2-PHASE METALLIC THERMAL INTERFACE MATERIALS

PROCESSED THROUGH LIQUID PHASE SINTERING

FOLLOWED BY ACCUMULATIVE ROLL-BONDING

r
Fo
Journal:

Manuscript ID:
Manuscript topic:

Date Submitted by the Author:

TCPMT-2015-036.R1
COMPONENTS: CHARACTERIZATION & MODELING
01-May-2015

Pe

Complete List of Authors:

Transactions on Components, Packaging and Manufacturing Technology

er

Sharma, Deepak; Indian Institute of Science, Materials Engineering

Tiwari, Rajesh; National Institute of Foundry and Forge Technology
(NIFFT), Manufacturing Engineering
Sharma, Rohan; Indian Institute of Technology, Kanpur, Department of
Materials Science and Engineering
Narayanan, Ramesh; Vikram Sarabhai Space Centre, Material
Characterisation Division
Kumar, Praveen; Indian Institute of Science, Bangalore, Materials
Engineering

ew

vi

Re

Keywords:

Page 1 of 16
Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

2-PHASE METALLIC THERMAL INTERFACE MATERIALS

PROCESSED THROUGH LIQUID PHASE SINTERING FOLLOWED
BY ACCUMULATIVE ROLL-BONDING
Deepak Sharma, Rajesh Kumar Tiwari, Rohan Sharma, Ramesh Narayan P. and Praveen Kumar

Abstract Thermal interface materials (TIMs) form

mechanical and thermal link between a heat source and a heat
sink. Thus, it should have high thermal conductivity and high
compliance to transfer heat efficiently and accommodate any
differential strain between the heat source and the sink,
respectively. This study reports on the processing and
characterization of potential metallic TIM composite solders
comprising of Cu, a high conductivity phase, uniformly embedded
in In matrix, a highly compliant phase. We propose fabrication of
such a material by a 2-step fabrication technique comprising of
liquid phase sintering (LPS) followed by accumulative rollbonding (ARB). To demonstrate the efficacy of the employed 2step processing technique, an In-40 vol. % Cu alloy was produced
firstly using LPS with short sintering periods (30 or 60 s at 160
C) followed by ARB up to 5 passes, each pass imposing a strain
of 50%. Mechanical response and electrical and thermal
conductivities of the fabricated samples were evaluated. It was
observed that processing through ARB homogenizes the
distribution of Cu in In matrix, disintegrates agglomerates of Cu
powder and also, dramatically increases the thermal and
electrical conductivities, almost attaining theoretically predicted
values, without significantly increasing the flow stress.
Furthermore, the processing technique also allows insertion of
desired foreign species, such as reduced graphene oxide, in the InCu for enhancing a target property, such as electrical
conductivity.

r
Fo

microelectronic devices demands development of new material

systems, most probably metal based, with (1) very high
thermal conductivity, , (2) high shear compliance with
moderate compressive stiffness under creep conditions, and (3)
low melting temperature for ease of reworking. Most often a
natural or pure material does not possess all of the above
properties; for example, Cu is thermally more conductive than
a polymer based thermal grease (Cu ~ 400 W/m-K and Ag filled
polymer ~7 W/m-K [10]); however, Cu is much stiffer than the
polymer. To simultaneously attain the above characteristics in
a material, a composite architecture, as shown in Figure 1,
comprising a high thermal conductivity phase (HTCP) (e.g.,
diamond, Cu, Ag, etc.) uniformly distributed in a high
mechanically compliance phase (HMCP) (e.g., In, Sn, etc.)
matrix, has been proposed [1,2]. Theoretically it has been
shown that a material with such architecture possesses very
high thermal conductivity, owing to presence of HTCP [1,2] as
well as high mechanical compliance attributed to the
continuous matrix of HMCP, which accommodates most of the
imposed strain [4].

er

Pe

I. INTRODUCTION

continuous increase in the computational power,

degree of miniaturization and adoption of 3-dimensional
(3-D) packaging philosophy in the advanced and
futuristic microelectronic devices is destined to drastically
increase the requirements for heat dissipation via thermal
interface materials (TIMs) which cannot be met by the existing
polymer based TIMs [1-3]. Hence, the packaging of advanced
Manuscript received February 02, 2015. This work was financially
supported by IISc-ISRO STC (ISTC0305) and DAE, India (DAEO0162). We
acknowledge help of Amit Kumar Verma of IISc, Bangalore in conducting a
few of the reported experiments.
D. Sharma and P. Kumar are with the Department of Materials
Engineering, Indian Institute of Science, Bangalore, 560012, India
(corresponding author (P. Kumar); phone: +91-80-22933369; fax: +91-8023600472; e-mail: [email protected]).
R. Sharma is with Indian Institute of Technology, Kanpur. R. K. Tiwari
was earlier with NIFFT, Ranchi and is currently with Indian Railways
R. Narayanan P is with the MCD/MMG, VSSC, ISRO, Trivandrum.

ew

Index Terms Thermal interface materials; 2-phase metallic

solders; liquid phase sintering; accumulative roll bonding;
mechanical and thermal characterization

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Fig. 1. Schematic of idealized microstructure with a uniform distribution of

HTCP and HMCP. The above material architecture is based on the design
proposed in ref [1].

The HMCP can be chosen in such a way that its melting

temperature is low to facilitate a low rework temperature.
Since the stiffness of materials, particularly metals, generally
increases with their melting temperature, it is possible to find a
HMCP which also has a low melting temperature. One such
example is In which has a melting temperature of only 157 oC
and elastic modulus and yield strength of ~11 GPa and ~5
MPa, respectively.
Sintering, especially liquid phase sintering (LPS) of a
HTCP and a HMCP above the melting temperature of the
HMCP has been suggested as a preferred method to produce a
composite solder with the architecture shown in Fig. 1 [13,5,6]. In these studies [1-3,5,6], a composite system with Cu
as the HTCP and In as HMCP was selected. One of the major
advantages associated with In-Cu system is that Cu has one of

Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

the highest thermal as well as electrical conductivity amongst

widely used metals whereas In is one of the softest metals.
Besides having high thermal conductivities, a metallic system
is expected to also possess high electrical conductivity; this
can qualify these metallic composite solders also for package
level interconnects in sensitive microelectronics.
Preliminary studies on In-Cu system showed that even
though In-Cu composite solders can, in theory, attain very high
thermal conductivity and high compliance [2,4], their
mechanical and thermal properties drastically deteriorated due
to the formation of numerous Cu-In intermetallic compounds
(IMCs), such as Cu16In9, Cu11In9, CuIn, etc. [3,11-13].
However, interfacial engineering involving coating Cu
powders with a ~10 to ~100 nm thick layer of Au or Al2O3 was
shown to significantly slow down the kinetics of reaction
between Cu and In, inhibiting the growth of IMCs during LPS
as well as under simulated service conditions [3,6]. As
expected, the compliance of In-Cu solders increased with
volume fraction of In but, at the same time, their thermal and
electrical conductivities decreased [3]. Finally, it was proposed
that In40 vol. % Cu solder had the optimum combination of
conductivity and mechanical compliance, which if interfacially
engineered showed 1.5 and 2 times larger thermal conductivity
and yield strength, respectively, than that of pure In [6]. Thus,
In-40 vol. % Cu solder with the architecture shown in Fig. 1
and produced through LPS can potentially be applied as TIM
as well as package level interconnects [3,6].
Although In-Cu solders produced through only LPS showed
properties suitable for thermal management and interconnect
applications, a critical observation of the microstructures
produced in the previous works [2,3,6] and as will be shown
latter in this study reveals that: (i) the dispersion of Cu
particles was non-uniform in In matrix1, resulting in
agglomeration of Cu dispersoids in certain regions; this will
result in non-uniform compliance (and equivalently, stiffness)
and form hot-spots potentially leading to pre-mature failure of
the component, and (ii) the minimum thickness of the In-Cu
composites cannot be arbitrarily small, as the thickness of such
a composite is controlled by the size of Cu powders and their
agglomerates. The latter makes these materials unsuitable for
the advanced microelectronic packages, which have space only
for few hundred micrometers thick TIM. Furthermore, LPS for
short period, which is important condition for minimizing
formation of IMCs, often leaves large fraction of residual
pores (~10 vol. %) and results in inferior contact between In
and Cu having large interfacial thermal resistance.
This work addresses the aforementioned challenges by
suggesting implementation of a 2-step composite processing
route comprising of LPS followed by accumulative roll
bonding (ARB) [7], which is a severe plastic deformation
(SPD) processing technique. It is speculated that ARB will not
only thin down the In-Cu solder to a desired thickness but will,
while maintaining the architecture shown in Fig. 1, also
increase the uniformity in mixing of Cu particles in the In
matrix as well as will improve the contact between In and Cu.
In addition, ARB also allows embedding desired foreign

r
Fo

phases, such as graphene, carbon nanotubes, etc., as distinct

layers within the In-Cu solder which can further increase the
conductivity and also, by acting as a lubricating layer,
increase mechanical compliance in the shear. Furthermore,
ARB also produces materials in sheet form, which is a
preferred form for easy integration of TIM into a
microelectronic package.
II. EXPERIMENTAL PROCEDURE
This work used In-40 vol. % Cu composite solder as the
experimental material because, as mentioned earlier and shown
in reference [3], the as-sintered In-Cu composite solder with
this composition possesses the optimum combination of
compliance and conductivity required for TIM and
interconnect applications.
Spherical Cu powders with average size of 10 m and
99.9% purity, were etched using 10% HCl solution for 10
minutes to remove the native oxide layer. The etched powder
was thoroughly rinsed using deionized water and iso-propyl
alcohol (IPA) and dried by storing in vacuum (10-4 Torr) for
12 hours. An appropriate amount of dry Cu powder (40 %
volume fraction) was mixed with10-40 m diameter, 99.99 %
pure In powder by vigorously shaking in a glass-vial inside a
glove-box filled with dry N2 gas. Subsequently, the mixed
powder was uniaxially compressed by using a hydraulic press
with a load of 4 MPa for 45 seconds, in a lubricated hardened
steel die to produce green pellets of 10.1 mm diameter and
1.5 mm height with a relative green density, defined as the
density of the green pellets, of 85 1%. The density of a green
pellet was measured by dividing its mass by its volume. The
density values, as measured using aforementioned geometry
method, were confirmed to be consistent with that of
Archimedes principle. To study the effect of green density on
the sintering process, a few samples were prepared with
different green densities. These samples were prepared using
different compaction pressure while all other parameters, such
as weight of powders, sintering time, etc., were kept constant.
The green pellet was tightly encapsulated in Al foil, and
LPS was performed at 160 C by dipping it into a pre-heated,
properly agitated2 400 mL silicone-oil bath for either 30 or 60
s, followed by quenching in large volume of water kept at
room temperature. The quenching step was observed to be
beneficial in inhibiting the formation of IMCs by limiting the
time available for reaction at high temperatures. The density of
sintered samples was measured using Archimedes principle so
as to avoid the errors caused by possible deviation in the
cylindrical shape of the pellets during sintering.
ARB, ranging from 1 to 5 numbers of passes3, and each
imposing a strain of ~50%, were conducted on the sintered
samples; Figure 2 schematically shows the slicing and stacking
steps of the ARB processing. As shown in Fig. 2, a sintered
sample was rolled, sectioned in two equal halves along the
original longitudinal direction, degreased by generously
applying acetone and IPA, vertically stacked, and again rolled

er

Pe

ew

1
A non-uniform distribution of Cu in In is probably due to poor mixing of In
and Cu owing to sticking of soft In on the walls of the mixer.

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 2 of 16
2

Agitation of oil was conducted using a magnetic stirrer for establishing a

uniform temperature distribution
3
A few samples were processed up to 13 ARB passes. However, these
samples were only used for thermal and electrical conductivity analysis.

Page 3 of 16
Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

for next cycle. For demonstrating the prowess of the proposed

2-step processing technique for embedding desired foreign
species, a few samples were also prepared where flakes of the
reduced graphene oxide (rGO), produced through chemical
exfoliation method [8], were placed in between the two
surfaces of the sample prior to the ARB processing.

Fig. 2. Main steps of ARB processing, involving slicing and stacking of the
rolled sample, employed in this study.

r
Fo

As-sintered pellets were cross-sectioned along transverse

direction while the ARB processed samples were crosssectioned along both transverse and longitudinal directions for
microstructural characterization. After metallographically
polishing up to 1 m diamond paste followed by 0.05 m
colloidal silica, the samples were observed under a scanning
electron microscope (SEM) (FEI Quanta 200), using both
secondary electrons (SE) and back-scattered electrons (BSE).
Since a suitable TIM should have both high thermal
conductivity and high mechanical compliance, a processing
route for producing TIM must not only improve the thermal
conductivity but should also produce highly compliant
microstructure. Hence, assessment of ARB on the mechanical
behavior of the proposed TIM is critical. Mechanical behavior
of the composite solders was evaluated under compression
using a universal testing machine (Instron 5967) equipped with
a 5 kN load cell. Samples were tested at room temperature at
various strain rates, ranging from 10-4 to 10-1 s-1.
The electrical conductivity of the composite solders was
measured using a custom designed fixture, as shown in Figure
3. For eliminating any contact resistance and to measure
conductivity with high accuracy, the developed fixture
employed the 4-wire probe method in conjunction with a nanovoltmeter (Keithley 2010). As shown in Fig. 3, the electric
current entered and exited the sample at the ends 1 and 2,
respectively, whereas the voltage drop was measured between
points 3 and 4. Usage of non-conducting ceramics and
Teflon baseplate precluded any electrical short-circuiting
between various electrical probes. The set-up was validated by
measuring electrical conductivity of pure Cu within an
accuracy of 5%.
Since in metallic materials, electronic conduction is mainly
responsible for the flow of both electric current and heat, the
electrical and the thermal conductivities are often proportional
to each other. Thus, the electrical conductivity, as determined
using the set-up shown in Fig. 3, was used to calculate thermal
conductivity of the metallic composite solders according to
Wiedemann-Franz Law given as follows:

= LT

(1)

where and are the thermal and the electrical conductivities,

respectively, L is Lorenz number, and T is the test temperature,
which was room temperature (i.e., ~303 K) in the present
study. Accordingly, the following steps were taken to
determine thermal conductivity of an In-Cu sample: (i) a
master plot showing the room temperature electrical and
thermal conductivities of pure metals, such as Al, Cu, Au, Sn,
In, etc., on abscissa and ordinate axes, respectively, was
prepared and a best fit line passing through all datum points
was drawn, (ii) electrical conductivity of a sample was
measured using the set-up shown in Fig. 3 and marked on the
abscissa of the master-plot, and (iii) thermal conductivity value
corresponding to the electrical conductivity data was
determined using the best fit curve drawn on the master plot.
Previous studies on as-sintered In-Cu solder composites have
shown a little difference between the thermal conductivity
values measured using the above indirect method and a direct
method using the laser flash analysis [2,3]. Thus, the thermal
conductivity of In-Cu composite solders as measured by this
method in this study should be reasonable.

er

Pe

Re

Fig. 3. Custom built fixture for measuring electrical resistivity (and

conductivity) of a sample using 4-wire method. The parallelepiped shaped
sample was mechanically affixed on the non-conducting platform using the
probes passing the current and the potential drop along the length was
measured using tungsten probes. The long axis of the samples with rGO
layer(s) was placed along direction 1-2.

ew

vi

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

III. EXPERIMENTAL RESULTS AND DISCUSSIONS

A. Microstructural Characterization
Figure 4a representatively shows the effect of green density
and sintering time on the densification of In-40 vol. %Cu (InCu) composite solder. In Fig. 4a, the density values are given
in terms of relative density (calculated by taking percentage
with respect to fully dense or theoretical density of composite
that is 7.962 g/cm3) and densification was calculated by taking
the increase in the density relative to the initial density. As
shown in Fig. 4a, a densification of >10 % occurred when the
green density was small (<85 %) and the densification
monotonically decreased with the green density. A decrease in
densification with green density can be attributed to the
increase in the volume fraction of closed pores which, unlike
open pores, cannot be removed during the short duration of the
flash LPS process employed in this study. Interestingly, there
was a negligible effect of sintering time on densification. This
suggests that the capillary flow induced densification was

Page 4 of 16
4

Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

completed in first 30 seconds and the solution and

reprecipitation, which is the second stage of LPS, did not
occur within the first 60 seconds.
94

12

In - 40 Vol. % Cu

Relative density (%)

Densification (%)

In - 40 Vol. %Cu

Sintered at160 C

10
8
6
4

Sintering time (s)

2
0

60
30
84

85

86

87

88

89

Relative Green Density (%)

90

Sintered at 160 for 60 s

92
90

(a)

(b)

88
86
84
0

Number of ARB passes

(a)
(b)
Fig. 4. Effects of (a) green density and sintering time on densification of In40 vol. % Cu composite solder and (b) ARB processing on the relative
density of the as-sintered samples.

r
Fo

Figure 4b shows the effect of ARB on the relative density of

In-Cu composite solder. Interestingly, the sintered samples did
not monotonically densify as a result of ARB processing: At
first, the density of the sample decreased; however, then it
increased monotonically approaching almost the as-sintered
density value after 5 passes of ARB. As will be shown later,
the decrease in the relative density of samples after the first
pass of ARB can be attributed to the collapse of Cu
agglomerates which were not well bonded under the employed
LPS conditions. Following the first pass of ARB, the bonding
between Cu and In improved due to the plastic deformation of
In leading to a monotonous increase in the density of In-Cu
samples. It should be noted that the densification became
saturated after several passes of ARB and a maximum density
of ~92-95 % was attained.
Figure 5 shows representative SEM micrographs of assintered samples. As shown in Fig. 5, closed pores or cavities
(black regions) were observed in the as-sintered samples. Such
an occurrence is usual in the samples produced through LPS
for short period of time because the sample was not fully
densified after such processing (Fig. 4a). Furthermore, Fig. 5a
clearly shows that the distribution of Cu in In was not uniform
and, as shown at high magnification in Figs 5b and 5c, there
were several regions where Cu powders were agglomerated
with little In in between them and vice versa. Such Cu
agglomerates did not have good bonding; this may explain the
initial drop in the density following the first ARB pass (Fig.
4b). Also, such inhomogeneous distribution of In and Cu in the
composite solder, which is an inherent outcome of LPS of Cu
and extremely soft In, may also lead to formation of
discontinuous matrix of In leading to formation of hard
skeleton of Cu (and hence leading to a decrease in compliance
of solder) and hot spots. A comparison of Figs 5b and 5c
reveals that microstructure of samples sintered for 60 seconds,
except for slightly improved homogeneous distribution of Cu
in In, was not significantly different than that attained after
only 30 seconds of sintering; this microstructural evidence is
consistent with the trend observed in Fig. 4a.

(c)
Fig. 5. SEM micrographs of as-sintered samples produced after LPS at 160 oC
for (a) 30 seconds (low magnification) (b) 30 seconds (high magnification)
and (c) 60 seconds (high magnification). Images were acquired using BSE.
Hence, the dark and the bright regions are Cu and In, respectively, whereas
black regions denote pores.

Figure 6 shows the effect of ARB processing on the

microstructure of In-Cu composite solders. It should be noted
that the microstructure of the ARB processed samples was
isotropic as it did not depend on the orientation of plane of
observation (i.e., same on both longitudinal and transverse
planes). A comparison of Figs 5 and 6 readily reveals the
following effects of ARB processing: (i) although ARB did not
eliminate closed pores, these pores became mainly confined
inside In and away from the Cu/In interface, (ii) Cu
agglomerates were broken and dispersed uniformly in In, and
(iii) In formed a continuous matrix around islands of Cu.
Although closed pores should be removed under large
compressive stress, the remnant pores may be attributed to
spring back action following rolling and, to some extent, to the
poor bonding between sliced and stacked surfaces. Since, as
shown in Fig. 4b, the density of In-Cu samples following
several passes of ARB was comparable to the density of the
as-sintered sample and was monotonically increasing with
ARB passes, the desired effect of homogeneous dispersion of
Cu in the continuous In matrix and the improvement of Cu/In
interfaces clearly indicate that ARB processing of liquid phase
sintered samples produced microstructures close to the ideal as
shown in Fig. 1.

er

Pe

ew

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

(a)

(b)

Page 5 of 16
Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology
125

50

In - 40 Vol. %Cu

In - 40 Vol. %Cu
Sintered at 160 for 30 s
ARB ( = 50 %) = 1

d = 8.9 3.6 m

d = 9.4 4.4 m

30
20
10
0

(c)
Fig. 6. High magnification SEM micrographs of In-40 vol. % Cu composite
solders processed through ARB for (a) 1, (b) 2 and (c) 5 passes. All samples
were sintered for 30 seconds at 160 oC. SEM images were acquired using
BSE. The dark and bright regions represent Cu and In, respectively, and the
tiny black regions show pores.

75
50
25

7 9 11 13 15 17 19
Diameter range (m)

7 9 11 13 15 17 19
Diameter range (m)

(a)

(b)

150
In - 40 Vol. %Cu
0

Sintered at 160 for 30 s

ARB ( = 50 %) = 5

125

d = 8.4 3.1 m

100
Count

Chemical analysis using electron probe micro-analyzer

(EPMA) was conducted to evaluate the extent of formation of
IMCs after several passes of ARB. Figure 7 shows
representative atomic concentration maps revealing the
distribution of Cu and In after 5 passes of ARB. Fig. 7 clearly
shows that large volumes of In and Cu remained unreacted
during ARB processing and the thickness of IMCs at Cu/In
interface was less than 1 m. This is dramatically (~ 10 times)
smaller than those reported for In-Cu composite solders
without using any diffusion or reaction barrier in between Cu
and In (i.e., the reacting species) [2].

Sintered at 160 for 30 s

ARB ( = 50 %) = 2

100
Count

Count

40

r
Fo

75
50
25
0

7 9 11 13 15 17 19
Diameter range (m)

(c)
Fig. 8. Histograms showing Cu dispersion size in In-40 vol. % Cu samples
processed through ARB for (a) 1, (b) 2 and (c) 5 passes. All samples were
sintered at 160 oC for 30 seconds. The legend also shows the average size (d)
and the corresponding standard deviation.

B. Mechanical Characterization
Figure 9 shows the effect of sintering time on stress-strain
() behavior of as-sintered samples at strain rates varying
from 10-3 s-1 to 10-1 s-1. To reduce the scatter in the data, all
tested samples had relative green density of ~85% and an
aspect ratio of ~0.5. Fig. 9 reveals that the flow stress of In-Cu
composite solders increased with the sintering time, which can
be attributed to an increase in the volume fraction of IMCs
with sintering period, and also with the applied strain rate. We
can speculate some additional strengthening due to Cu/In
interface, which may act as the dislocation source due to the
mismatch in thermal expansion coefficients and elastic moduli
of Cu and In [3]. With a few exceptions, all samples showed
limited strain hardening. This may be attributed to the fact that
for In, which accommodates most of the imposed strain [4],
the test temperature was equal to ~0.7Tm, where Tm is melting
temperature, enabling rapid recovery. Interestingly, the yield
strength of the as-sintered In-Cu sample was only 2 to 3 times
that of pure In at these strain rates.

er

Pe

The size of Cu dispersions was determined by measuring at

least 400 of them after each ARB pass; Figure 8 shows the
measured size distribution for a few processing conditions.
Fig. 8 clearly reveals that continued ARB processing not only
decreased the average size of Cu dispersions but also narrowed
the distribution of the particle size. The above findings are
consistent with the representative micrographs shown in Fig. 6
and support the proposed hypothesis of homogenization of Cu
dispersion size with ARB processing. Although a decrease in
the dispersion size increased the total interfacial area;
however, it should be noted that the ARB processing also
improved the quality of Cu/In interface resulting in a decrease
in interfacial thermal resistance (Rcth). The effect of this
decrease in Rcth on thermal conductivity of In-Cu samples was
drastic and will be discussed latter.

2.5 10

In - 40 Vol. % Cu
Sintered:160oC

2 10

True Stress (Pa)

Fig. 7. EPMA micrographs showing the atomic distribution of In (top image)

and Cu (bottom image) in a In-40 vol. %Cu solder after 5 passes of ARB. The
In-Cu sample was sintered for 30 s.

ew

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Sintering time: 30 s
1.5 107
Sintering time: 60 s
1 107
Strain Rate (s-1)
-1
10
-3
10

5 106
0
0

0.25

0.5

True Strain

Fig. 9. True stress-strain plot of as-sintered In-40 vol. % Cu samples at

various strain rates. For plotting stress-strain data for 30 s sintered samples,
strain axis was shifted by 0.25.

Page 6 of 16
6

Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

To further analyze the effect of strain rate on the

mechanical response of as-sintered In-Cu composite solders,
the yield strength (YS) and the strain rate sensitivity was
calculated as follows:

= 0.2% = YS = C& m

(2)

where C is a constant and m is strain rate sensitivity. Figure 10

shows the effect of strain rate and also, sintering time on the
yield strength of the as-sintered samples. As shown in Fig. 10,
the as-sintered In-Cu composite solders showed significant
strain rate sensitivity. However, Fig. 10 also reveals that an
increase in the sintering period not only decreased the strain
rate sensitivity but also significantly reduced the scatter in the
data, improving the curve fitting parameter, R, relative to Eq.
2. As mentioned previously (Fig. 5), both observations can be
attributed to the relatively improved distribution of Cu in In.
7

3 10

In - 40 Vol. % Cu

r
Fo

Sintered:160 oC
Sintering time (s)
30
60

YS-0.2 %

(Pa)

2 10

9
8
7
6

106
10
106
6
10
6
10

5 106
10-5

y = 1.5e+07 * x^(0.067) R= 0.73

y = 1.7e+07 * x^(0.023) R= 0.94
10-4

10-3

10-2

-1

10-1

Strain rate (s )

Pe
100

Fig. 10. Variation of yield strength of as-sintered In- 40 vol. % Cu samples as

function of applied strain rate.

C. Electrical and Thermal Characterization

The electrical conductivities of as-sintered samples
produced following 30 and 60 s of sintering were 2.1 106
and 5.9 106 /m, respectively. This increase in electrical
conductivity with the sintering period can be attributed to the
improvement in both the distribution of Cu in In and the
contact between In and Cu. After 2 passes of ARB, the
electrical conductivity of sample sintered for 30 s increased to
1.4 107 /m, clearly indicating beneficial effects of ARB on
increasing the electrical conductivity.
Figure 12 shows the effect of ARB on thermal conductivity,
as predicted by Eq. (1) (i.e., Wiedemann-Franz Law) of In-40
vol. % Cu samples sintered for 30 s. Fig. 12a shows that
thermal conductivity of as-sintered samples was considerably
smaller as compared to pure In and Cu, and this can be
attributed to poor densification of the composite solder and
existence of pores, especially inside Cu agglomerates, as
shown in Fig. 5. However, the conductivity of the In-Cu
composites significantly increased with the ARB passes. As
shown in Fig. 12a, the thermal (as well as the electrical)
conductivity of In-Cu composites became greater than that of
pure In and even Ni after 13 passes of ARB: A maximum
thermal conductivity of 151 W m-1K-1 was recoded after 13
passes of ARB, which is ~2 times that of pure In and 65 % and
40 % of Al and Cu, respectively. Such an improvement is
comparable to that attained after placing a thin coat of Au or
Al2O3 on Cu powders, i.e., through interfacial engineering
route as reported previously [3,6].
Since, as shown in Fig. 6, the shape of Cu particles
remained spherical even after several passes of ARB, it can be
concluded that most of the strain during ARB also was
endured by the soft In phase. Hence, it is speculated that ARB
of interfacially engineered as-sintered samples may not
deteriorate the interfacial layer. Therefore, application of the
proposed 2-step process for fabricating interfacially
engineered In-Cu composite solders may further improve the
thermal conductivity as well as retain its value during service.

er

Figure 11 shows the stress-strain behavior of samples

sintered for 30 seconds and processed through ARB. Careful
observation of Fig. 11 in conjunction with Fig. 9 reveals the
following: (i) the flow stress of In-Cu composite solder
monotonically increased with ARB passes; however, the
increase was marginal after the first pass, (ii) similar to assintered samples, the samples processed through ARB also
showed considerable strain rate sensitivity, and (iii) with an
increase in ARB passes, the increase in the strain hardening
rate became saturated. A relative saturation in stress-strain
behavior of In-Cu composite solder after a few passes of ARB
can be attributed to the fact that ARB proffered attainment of
continuous channels of In (Fig. 6), which then accommodated
most of the imposed strain, as envisioned and predicted for
such material systems [2,4].

Sintered:160 C, 30 s

True stress (Pa)

-1

In - 40 vol. % Cu

Au

Sintered: 160 C for 30 s

400

Wiedemann-Franz Law
Circles: Pure metals
Squares: In-Cu

300

Cu

Ag

Al

200

13 Passes
ARB passes

100

In
Sn

Ni

Pb
1

1.5 107

(b)

Electrical Conductivity (x 10 / -m)

1 10

-1

ARB Pass Strain Rate (s )

-1

5 106

10

-3

10

10-3

0
0

0.1

0.2

0.3

0.4

0.5

True Strain

Fig. 11. True stress-true strain behavior of In-40 vol. % Cu samples processed
through 1 and 2 passes of ARB and tested at room temperature at different
strain rates.

200

In-Cu

-1

-1 -1

500

Thermal Conductivity (W m K )

2 10

(a)
Thermal Conductivity (W m K )

In - 40 Vol. % Cu

ew

2.5 10

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

FEM
Maxwell-Garnett
LPS only
LPS+ARB

150

ARB passes

100

50

0.1

0.2

0.3

0.4

0.5

0.6

Volume Fraction of Cu

Fig. 12. Variation of thermal conductivity of In-vol.40 % Cu samples

processed through ARB: (a) Thermal conductivity calculation using the
master plot of pure metals prepared based on Wiedemann-Franz Law as
explained in Section II and (b) comparison of experimental data (LPS only
and LPS + ARB) with the theoretical predictions using FEM and MaxwellGarnett theory for spherical inclusions. The experimental data of LPS only
and theoretical predictions in (b) were taken from references [3] and [4],
respectively. The master-plot in (a) was prepared using the reported values
of electrical and thermal conductivities of pure metals. Thus, the exact
value of Lorentz numbers for elements was not used in preparing this plot.

Page 7 of 16
Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

Fig. 12b is a comprehensive plot showing the effect of only

LPS and the 2-step composite process technique comprising of
LPS followed by ARB on the thermal conductivity of In-Cu
composites. Fig. 12b also shows the prediction of finite
element modeling (FEM) [4] as well as Maxwell-Garnett
theory for spherical inclusions. The Maxwell-Garnett equation
for spherical inclusions is given as follows [9]:
p + 2 m + 2 f p ( p m )

(3)
= m
p + 2 m f p ( p m )
where f is the volume fraction and the subscripts m and p
represent matrix and particles, respectively. It should be noted
that both FEM and Maxwell-Garnett theory neglect the
interfacial thermal resistance between different constituents of
a composite.
Interestingly, Fig. 12b reveals that continued ARB
processing can lead to attainment of the theoretical thermal
conductivity suggesting negligible effect of interface thermal
resistance; this is consistent with previous discussion on a
decrease in Rcth with ARB (Figs 6 and 8). Hence, although the
maximum value of thermal conductivity attained in this study
is still smaller than that of Al or Cu, it clearly shows that ARB
can greatly improve the thermal conductivities of In-Cu
composite solders.
Fig. 13 schematically summarizes the effect of ARB on the
microstructure of the In-Cu composite solder. As shown in Fig.
13, ARB affects the microstructure of In-Cu samples in mainly
the following 4 ways: (i) disintegration of Cu agglomerates
and establishment of a relatively uniform distribution of Cu in
continuous channel of In, (ii) reduction and uniformaization of
Cu dispersion size, (iii) reduction in volume fraction of pores
and their isolation away from Cu/In interface, and (iv)
improvement in the Cu/In interfacial adhesion. The effects (i)
and (ii) proffer maintaining the high compliance whereas (iii)
and (iv) improve thermal and electrical conductivities. Thus,
the proposed 2-step fabrication methodology comprising of
LPS followed by ARB is proffers actual realization of the ideal
microstructure of the TIM as shown in Fig. 1.

r
Fo

In channels become continuous after several passes of ARB,

the requirement of high volume fraction of In for ensuring the
same in the as-sintered sample can be relaxed if, instead of
LPS only processing technique, the proposed 2-step processing
technique is used. An increase in the high conductivity phase,
especially at the expense of the low conductivity phase, will
lead to relatively higher thermal conductivity.
D. Effect of rGO on Thermal Conductivity
As mentioned earlier, preliminary study was conducted to
evaluate the effectiveness of ARB processing in encapsulating
rGO, a foreign desired species, into the as-sintered samples. In
this preliminary study, 0.6 mg of rGO flakes was placed in
between the mating surfaces (Fig. 2) before the first pass of
the ARB processing. Figure 14 shows a representative SEM
micrograph of In- 40 vol. % Cu composite solder with rGO
processed through 2 passes of ARB. The volume fraction of
rGO in the composite after two ARB passes was ~18.5 % and
as shown in Fig. 14, the average thickness of each rGO layer
was ~115 m whereas the total thickness of the sample
remained equal to 1.2 mm. Fig. 14 clearly indicates that highly
conductive materials, such as graphene, carbon nanotubes, 2-D
or 3-D diamond, etc., can easily be embedded in In-Cu
composite solders through ARB processing of the as-sintered
sample. The number of such inter-layers will increase with
ARB passes. It should be noted that the total number of rGO
layers may determinately affect the conductivity along the
direction perpendicular to the layers as it would increase the
total interfacial area/volume in the composite. However, in this
study, the electrical conductivity was measured along the
length of the rGO layer and also the thicknesses of all layers of
the sample with rGO (>100 m) were too large to have
significant effect on the electrical conductivity, the total
number of rGO layers may not significantly affect
conductivity.
Electrical conductivity of the In-Cu composite with rGO
layers after two consecutive passes of ARB was equal to ~6.9
107 /m whereas, as mentioned previously, the samples
processed to two passes of ARB without rGO layer showed an
electrical conductivity of only ~1.4 107 /m. Thus, the
electrical conductivity of the monolithic In-Cu composite
solders along a particular direction can be dramatically
enhanced through the inclusion of a few layers of highly
conductive foreign phase by the proposed 2-step processing
technique.
Since phonons are often considered to be the majority
carrier of heat in graphene and graphene oxides [15-17],
Wiedemann-Franz law cannot be applied to predict the thermal
conductivity of the samples with rGO layers. Thus, there is a
need for direct measurement of thermal conductivity in these
samples, such as through laser flash analysis, etc. However,
since the electrical and thermal conductivities of rGO
simultaneously either increase or decrease due to a thermomechanical treatment (such as thermal treatment at high
temperature [17], encapsulation in polymer composites [18],
etc.), it can be speculated that a dramatic increase in the
electrical conductivity, as observed in this study, is an
indication of a simultaneous increase in the thermal
conductivity of the samples with rGO layers.

er

Pe

(b)

Fig. 13. Schematics illustrating effects of ARB on microstructure of In-Cu

composite solder: Configurations in (a) as-sintered condition, and (b) after
several ARB passes. Black solid region and black annulus around Cu
represent pores and poor interfacial adhesion between Cu and In,
respectively.

It should be noted that ARB processing drastically affects

the microstructure and hence the mechanical, electrical and
thermal properties of In-Cu solder composites. Thus, the
optimum composition of as-sintered In-Cu solders may shift
from the previously reported 40 volume % of Cu [3]. Since the

ew

(a)

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 8 of 16
8

Submitted to IEEE Transactions on Components, Packaging and Manufacturing Technology

REFERENCES
[1]

[2]

[3]

(a)
(b)
Fig. 14. A representative micrograph showing embedding of reduced
graphene oxide (rGO) in In-Cu composites through ARB. The black
elongated region in the center of the micrograph shows one of the two rGO
layers.

r
Fo

[4]

[5]

IV. CONCLUSION

Short term liquid phase sintering was conducted to prepare

In-40 vol. % Cu composite solders. Cu and In are the HTCP
and HMCP, respectively. The produced microstructure
consisted of several isolated agglomerates of Cu powders and
closed pores. Although increase in the sintering period did not
significantly affect densification, the distribution of Cu in In
was slightly improved with an increase in the sintering period
from 30 seconds to 60 seconds.
ARB did not cause a monotonous increase in the density of
the sample; however, the distribution of Cu in In became
uniform with ARB passes. ARB processing actually led to
formation of continuous channels of In around Cu particles
which was never observed in as-sintered sample. Processing
through ARB disintegrated Cu agglomerates and led to a
decrease in the Cu particle size as well as uniformaization of
the particle sizes.
As-sintered In-Cu composite solders showed significant
strain rate sensitivity. Although the mechanical behavior of InCu composite solders, as revealed through stress-strain plots,
fundamentally did not change with ARB processing,
subsequent ARB processing did not result in significant
increase the flow stresses.
Electrical and thermal conductivities of In-Cu composite
solders increased with number of ARB passes. Thermal
conductivity of the samples can be increased to the theoretical
limit by ARB processing while only moderately decreasing the
high compliance of the as-sintered material.
ARB processing can be used for embedding a desired
foreign species, such as rGO, along a certain direction in In-Cu
composite solders. Electrical conductivity of In-Cu samples
increased by a factor of five upon inclusion of two thin layers
of rGO through ARB.
These results on mechanical response, thermal and electrical
conductivities and the flexibility of embedding a desired phase
for improving a target property confirm that combining LPS
with ARB is a promising route to meet the requirements of a
TIM as well as a package level interconnects in advanced
microelectronic packages.

[6]

[7]

[8]

er

Pe

[9]

[10]

P. Kumar, I. Dutta, R. Raj, M. Renavikar, and V. Wakharkar, Novel

liquid phase sintered solders with indium as minority phase for next
generation thermal interface materials applications Proc. Conf. on
Thermal Issues in Emerging Technologies (ThETA 2), Cairo, Egypt,
IEEE, 17-20 Dec.(2008) pp. 339-346.
I. Dutta, R. Raj, P. Kumar, T. Chen, C. M. Nagaraj, J. Liu, M.
Renavikar and V. Wakharkar, Liquid phase sintered solders with
indium as minority phase for next generation thermal interface material
applications J Electron Mater Vol. 38, No. 12, pp. 2735-2745,
August 2009.
J. Liu, P. Kumar, I. Dutta, R. Raj, M. Renavikar and V. Wakharkar,
Liquid phase sintered composite solders for next generation thermal
interface material and interconnect applications J Mater Sci Vol. 46,
No. 21, pp. 7012-7025 June 2009.
P. Kumar and S. Awasthi, Mechanical and thermal modeling of in-cu
composites for thermal interface materials applications J Comp
Mater. Vol. 48, No. 11, pp. 1391-1398, May 2014.
J. Liu, P. Rottmann, S. Dutta, P. Kumar, R. Raj, M. Renavikar and I.
Dutta, Next generation materials for thermal interface and high
density energy storage applications via liquid phase sintering
Proceeding of the 12th Electronics Packaging Technology Conference
(EPTC) (IEEE, Singapore) (2009) pp. 506-511.
J. Liu, U. Sahayam, I. Dutta, R. Raj, M. Renavikar, R. S. Sidhu and R.
Mahajan, Interfacially engineered liquid-phase-sintered composite
solders for thermal interface material applications, J Mater Sci, Vol.
49, No. 22, pp. 7844-7584, August 2014.
Y. Saito, H. Utsunomiya, N. Tsuji and T. Sakai, Novel ultra-high
straining process for bulk materialsdevelopment of the accumulative
roll-bonding (ARB) process, Acta Mater, Vol. 47, No. 2, pp. 579
583, January1999.
I. K. Moon, J. Lee, R. S. Ruoff and H. Lee, Reduced graphene oxide
by chemical graphitization, Nature Comm, Vol. 1, Article No. 73,
September 2010.
J. C. Maxwell-Garnett, Colours in metal glasses and in metallic films
Phil. Trans Royal Soc (London), Vol. 203, pp 385420, January 1904.
Solder Thermal Interphase Materials. Available online:
http://www.indium.com/TIM/solutions/index.php.
D. G. Kim, J. W. Yoon, C. Y. Lee and S. B. Jung, Reaction diffusion
and formation of Cu11In9 and In27Ni10 phases in the couple of indiumsubstrates Mater Trans Vol. 44, No. 1, pp. 72-77, January 2003.
S. L. Yu, S. S. Wang and T. H. Chuang, Intermetallic compounds
formed at the interface between liquid indium and copper substrates J
Electron Mater Vol. 31, No. 5, pp. 488-493, May 2002.
H. S. Liu, X. J. Liu, C. P. Wang, I. Ohnuma, R. Kainuma, Z. P. Jin and
K. Ishida, Thermodynamic assessment of the Cu-In binary system J
Phase Equilib Vol. 23, No. 5, pp. 409-415, October 2002.
N. Tsuji, Y. Saito, H.-S. Lee, and Y. Minamino, ARB (accumulative
roll-bonding) and other new techniques to produce bulk ultrafine
grained materials, Adv. Engineering Mater, Vol. 5, No. 5, pp. 338344, 2003.
Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts and R. S.
Ruoff, Graphene and graphene oxide: Synthesis, properties, and
applications, Adv. Mater., Vol. 22,No. 35, pp. 39063924, 2010.
N. K. Mahanta and A. R. Abramson, Thermal conductivity of
graphene and graphene oxide nanoplatelets, 13th IEEE Intersociety
Conference on
Thermal and Thermomechanical Phenomena in
Electronic Systems (ITherm), San Diego (US), May 30-June 1 2012,
pp. 1-6
T. Schwamb, B. R. Burg, N. C. Schirmer and D. Poulikakos, An
electrical method for the measurement of thermal and electrical
conductivity of reduced graphene oxide nanostructures, Nanotechnol,
Vol. 20, No. 40, pp. 405704 1-5, 2009.
W. Park, J. Hu, L. A. Jauregui, X. Ruan and Y. P. Chen, Electrical
and thermal conductivities of reduced graphene oxide/polystyrene
composites, Appl. Phys. Lett., Vol. 104, No. 11, pp. 113101-113104,
2014.

[11]

[12]

[13]

[15]

[16]

[17]

[18]

ew

[14]

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 9 of 16

Review of TCPMT-2015-036:
We thank the reviewers for patiently going over the manuscript and providing us
extremely helpful comments on our paper. Our responses are given below in bold type and
the changes in the manuscript are marked using yellow highlight.

Reviewer: 1
1. The value of ARB in improving electrical or thermal properties has been well-established
through this work. However, the flow stress was shown to increase with ARB. How do the
authors claim that ARB led to high compliance.

r
Fo

The reviewer is correct in pointing out that ARB led to an increase in the compliance of
Cu-In samples. However, as shown in Fig. 11, such an increase was quite moderate as
compared to the effects of other parameters, such as strain rate (Fig. 9) and sintering
times (Fig. 10). Nevertheless, we understand the concern raised by the reviewer in

Pe

stating ARB led to high compliance. Hence, ..while keeping high mechanical
compliance. in the conclusion section has now been replaced by .while only

er

moderately decreasing the high compliance of the as-sintered material. in the page 8 of
the revised manuscript.

Re

2. Closed pores can be sintered if they meet certain energy criterion. It is not accurate to say

vi

only open pores can be sintered. Can the authors justify or revise?
The reviewer has correctly pointed out that closed or isolated pores can be sintered.

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

However, it often occurs if the pores are attached to grain boundaries or other fast
diffusion paths. Hence, being a diffusion controlled phenomenon, sintering of closed
pores is a slow process and it may not readily occur under the flash sintering conditions
employed in this work. We have now inserted this clarification in page 3 of the revised
manuscript.

3. How are samples prepared with different green densities? The experimental part does not
show how different green densities are achieved.
We thank the reviewer for pointing out this lack of information. The samples used to
study the effect of green density on the sintered density, as shown in Fig. 4a, were
1

Page 10 of 16

fabricated using different compaction pressure while keeping the weight of powder
mixtures, time of compaction and sintering conditions the same. This is now mentioned
in page 2 of the revised manuscript.

4. Conclusions section: It has some repetition in the second para (ARB led to uniform
distribution) and can be condensed to make it read better. It talks about electrical and
thermal, then talks about mechanical and again talks about electrical/thermal. So, it goes
back and forth a bit. This can be edited for better readability.
We thank the reviewer for mentioning the repetition. The repeating lines have been

r
Fo

removed from the conclusion section of the revised manuscript.

5. The reason for mechanical characterization is not discussed in the Introduction. It does not
set the stage for why the authors spent so much time doing mechanical testing with different

Pe

strain rates, identifying the behavior of flow stress as a function of strain rate and ARB
passes. Is the mechanical characterization just to show that ARB improves softness?

er

Mechanical characterization is an important aspect of any study on TIM. This is so as

an ideal TIM should be highly compliant. Thus, a processing technique which although

Re

increases the thermal conductivity (and electrical conductivity), but significantly

decreases the compliance of the material cannot be suitable for processing TIM.

vi

Therefore, study of effect of ARB on the mechanical response of Cu-In sample is critical
for this study. This is now mentioned in page 3 of the revised manuscript.

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

6. A schematic (cartoon drawing) of the composite microstructure of as processed, LPS- and

ARB-processed samples, along with how that leads to different thermal properties and flow
stress is also recommended to be included. That will convey the key message well and fast.
We thank the reviewer for the excellent suggestion. We have now inserted a schematic
(Fig. 13) and associated explanatory text in page 7 of the revised manuscript.
Accordingly, the figure numbers and their citations in the revised manuscript have
been changed. These changes in figure numbers and their citations in the text are not
highlighted in yellow.

Page 11 of 16

7. What is the thermal conductivity of rGO? Is the thermal conductivity reported along the rGO
plane? What is the volume fraction and estimated K with rGO?
As mentioned by the reviewer # 3, the thermal conductivity of rGO cannot be estimated
using Wiedemann-Franz law as the majority carrier of heat in graphene is phonons and
not electrons. Thus, the mention of thermal conductivity of samples with rGO has now
been replaced by the electrical conductivity the quantity which was directly measured
in this study.
The method of measurement of electrical conductivity and a discussion on its
implications on thermal conductivity are now mentioned in the caption of Fig. 3 and

r
Fo

page 7 of the revised manuscript.

8. Can you show the Cu-In interface to establish that the intermetallic thickness is minimal?
EPMA analysis was conducted to evaluate the extent of IMC formation as a result of

Pe

ARB processing. The EPMA analysis confirmed formation of IMC layer of

dramatically smaller thickness than those previously reported in Cu-In samples without

er

any reaction barrier layer. A new figure (Fig. 7) representatively showing the atomic
distribution of Cu and In in an ARB processed sample and the related explanatory text

Re

have been inserted in page 5 of the revised manuscript. Accordingly, the figure
numbers and their citations in the revised manuscript have been changed. These

vi

changes in figure numbers and their citations are not highlighted.

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

9. It will be more useful if the SEM pictures are labeled so that the audience dont have to read
the detailed figure captions;

We thank the reviewer for the helpful comment. We have marked different phases in
SEM images wherever possible without compromising with the clarity of images (i.e.,
all components of Figs 5 and 6 except 5a). These changes are not highlighted in yellow.

Page 12 of 16

Reviewer: 2
Major comments:
1. The value of thermal conductivity reported in this work seem to be an overestimation. This
may be due to the fact that only electrical conductivity has been experimentally measured.
Thermally conductivity has been estimated using WiedeMann-Franz law. The assumption
that metallic composite abides by the Wiedemann-Franz Law and the corresponding Lorentz
number will be equal to that obtained by fitting a curve for electrical vs thermal conductivity
of various metals only needs to be justified either through experimental measurement of
thermal conductivity or through literature supporting the estimation methodology for
mixtures.

r
Fo

The reviewer has raised an excellent point related to the validity of the Lorentz
number(s) for In-Cu composite solder. However, in a previously reported study on assintered In-Cu samples, we simultaneously measured the thermal conductivities using

Pe

laser flash apparatus and through electrical conductivity using Wiedemann-Franz law
(as used in this study) and observed a little difference (<10 %) in these two values of

er

thermal conductivities. We have highlighted the above in page 3 and cited appropriate
references in the revised manuscript.

Re

2. The authors need to provide references for the Lorenz number values for In and Cu and so on

vi

since it varies in literature.

We prepared the master-plot (i.e., Fig. 12a) using the reported values of thermal and

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

electrical conductivities of various pure metals. Thus, we have not used the respective
Lorentz numbers for the master plot (Fig. 12a). We have now mentioned it clearly in
the caption of Figure 12a in page 6 of the revised manuscript. Also, the methodology for
calculating thermal conductivity using electrical conductivity is now reworded in page 3
for clarity.

3. Will the thermal conductivity not depend on the thickness of the different layers (In-Cu as
well as rGO) in the sandwich structure considering it is not a homogeneous mixture? Please
clarify.

Page 13 of 16

The reviewer has raised an excellent point here. Indeed, the thermal conductivity of a
composite structure is a strong function of the volume fraction of different constituents.
The thickness of individual layers will affect the thermal conductivity in the direction
perpendicular to the rGO layer (as there the interfacial layers are configured in series).
However, the reported conductivity in this manuscript is along the length of the rGO
layer and the thickness of all layers are very large (>100 m) to have any effect due to
scattering from the interfaces. Hence it will be less affected by the thickness of the layer
or number of layers. A mention of the above and the dimensions of Cu-In and rGO are
now given in page 7 of the revised manuscript.

Minor Comments:

r
Fo

1. Please avoid using very long sentences like a few in the abstract.
We thank the reviewer for the suggestion of keeping the sentences short. We have kept

Pe

it in mind and have put significant efforts in improving the language of the revised
manuscript, including the abstract. We have not highlighted these changes.

er

2. Electrical conductivity is plotted in Fig 11 while the text says resistivity.

Re

We have changed all instances of resistivity to conductivity.

vi

3. Many symbols in the Figures and the legends do not show up (color scheme) when printed.
We thank the reviewer for the raising this issue. We have now ensured that all symbols,

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

text and pictures are discernable when printed with grey scale or in color.

4. Please define green density at the instant of first use.

We have now defined green density in page 2 of the revised manuscript.

Page 14 of 16

Reviewer: 3
This paper describes the development of a processing technique for the creation of copperindium composites for use as thermal (and electrical) interfaces. The paper builds on prior work
associated with liquid phase sintering, with the new addition of accumulative roll bonding. The
results include imaging, mechanical, and electrical results, with thermal conductivity deduced
indirectly from the latter. The paper is well written, informative, and useful to the community.
However, the paper contains a major problem and should not be published until it is addressed.

Major comments:

r
Fo

1. The discussion of the addition of rGO flakes in Section II is very cavalier, without any clarity
about the amount of material added. This issue is discussed briefly in section III.D, but
without a clear statement concerning the amount of rGO added. Nevertheless, the same
section includes a brief yet astounding result thermal conductivity of 467 W/mK with rGO

Pe

a 5-fold increase! One would expect much more discussion and data substantiation for
such a result, yet none exists.

er

We thank the reviewer for raising the issue of limited information and discussion on
rGO. We have now added vital information about the weight, thickness and volume

Re

fraction of rGO in page 7 of the revised manuscript.

The result on dramatic increase in the electrical conductivity is from our preliminary

vi

work where we wanted to show the ability of the proposed 2-step processing technique
in embedding desired foreign species, such as rGO, etc., in the otherwise monolithic

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

composite solder. The demonstration of this ability of the proposed fabrication method
has been the core focus of this manuscript. Thus, discussion on various results has been
scantier. We are currently conducting more systematic work on effect of rGO on
microstructure, mechanical behavior and conductivities, and they will be reported
later. Nevertheless, we have inserted some more discussion on these results in section
III.D.

2. The veracity of this result is highly doubtable, as all thermal conductivity data in the paper
derive from electrical measurements and the Wiedemann-Franz law. However, the paper fails
to mention that the Wiedemann-Franz law does NOT apply to graphene (and similar
6

Page 15 of 16

materials). Consequently, the entirety of this single result (only a number given in the last
section) does not contain the rigor expected for scholarly publication.
We thank the reviewer for highlighting the limitation of Wiedemann-Franz law for
graphene as phonons are the majority carrier of heat in graphene. However, several
reports on rGO have shown a simultaneous increase in the thermal and the electrical
conductivities after certain processing or treatment (say, exposure to high temperature,
fabrication of composites with different volume fraction of rGO, etc.). Thus, we can
infer that an increase in the electrical conductivity upon addition of rGO will lead to a
simultaneous increase in the thermal conductivity also. We have mended the mistake in

r
Fo

reporting the probable number for thermal conductivity of the sample with rGO (using
Wiedemann-Franz law) and highlighted the need for direct measurement of thermal
conductivity in these cases in page 7 of the revised manuscript.

Minor comments:

Pe

1. General, use of the symbol K for thermal conductivity is a bit unusual. More common

er

forms are k (lowercase), Greek kappa, or Greek lambda.

We have changed the symbol for thermal conductivity to (Greek kappa) in the revised
manuscript.

vi

Re

2. Section II, top. The choice of a fixed 40% Cu volume fraction is based on LPS only and the
resultant compliance from prior work. However, one would expect (and the later results

ew

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

prove) that the ARB process substantially changes mechanical behavior. Therefore, this
optimum is questionable, and this issue should be discussed.

We thank the reviewer for highlighting this interesting issue. We have discussed it in
page 7 of the revised manuscript.

3. Page 2, left column. This is a very long paragraph that could be broken thematically (e.g., at
line 49).
We thank reviewer for the suggestion. The paragraph has been broken into two
paragraphs in the revised manuscript. Following this advice, we did similar breaking of

Page 16 of 16

a few other paragraphs also in the revised manicurist. These are not highlighted in
yellow.

4. Section III.B. Results from rGO samples are curiously missing in this section.
As mentioned previously, the work on rGO samples is in the preliminary stage. We are
in process of conducting full set of tests including mechanical testing, detailed ARB
processing and microstructural analysis, etc. Unfortunately, we are still not in a
position to comment on the effect of rGO on the mechanical behavior and it will be
reported later.

r
Fo

5. Section III.C. An equation form of the Maxwell-Garnett theory should be included for
completeness.

We have inserted the equation form of Maxwell-Garnett theory, as Eq. (3), in page 7 of
the revised manuscript.

er

Pe

ew

vi

Re

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60