation of the deformation fault probability in cold-
worked fcc alloys with solute concentration.
It is of interest to compare the values for compound
faulting probabilities (1.5o:+f3)M+A obtained from peak
38 C. S. Barrett, Structure of Metals (McGraw-Hill Book
Company, Inc., New York, 1952), p. 533.
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3458
R. P. I. ADLER AND C. N. J. WAGNER
'0
".40

;.;
:t:
30
..0
o
L-
a..
= 20
u:
8
'B 10
E

-.
+.
' .
-+
+
+
Alloy;
Aq-Cd
Aq-In
+ Aq-Sn
-Cu-Sn
o
1.0 1.1 1.2 1.3 114
Electron Concentration, eja
FIG. 6. Variation of de-
formation fault probability
'" as a function of electron
concentration e/a. Values
for copper base-tin alloys
taken from reference 20 .
shift (a) and asymmetry ({:/) with the corresponding
term (1.5a+{:/)B calculated from anisotropic particle
sizes (Tables II and III). Since the faulting probabilities
were derived by two different methods (recognizing the
limits of accuracy involved), the essential agreement is
an experimental verification of the theory and sub-
stantiates the experimental measurements.
IV. CONCLUSIONS
(1) Alloying cadmium, indium, and tin with silver
increases the probability of stacking faults in these cold-
worked alloy filings with respect to pure silver.
(2) Corresponding to the increase in stacking fault
frequency as a function of alloying, there is a significant
decrease in the effective particle sizes and the domain
size in the plane of the stacking fault.
(3) Using the relative changes in lattice parameter
calculated from the individual reflections, both defor-
mation fault probability and lattice parameter of a
deformed material can be determined.
(4) After cold-working there was a detectable de-
crease in lattice parameter of the tin but not the indium
ailoys.
(5) Values of the compound faulting probability
(1.5a+{:/) obtained from asymmetry and peak maxima
measurements were consistent with the ones obtained
from anisotropic particle size measurements.
ACKNOWLEDGMENT
This work was conducted under contract with the
Office of Naval Research.
JOURNAL OF APPLIED PHYSICS VOLUME 33, NUMBER 12 DECEMBER 1962
Formation Conditions and Structure of Thin Epitaxial Germanium
Films on Single-Crystal Substrates*
BILLY W. SLOOPE AND CALVIN O. TILLER
Virginia Institute for Scientific Research, Richmond, Virginia
(Received April 4, 1962)
An experimental investigation of the effects of formation conditions on the structural characteristics of
thin Ge films vacuum deposited onto synthetic single crystals of CaF
2
, NaCI, NaF, and MgO is reported.
Formation conditions include substrate temperature during deposition, rate of deposition, and heat treat-
ment. The amorphous to crystalline transformation of Ge was found to occur in the 300-350C substrate
temperature range. It is shown that single--crystal films, 1500 A thick, can be formed on CaF
2
substrates
at temperatures between 450
0
and 700C by proper choice of rate of deposition. Crystalline structure
complexity of imperfections, and film adhesion are dependent on the rate of deposition and
tlOn temperature,
INTRODUCTION
E
XTENSIVE research has been carried out on
various aspects of growth, structure, and physical
properties of thin evaporated films. Many of these
studies have made use of single-crystal films formed by
deposition onto single-crystal substrates, a process
called epitaxy. In a previous study on epitaxial Ag
filmsl it was found that formation conditions such as the
rate of deposition, deposition temperature, and anneal-
ing treatments controlled the microstructure as well as
the crystalline structure of the deposited film. There-
fore, in order to produce films of a particular material
* Supported by the United States Department of Defense.
1 B. W. Sloope and C. O. Tiller, J. Appl. Phys. 32, 1331 (1961).
with specific structural characteristics, it is necessary to
make use of particular formation conditions. These
structural characteristics also depend upon the sub-
strate on which the film is deposited and, at present, it
is not possible to choose a substrate with assurance that
a single-crystal film can be formed epitaxially. This is
due to the fact that a satisfactory theory of the mecha-
nism of epitaxial growth has not been developed. This
report contains additional experimental information on
the relation between orientation and formation condi-
tions which may be of value in understanding the epi-
taxial process.
Recently, researches have been directed toward the
use of epitaxial semiconductor films because of their
Downloaded 10 May 2011 to 157.138.23.35. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

You might also like