Composites Engineering, Vol. 5, No. 7, pp. 935-950, 1995
Copyright © 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0961-9526/95 $9.50 + .00
Pergamon
0961-9526(95) 00034-8
ARTIFICIALLY LAYERED NANOCOMPOSITES
FABRICATED BY JET VAPOR DEPOSITION
H . N. G. W a d l e y , L. M. H s i u n g a n d R. L. L a n k e y
Department of Materials Science and Engineering, School of Engineering, Thornton Hall,
University of Virginia, Charlottesville, VA 22903-2442, U.S.A.
(Received 20 December 1994; final version accepted 20 February 1995)
Abstract--Novel jet vapor deposition (JVD) processes offer considerable promise for the
inexpensive synthesis of functionally graded (composite) materials (FGMs). Here, we explore
microstructure-mechanical property relationships for a model A1/Cu metal-metal system and an
A1/A120~ metal-metal oxide multilayered nanocomposite system fabricated by the JVD process.
The 10 pm thick A1/Cu multilayers were deposited on silicon wafers at a substrate temperature of
-140°C. The A1 and Cu layers were of approximately equal thickness and were systematically
varied from -20 to ~ 1000 nm. The 20pm thick A1/A1203 multilayers were deposited on glass
slides at -250°C. The oxide layer thickness was held constant in the -2-6 nm range, whilst the
AI layer thickness was systematically varied from - 3 to -50 nm. The structure of the AI/Cu
multilayers was polycrystalline and had a strong l111] texture, whereas the AI/A12O3 multilayers
consisted of amorphous aluminum oxide layers and polycrystailine metal layers with randomly
oriented grains. The yield strength of the AI/Cu multilayers exhibited an inverse dependence upon
layer thickness when the layer spacing exceeded -50nm. When the AI/Cu layer spacing was
thinner than -50 nm, the strength was better predicted by a Koehler image force model. A similar
phenomenon was also found in the AI/AI203 multilayers. In this case the critical metal layer thickness for the transition from an Orowan to a Koehler type behavior was approximately 25 nm. This
is consistent with theoretical predictions which indicate that the critical layer thickness of the low
modulus consistuent decreases as the difference in shear moduli between the two constituent layers
increases.
1. INTRODUCTION
N o v e l v a p o r p h a s e synthesis m e t h o d s are b e g i n n i n g to emerge with the c a p a b i l i t y o f
creating c o m p o s i t e structures with i n d e p e n d e n t l y c o n t r o l l e d v o l u m e f r a c t i o n s a n d thickness o f two or m o r e c o n s t i t u e n t s ( H s i u n g et al., 1993, 1995). T h e y offer the o p p o r t u n i t y
to synthesize l a m e l l a r c o m p o s i t e s with n a n o m e t e r scale d i m e n s i o n s a n d to v a r y the layer
t y p e a n d spacing t h r o u g h o u t the c o m p o s i t e ' s thickness, thus creating g r a d e d m i c r o s t r u c t u r e materials. H e r e , we e x p l o r e the use o f a novel jet v a p o r d e p o s i t i o n (JVD)
process d e v e l o p e d b y Jet P r o c e s s C o r p o r a t i o n , New H a v e n , C T ( H a l p e r n , 1982; Schmitt,
1988; H a l p e r n et al., 1992; H a l p e r n a n d S c h m i t t , 1994). T h e J V D process uses sonic, high
p u r i t y h e l i u m (He) gas jets to e n t r a i n a t o m i c a l l y d i s p e r s e d v a p o r a n d d e p o s i t f i l m s / c o a t ings. It is o f interest here b e c a u s e several jets can be used sequentially to d e p o s i t e multil a y e r e d m a t e r i a l s inexpensively at high rates. Its p o t e n t i a l value for this is e x a m i n e d by
d e p o s i t i n g a n d m e a s u r i n g the strength o f a m o d e l m e t a l - m e t a l a n d a p r o t o t y p i c a l
m e t a l - m e t a l oxide system.
1.1. Background and theory
M a t e r i a l s with thin, m u l t i l a y e r e d structures are o f interest because t h e y p o t e n t i a l l y
o f f e r novel m e c h a n i c a l a n d p h y s i c a l p r o p e r t i e s ( C o h e n et al., 1983; B a r n e t t , 1993). T h e
m o v e m e n t o f d i s l o c a t i o n s in these structures has been an a r e a o f g r o w i n g interest
( E m b u r y a n d H i r t h , 1994). K o e h l e r (1970) used an i m a g e force m o d e l to a n a l y z e the
strength o f l a y e r e d m a t e r i a l s o f very thin layer thicknesses. H e p r e d i c t e d t h a t if one o f the
layer c o n s t i t u e n t s h a d a significantly lower shear m o d u l u s , d i s l o c a t i o n s w o u l d have to
o v e r c o m e a large repulsive i m a g e force in o r d e r to m o v e f r o m m e t a l layer B o f lower shear
m o d u l u s (PB) into the A m e t a l layer (with shear m o d u l u s g n >/~B). This i m a g e force ( F )
is given a p p r o x i m a t e l y by:
F = R/.t B b 2 sin O/4nr
935
(1)
936
H . N . G . Wadley et al.
where R = (/~A -- /~B)/(/~A + /tB), b is the magnitude of Burger's vector in metal layer B,
0 is the angle between the A / B metal interface and the slip plane, and r is the distance
between the dislocation and the nearest interface. Equation (1) indicates that the image
force is proportional to the difference in shear moduli for the two metal layers and rapidly
increases as the dislocation approaches the interface. The maximum stress (rm = F/b) will
be reached when the dislocation reaches the interface, i.e. when r is approximately equal
to the radius of a dislocation core ( - 2 b ) . In this case r m~ Rp B sin O/8n and is independent
of the B layer thickness 0-B). When Koehler's mechanism controls yielding and if isostrain
conditions are assumed to apply, application of a rule-of-mixtures model to the yield
stress of a multilayer sample with equal A and B layer thicknesses gives:
try = R/.tB sin 0[1 + (EA/EB)]/16nm
(2)
where EA and EB are Young's moduli of A and B, respectively, and m is the Schmid
factor for the operative slip system. Again, this is independent of the layer thickness. For
different A, B layer thicknesses, the yield strength under isostrain conditions would be:
tTy = e f l B sin 0{[,~B/(,~ A + '~B)] q- [(~'A/2A if" "~B)(EA/EB)]}/87rm
(3)
where 2 A and 2B are the layer thicknesses of A and B, respectively.
Lehoczky (1978) experimentally demonstrated that the yield strength of Al/Cu
multilayers with equal A1 and Cu layer thicknesses conformed to the Koehler model
when the layer spacing was below a critical value of - 7 0 nm. However, when the layer
thickness was greater than - 7 0 nm, the strength was better fitted to a Hall-Petch type
relation:
try = tr* + k " 2~"
(4)
where a* is a frictional stress, k is a constant, n lies between unity and 0.5, and 2B is the
B layer thickness. These results suggest that an Orowan or Frank-Read type of
mechanism operates in the A1/Cu multilayers and governs yielding for 2B > - 7 0 nm. In
this regime, try is strongly dependent on the metal layer thickness. On the other hand, the
Koehler mechanism only controls yielding when the metal layer is so thin that Frank-Read
a n d / o r Orowan mechanisms are not operative in either layer. In this situation, the movement of dislocations towards and into the interface would presumably be the only
mechanism by which plastic deformation could occur.
To estimate the strength in this case, suppose a slip plane within metal layer B is
inclined to the A / B interfaces at an angle and intercepts the interfaces at P and Q (Fig.
1 (a)). The plan view of the slip plane with a single dislocation loop is shown in Fig. l(b).
The dislocation loop is confined in a very small space (J.B/sin 0) between two A / B interfaces with two dislocation segments pinned along the A / B interfaces, and the rest of the
loop is movable in the slip plane. The movable dislocation segments can be viewed as
Orowan dislocations. The critical resolved shear stress (re) required to move the Orowan
dislocations is:
"tf -- 2~/~Bb sin 0/2B
(5)
where o~ is a constant between 0.5 and unity (Hull and Bacon, 1984). As ;tB is decreased,
rf increases. For 2B less than a critical thickness, rf eventually exceeds the maximum image
stress (rm -~ R/tB sin 0/8z0, and the image force mechanism will then be responsible for
yielding. The critical thickness (2c) at which this occurs can be found by setting rf equal
to r m from which:
2c ~ 16~rab/R.
(6)
Thus, multilayers with a metal layer thickness less than a critical thickness, defined by
eqn (6), are expected to exhibit a different deformation behavior from that observed in
any monolithic material or in multilayers with a layer thickness greater than 2c. We also
note that, according to eqn (6), the greater the shear modulus difference (R), the smaller
the critical metal thickness for the transition from a Hall-Petch to a Koehler type
937
Artifically layered nanocomposites
-----U
(a)
-d
)
o
::::!:iiiii~i~iiii~iii!i~!iiU
Fig. 1. Schematic representations of: (a) the geometry of a slip plane within a low modulus
constituent layer; (b) a dislocation loop confined in a small space between two A / B interfaces with
two dislocation segments pinned along the A / B interfaces and the rest of the loop movable in the
slip plane.
behavior. Together, eqns (2), (3), (5) and (6) suggest that modifying layer spacings and
thicknesses will provide great flexibility for independently manipulating many of the
mechanical properties (such as modulus, strength, toughness . . . . ) of nanocomposite
materials. If they can be shown to be a valid description of the strength-nanostructure
relationship, they could be used to design the mesoscale structure gradients of future
FGMs.
1.2. Synthesis techniques
If tomorrow's FGM designs are to be realized, inexpensive, flexible approaches for
their near net shape synthesis/processing are needed. There are many potential approaches
for fabricating layered nanocomposites. Directionally solidified eutectic reactions have
been traditionally used to produce in situ composites with lamellar structures. In particular,
Ni/Ni3AI-Ni3Nb ( 7 / Y ' - ~) or Ni3AI-Ni3Nb ( 7 ' - ~ ) lamellar eutectics (Thompson
and Lemkey, 1974; Stoloff, 1976) have been extensively explored. The yield strength
of the Ni 3A1-Ni 3Nb lamellar eutectics obeyed a Hall-Petch type relation. For example,
the yield strength increased from -750 MPa at a 5/~m lamellar spacing to 1.5 GPa at a 1 ~m
spacing. However, attempts to further elevate the strength by decreasing the spacing below
1/~m have proven unsuccessful because of the onset of cellular solidification and a
consequent loss of the lamellar structure. Other techniques for synthesizing layered
materials include physical vapor deposition (Springer and Catlett, 1978; Bickerdike et al.,
1984-85; Alpas et al., 1990; Chou et al., 1992) and electrodeposition (Foecke and
Lashmore, 1992). Two advanced physical vapor deposition techniques--electron-beam
vapor deposition (EBVD) and jet vapor deposition (JVD)--show particular promise for
938
H.N.G.
Wadley et al.
the synthesis of graded layer nanocomposites because of their potentially high deposition
rate (and thus reduced cost) and their multi-material capability.
Alpas et al. (1990) recently reported the tensile properties of A1-A1203 laminar
composites deposited at ambient temperature in an EBVD system using a pulsed gas
process. The thickness of the aluminum oxide layer was kept constant at - 5 nm, whilst
the thickness of the metal A1 layer was varied from 50 nm to 500 nm. Only a Hall-Petch
type behavior was observed in this case. The yield strength (try) increased to a maximum
of -475 MPa when the A1 layer thickness decreased to 50 nm, and they fitted their data
to a Hall-Petch type equation: try(MPa) ~ 27 + 0.1 • 2-1/2(m-t/2). It is possible that the
metal layer thickness was not thin enough to exhibit a Koehler type behavior. When
Al/Al203 multilayers with the Al layer thicknesses less than -50 nm are deposited using
EBVD techniques, one problem is that the pulsed 02 "spills-over" oxygen into the Al
layer (Springer and Catlett, 1978). The origin has been thought to be residual gas
circulating within the deposition chamber (while the metal is being deposited) as a
consequence of the slow vacuum pumping speed and fast duty cycle needed for
fabricating nanoscale multilayers. This problem is significantly reduced with the "fast
flow" JVD approach.
A schematic representation of the JVD process is shown in Fig. 2 (Halpern and
Schmitt, 1994). The JVD vapor source is a nozzle, incorporated in a "low vacuum"
(several hundred Pa), mechanically pumped, fast flow system. High purity helium carrier
gas emerges from the nozzle as a highly collimated sonic or near sonic jet. Atomic or
molecular vapor, generated in the nozzle by evaporation, is entrained in the jet and forms
a localized deposit on a downstream substrate. One or more jets can be aimed at a
rotating/oscillating substrate carousel. The moving substrates are thereby "scanned" and
coated uniformly. Multiple jets operated simultaneously yield alloys, or if operated
sequentially, give multilayered composites. Because the JVD process incorporates a fast
flow helium carrier gas and a powerful pumping system, the oxygen residence time in the
JVD deposition chamber is short. Substrate temperatures can also be kept low, therefore
avoiding layer interdiffusion. However, when JVD fabricated multilayers are deposited at
too low a temperature (Hsiung et al., 1993, 1995), metal layers in the multilayer samples
tend to be discontinuous, and aluminum oxide can fill the voids between the Al clusters,
creating discontinuous metal layers. This can be overcome by slightly elevatedtemperature deposition (Hsiung et al., 1995).
lnert
s in
.:.::::i'i:!:i
-1 torr ~
Vapor source
~:~iiilCollimatedgas jet
Substrate~__
"*"7
~
Localizeddeposit
Fig. 2. Schematic representation of jet vapor deposition (1 torr = 133.3 Pa). The deposit vapor is
generated near the nozzle exit by mechanisms such as thermal vaporization. The dotted arrows
indicate substrate motion in two dimensions for large area coverage and uniformity.
Artificially layered nanocomposites
939
2. EXPERIMENTAL PROCEDURES
2.1. Specimen preparation
In this study, aluminum-copper multilayers were grown by means of an A1 and a Cu
"wire feed" jet source, whilst the aluminum-aluminum oxide multilayers were grown by
means of an AI "wire feed" jet source in association with a pulsed 02 source. Both the
A1 and Cu vapors were created in separated nozzles by feeding aluminum and copper
wires under computer control to a resistively-heated (BN-sheathed) tungsten filament. To
grow the A1/AlzO3 multilayers, high purity (99.99°7o) molecular 02 was periodically
admitted through a pulsed valve just downstream of the nozzle. When the O2 flux was off,
Al layers were deposited; when the O 2 flux was on, aluminum oxide was formed by reactive deposition at the surface of the growing film. The pulsed valve duty cycle, Al wire
feed velocity, and carousel rotation frequency were chosen to allow alternating deposition
of nanometer dimension Al and aluminum oxide layers at nominal rates of - 1 0 n m / s over
an area of - 5 0 cm 2. Either silicon wafers (AI/Cu) or glass slides (A1/AI203) were used as
substrates. The A1/Cu multilayers were - 1 0 tim thick and had approximately equal thicknesses of A1 and Cu. The samples were deposited at a substrate temperature of - 1 4 0 ° C ,
and the alternating layer thickness was systematically varied from - 2 0 to -1000 nm. The
20/~m thick A1/AI203 multilayers were deposited at substrate temperatures of either
- 2 5 ° C or - 2 5 0 ° C . The oxide layer thickness was held constant between - 2 and - 6 nm,
whilst the A1 layer thickness was systematically varied from - 3 nm to - 5 0 nm.
2.2. Microstructural characterization and hardness measurement
Microsctructures of as-deposited A1/Cu and AI/A12Oa multilayer samples were
examined using a JXA-840A scanning electron microscope (SEM) and a Phillips-400T
transmission electron microscope (TEM). Both plan-view and cross-sectional TEM
specimens were prepared from multilayer samples by standard dimpling and ion-milling
procedures. To reduce irradiation damage, ion-milling was performed at 4 V and 0.5 mA
at liquid nitrogen temperature.
Hardness of both the A1/Cu and the A1/AIzO 3 samples were measured at the
National Institute of Standards and Technology (NIST) using a nanoindentation technique (Oliver and Pharr, 1992) on a Nanolnstruments Nanolndenter II equipped with a
triangular pyramid diamond tip. The sides of the pyramid made an angle of 65.3 ° with the
normal to the base of the indenter. The indentations were made at a maximum load of
100 mN in a load control mode. Loads were recorded at 1 mN load increments. Hardness
values were corrected for thermal drift, machine compliance, elastic recovery and
deviation of the indenter tip from perfect Berkovich geometry (Oliver and Pharr, 1992).
3. RESULTS AND DISCUSSION
3.1. Microstructure
3.1.1. AI/Cu multilayers. Figures 3(a) and (b) show the distinct multilayered
structure of AI/Cu samples of nominal layer spacings of 100nm and 1000nm,
respectively. In the micrographs, the bright contrast layers are copper; those with a darker
contrast are aluminum. Figure 4(a) is a plan-view TEM micrograph showing a typical
microstructure observed in an as-deposited A1/Cu multilayer sample (with a nominal
layer spacing of 100 nm). The appearance of Moir6 fringes in Fig. 4(a) indicates that the
A1 and Cu layers are coherent. A preferred [111} growth direction can be readily seen
from the selected-area diffraction (SAD) patterns (Fig. 4(b)) generated from the region
shown in Fig. 4(a). Both Figs 4(a) and 4(b) reveal that the microstructure of the asdeposited AI/Cu multilayers is polycrystalline but strongly {111 ]-textured. The grain sizes
of A1 and Cu varied between - 10 nm and - 2 0 0 nm. The grain size of the aluminum was
usually greater than that of copper.
When the interface was carefully examined, fine domains (several nm in lateral size)
of an A1Cu3 intermetallic phase were occasionally observed. Since they were extremely
940
H.N.G.
Wadley et al.
small and only occasionally observed, they are presumed to have a minor effect on the
strengthening of the multilayers.
3.1.2. AI/AI203 multilayers. Figures 5(a) and 6(a) are plan-view, bright-field TEM
images showing typical microstructures observed in AI/AI20 3 multilayer samples
deposited at - 2 5 ° C and -250°C, respectively. Clearly, the continuity of the metal layers
is significantly improved by an elevated-temperature deposition. The encapsulation of AI
clusters by oxide in the - 2 5 ° C deposited samples is no longer observed in the -250°C
deposited samples; AI grains formed within the -250°C deposited samples are also larger
in size ( - 1 0 0 nm to -200 nm) compared to the AI grains ( - 5 to - 5 0 nm) formed within
the - 2 5 ° C deposited samples. Notice that only the reflections corresponding to crystalline FCC A1 could be found in the selected-area diffraction (SAD) patterns shown in Figs
5(b) and 6(b), indicating that the aluminum oxide layers deposited at both - 2 5 and
-250°C are amorphous. Figure 7 is a bright-field, cross-sectional TEM image showing
the layer structure of an A1/A1,O3 sample (with an A1 layer thickness of - 5 0 n m )
deposited at -250°C. Notice that the length of the AI grain (as marked by A) is about
twice the metal layer thickness. We also see that the metal layer has few grain boundaries
parallel to the layer (i.e. both the top and bottom of each aluminum grain is usually in
contact with the alumina layers).
3.2. Hardness and estimated yield strength
Hardness measurements were made on twelve AI/Cu (all deposited at -140°C) and
six AI/A12 O3 (all deposited at -250°C) multilayer samples of each layer thickness. The
indentation contact length, hc (nm), hardness value, H (GPa), and estimated yield
strength, Oy (MPa), for each different layer thickness 2B are listed in Tables 1 and 2 for
A1/Cu and A1/AI 203 , respectively.
3.2.1. AI/Cu multilayers. The average contact depths of the 100 mN load indentations for the A1/Cu laminate samples decreased from 1607 + 30 nm to 1189 5:67 nm
when the nominal layer spacing decreased from 1000 nm to 20 nm. This corresponded to
an average hardness (H) increase from 1.54 + 0.06GPa for 2B ~ 1000nm to
2.8 + 0.29 GPa for 2B = 20 nm. These hardness values correspond to estimated yield
strengths ( - H / 3 ) of 512 5:19 MPa and 933 + 98 MPa, respectively. To investigate the
yield strength-layer spacing relation for the JVD A1/Cu multilayers, the data points were
plotted against the inverse of the layer spacing, Fig. 8. The results reveal a similar
behavior to that reported by Lehoczky (1978) on evaporated A1/Cu multilayers" the yield
strength appears governed by an Orowan mechanism when the layer thickness was greater
Table 1. Hardness and yield strength of A1/Cu multilayers
Sample
no.
/1.B
(nm)
1
2
3
4
5
6
7
8
9
10
11
12
1000
500
200
100
80
70
60
50
40
30
25
20
hc
(nm)
1607
1588
1305
1264
1228
1164
1238
1126
1213
1287
1020
1189
±
±
±
±
±
±
±
±
±
±
±
±
H
(GPa)
30
38
19
39
73
31
27
71
74
5
45
67
1.54
1.58
2.32
2.47
2.64
2.90
2.56
3.12
2.70
2.38
3.76
2.80
±
±
±
±
±
±
±
±
±
±
±
±
0.06
0.08
0.06
0.15
0.30
0.14
0.12
0.36
0.33
0.02
0.30
0.29
try
(MPa)
512
526
772
825
879
968
854
1039
901
793
1255
933
±
±
±
±
±
±
±
±
±
±
±
±
19
26
21
50
99
48
38
121
110
6
100
98
Note: 2 a = nominal A1 layer spacing, h c = the contact depth, H = calculated
hardness, try = estimated yield strength. The error cited is + 1 standard deviation of
the experimental values.
Fig. 3. Cross-sectional SEM micrographs showing layer structures of A1/Cu multilayers with nominal layer spacings of: (a) 100 nm; (b) 1000 nm.
4~
C)
?
g~
>
Fig. 4. (a) Plan-view bright-field (BF) TEM image showing a typical microstructure observed in an A1/Cu multilayer sample (nominal layer spacing of 100 rim); (b) selected-area
diffraction patterns generated from the region in (a) showing a preferred 1111] growth direction of the multilayer.
e,
0
Z
b~
rig. 5. (a) Plan-view, bright-field TEM image showing a typical microstructure observed in an A1/A1203 sample deposited at -25°C; (b) selected-area diffraction patterns generated
from the region in (a).
v,"
Fig. 6. (a) Plan-view, bright-field TEM image showing a typical microstructure of an A1/Ala O 3 multilayer deposited at -250°C; (b) selected-area diffraction patterns generated from
the region in (a).
t~
t~
Z
Artifically layered nanocomposites
Fig. 7. Bright field, cross-sectional TEM image showing layer structures observed in an A1/AI203
multilayer sample (AI layer thickness -50 nm) deposited at -250°C.
945
Artificially layered nanocomposites
947
Table 2. Hardness and yield strength of AI/AI203 multilayers
Sample
no.
it s
(nm)
1
2
3
4
5
6
50
25
15
10
5
3
he
(nm)
1631
1351
1339
1328
1311
1274
±
±
±
±
±
±
H
(GPa)
50
18
16
72
16
19
1.50
2.17
2.20
2.29
2.30
2.43
±
±
±
±
±
±
try
(MPa)
0.08
0.06
0.05
0.23
0.05
0.06
500±
723 ±
733 ±
763 ±
767 ±
810 ±
;t/As
27
2
17
77
18
20
0.42
0.55
0.54
0.59
0.76
1.03
Note: 2 B = A1 layer thickness, h c = the contact depth, H = calculated hardness, ay = estimated yield strength. The error cited is + 1 standard deviation of the
experimental values.
than a critical thickness of - 5 0 nm. The best fit was obtained with a ).~1 relation: O'y
(MPa) = 538.6 (MPa) + 25375.2 (MPa nm)" 2~1 (nm-1).
The theoretical yield strength predicted by Koehler's equation (2) is also included in
Fig. 8. It was assumed that EA = 101.4 GPa (measured by indentation), EB = 58.6 GPa
(measured by indentation), /ZA = EA/(1 + V) = 39.0GPa, where v (=0.3) is Poisson's
ratio,/Za = 22.6 GPa, A denotes Cu and B denotes AI. Predictions are shown for both
equiaxed and 1111}-textured (m = 0.244) multilayers deforming under isostrain conditions. The estimated yield strength of the JVD A1/Cu samples with alternating layer
thickness smaller than - 5 0 nm is better predicted by a Koehler type relation. The critical
layer spacing of - 5 0 nm for transition from a Hall-Petch to a Koehler type relation is
close to the thickness range (depending on ct) of 26.9-53.8 nm predicted by eqn (6), using
b = 0.286 nm and R = (/zA -/za)/(/z A +/Za) = 0.266.
Embury and Hirth (1994) have recently proposed an alternative view of the dislocation mechanism of yielding in multilayered materials. They incorporated the interaction
energy of the dislocation with the A-B interface in the Orowan relation. This results in an
expression for the shear yield strength of the form:
2Se In (~--~)
r = b2B
(7)
Layer thickness / nm
1400100o
1oo
; .5341.6(MPa)
Oy
50
.
. :
h
30
25
20
/"/"
:
Q..
~S
40
1
looo
T
(dislocationspacin9
e-
2
800
~
~
.,_
N,
ID
6OO
W
400
20O
0.00
~/¢Koehler'smodel-equiaXedtexlute
0.01
0.02
0.03
Inverse layer thickness,
0.04
0.05
~. B1/ rim-1
Fig. 8. A plot of the yield strength versus the nominal layer spacing of the JVD A l / C u multilayers.
The m e a n deviation is indicated by vertical bars. The curve fit to a Hall-Petch type relation and the
yield strength predicted by a Koehler image force model are also included.
H.N.G. Wadley et al.
948
where 2 is the distance between dislocations and Se is the edge dislocation interaction
energy with the interface boundary. The predicted 2B dependence is plotted on Fig. 8
using a best fit value for 2 of 0.44 2a for all samples.
3.2.2. A//AI2 03 multilayers. The average contact depths of the 100 mN load indentations for the A1/Al203 samples decreased from 1631 + 50 nm to 1274 ± 19 nm when
the metal layer thickness (2B) decreased from - 5 0 nm to - 3 nm. This corresponds to an
average hardness (H) increase from 1.50 ± 0.08 GPa for ;t B -~ 50 nm to 2.43 ¢ 0.06 GPa
for '~a ~ 3 nm, and the yield strengths are 500 + 27 MPa and 810 ± 20 MPa, respectively. In order to compare the results of JVD A1/A1203 to the results of EBVD Al/AI203
reported by Alpas et al. (1990), the estimated yield strengths of JVD A1/Al203 were
plotted versus inverse square-root of the Al layer thickness, Fig. 9. Here, both the
extrapolated curve based on Alpas's result (1990), try (MPa) --- 27 + 0.1 • ,~.-1/2(m-1/2),
and the predicted yield strength based on eqn (3) are included. Note that with A denoting
AI2 03 and B denoting AI, values of EA (for amorphous alumina) = 150 GPa (Bickerdike
et al., 1984-85), E B = 58.6 GPa, BA(amorphous) = 58 GPa, BB = 22.6 GPa, )'A = 2 nm,
2B = 3-100 rim, and an average Schmid factor m = 0.5 were used to calculate the yield
strengths of AI/A120 3 multilayers with randomly oriented polycrystalline AI layers
deformed under isostrain conditions. As can be seen from Fig. 9, although only two data
points are available (which are insufficient for a line fit), the estimated yield strength of
the JVD AI/A12 0 3 samples with A1 layer thicknesses greater than - 2 5 nm is consistent
with an Orowan type relation suggested for the EBVD AI/AI203 (Alpas et al., 1990),
whereas the strength of samples with an AI layer thickness smaller than - 2 5 nm is better
predicted by the Koehler image force model. Note that the critical A1 layer thickness
of - 2 5 nm is within the thickness range of 16.4-32.8 nm theoretically predicted by
eqn (6), where b = 0.286 nm and R = (PA -- PB)/(PA + PB) = 0.439 were used for the
calculation.
The Embury and Hirth model poorly fit the data when a single 2/2B ratio was used
for all the samples. By treating the dislocation spacing parameter, 2, in the Embury and
Hirth model as a free " f i t t i n g " parameter for each layer spacing, it was possible to fit
eqn (7) to the data. The best fit )./2a for each layer thickness are given in Table 2. The
2/). B ratios are difficult to reconcile with the usually inverse dependence of 2 on strain
ke/nm
1.1
50
25
15
10
5
I
I
I
~
I
L"
3
/
I
1.0
0.9
0.8
"O
0.7
E
aa
0.6
i
~.,'
W
0.5
#
ae
0.4
0.0
I
!
t
I
I
I
I
I
I
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
~.B1/2 X 1 0 - 4 / m 1/2
Fig. 9. A plot of the yield strength versusthe AI layer thickness of the JVD A1/AI203 multilayers.
The mean deviationis indicated by verticalbars. The extrapolationof the curve fit to Alpas's data
(1990) and the curve predicted by a Koehler image force model are also included.
Artificially layered nanocomposites
949
(i.e. he). The "best fit" dislocation spacing, 2, decreased from - 2 0 nm for )~B ---- 50 nm
to --3 nm for 2 B = 3 nm, while for the same samples hc (and therefore the strain)
decreased from 1631 to 1274 nm.
3.2.3. Nanocompositing implications. Two possible deposition schemes for creating layered materials from two constituents have been demonstrated. In Scheme (I), the
thickness of equally thick layers (i.e. equal volume fractions) is systematically varied. In
Scheme (II), the layer thickness of the low modulus constituent is systematically varied
whilst that of the high modulus constituent is held constant (i.e. unequal and varied
volume fractions). The yield strength of layered materials deposited by both schemes
increases with decreasing layer thickness. When the layer thickness of the lower modulus
constituent is greater than a critical thickness, the data are consistent with an Orowan type
mechanism, and the strength varies approximately inversely with the layer thickness. The
existing theoretical relationships appear suitable for designing layer architectures in this
layer spacing regime. However, when the layer thickness of the low modulus constituent
is less than a critical thickness, the yield strength of the materials deposited by both
schemes appears to be controlled by an interaction with the interface. Both the Koehler
image force model and the Embury and Hirth mechanism result in similar agreement with
the data for the AI/Cu (Scheme I) system. However, poorer agreement with data is seen
in the AI/A120 3 system. The Koehler model predicts that the greater the shear modulus
difference between two constituent layers, the smaller will be the critical layer thickness
where interface interactions take over, and this is supported by the experiments conducted
here. The yield strength of the materials deposited by Scheme (I) is shown to level off
when the layer thickness is thinner than the critical thickness, whereas that of the
materials deposited by Scheme (II) still increases with decreasing layer thickness because
of the increase in the volume fraction of higher modulus constituent.
4. SUMMARY
Model A1/Cu metal-metal and A1/A1203 metal-oxide multilayered nanolaminates
have been successfully fabricated by a novel jet vapor deposition (JVD) process. The
microstructure and mechanical properties of these JVD fabricated multilayered nanocomposites have been investigated using scanning electron microscopy (SEM), transmission
electron microscopy (TEM), and nanoindentation methods. The structure of the A1/Cu
multilayers was polycrystalline and had a strong Ill 1] texture, whereas the A1/A1203
multilayers consisted of amorphous aluminum oxide layers and polycrystalline metal
layers with randomly oriented grains. The yield strength of the Al/Cu multilayers was
controlled by an Orowan type process when the layer spacings exceeded - 5 0 nm, whereas
that of the nanocomposites with a layer thickness smaller than - 5 0 nm is controlled by
interfacial interactions. This transition is consistent with predictions of the Koehler
model, which indicates that the critical layer thickness of the low modulus constituent
decreases as the difference in shear moduli between the two constituent layers increases.
A similar phenomenon was also observed in the A1/A1203 multilayers. In this case the
critical metal layer thickness for the transition from an Orowan type to an interface controlled behavior was approximately 25 nm. This study has demonstrated two possible
deposition schemes which can be implemented for making functionally graded
nanolaminates. In Scheme I, the thickness of equally thick layers (i.e. equal volume
fractions) is systematically varied. In Scheme II, the layer thickness of the low modulus
constituent is systemically varied whilst that of the high modulus constituent is held
constant. Together, they offer interesting possibilities for independently varying the local
modulus and strength in functionally graded materials.
Acknowledgements--This research was supported by the Advanced Research Projects Agency(Mr W. Barker,
program manager), and the National Aeronauticsand Space Administration (Mr R. Hayduk, program monitor)
under grant No. NAGW 1692. The authors would like to express their gratitude to Mr J. Schmitt and Dr B.
Halpern of Jet Process Corporation (JPC) for providing access to their JVD facilitiesfor preparing the samples
used for this research. Two of the authors (R. L. Lankey and L. M. Hsiung) would also like to express their
950
H . N . G . Wadley et al.
thanks to Dr J. Z. Zhang, J. W. Golz and S. M. Karecki of JPC for their technical assistance during the preparation of the samples. The authors are also indebted to Dr D. T. Smith of NIST for his helpful assistance whilst
performing the nanoindentation experiments and during the interpretation of the hardness measurement results.
REFERENCES
Alpas, A. T., Embury, J. D., Hardwick, D. A. and Springer, R. W. (1990). The mechanical properties of
laminated microscale composites of A1/Al203. J. Mater. Sci. 25, 1603-1609.
Barnett, S. A. (1993). Deposition and mechanical properties of superlattice thin films. In Physics o f Thin Films
(Edited by M. H. Francomhe and J. L. Vossen), Voi. 17, p. 1.
Bickerdike, R. L., Clark, D., Easterbrook, J. N., Hughes, G., Mair, W. N., Partridge, P. G. and Ranson,
H. C. (1984-85). Microstructures and tensile properties of vapour deposited aluminum alloys. Part 1: Layered
microstructures. Int. J. Rapid Solidif. 1, 305-325.
Chou, T. C., Nieh, T. G., McAdams, S. D., Pharr, G. M. and Oliver, W. C. (1992). Mechanical properties and
microstructures of metal/ceramic microlaminates: Part II. A Mo/Ai203 system. J. Mater. Res. 7(10),
2774-2784.
Cohen, U., Koch F. B. and Sard, R. (1983). Electroplating of cyclic multilayered alloy (CMA) coatings. J. Electrochem. Soc. 130, 1987.
Embury, J. D. and Hirth, J. P. (1994). On dislocation storage and the mechanical response of fine scale microstructures. Acta Metall. 42(6), 2051-2056.
Foecke, T. and Lashmore, D. S. (1992). Mechanical behavior of compositionally modulated alloys. Scripta
Metall. Mater. 27, 651-656.
Halpern, B. L. (1982). Fast flow deposition of metal atoms on liquid surfaces: gold sol formation and surface
lifetime. J. Colloid Interf. Sci. 86, 337.
I-Ialpern, B. L. and Schmitt, J. J. (1994). Jet vapor deposition. In Handbook of Deposition Technologies for
Films and Coatings (Edited by R. F. Bunshah), p. 822. Noyes Publications, Park Ridge, NJ.
Halpern, B. L., Schmitt, J. J., Di, Y., Golz, J. W., Johnson, D. L., McAvoy, D. T., Wang, D. and Zhang,
J. Z. (1992). Jet vapor deposition of single and multicomponent thin films. Metal Finishing 37, December.
Hsiung, L. M., Zhang, J. Z., McIntyre, D. C., Golz, J. W., Halpern, B. L., Schmitt, J. J. and Wadley,
H. N. G. (1993). Structure and properties of jet vapor deposited aluminum-aluminum oxide nanoscale
laminates. Scripta Metall. Mater. 29, 293.
Hsiung, L. M., Lankey, R. L., Wadley, H. N. G., Smith, D. T., Zhang, J. Z., Golz, J. W., Halpern, B. L. and
Schmitt, J. J. (1995). Jet vapor deposited aluminum=aluminum oxide nanolaminates. In Proc. Int. Syrup.
Novel Techniques in Synthesis and Processing of Advanced Materials (Edited by J. Singh and S. M. Copely).
TMS, Warrendale, in press.
Hull, D. and Bacon, D. J. (1984). Introduction to Diffocations, 3rd edn. Pergamon Press, Oxford.
Koehler, J. S. (1970). Attempt to design a strong soli¢~. Phys. Rev B2, 547-551.
Lehoczky, S. L. (1978). Strength enhancement in thin-layered A1-Cu laminates. J. Appl. Phys. 49(11),
5479-5485.
Oliver, W. C. and Pharr, G. M. (1992). An improved technique for determining hardness and elastic modulus
using load and displacement sensing indentation experiments. J. Mater. Res. 7(6), 1564-1582.
Schmitt, J. J. (1988). U.S. Patent No. 4, 788,082, November 29.
Springer, R. W. and Catlett, D. S. (1978). Structure and mechanical properties of Al/AlxOy vacuum-deposited
laminates. Thin Solid Films 54, 197-205.
Stoloff, N. S. (1976). Composite strengthening. In Alloy and Microstructural Design (Edited by J. K. Tien and
G. S. Ansell), p. 65. Academic Press, New York.
Thompson, E. R. and Lemkey, F. D. (1974). Directionally solidified eutectic superalloys. In Metal Matrix
Composites (Edited by K. G. Kreider), pp. 539. Academic Press, New York.