Aluminum matrix nanocomposites for applications in

automobile industry
Binghua Ma

To cite this version:

Binghua Ma. Aluminum matrix nanocomposites for applications in automobile industry. Material
chemistry. Sorbonne Université, 2021. English. �NNT : 2021SORUS110�. �tel-03680513�

HAL Id: tel-03680513

https://theses.hal.science/tel-03680513
Submitted on 28 May 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Sorbonne Université
Ecole doctorale 397 - Physique et Chimie des Matériaux
Laboratoire de Chimie de la Matière Condensée de Paris

NANOPARTICULES DANS DES MATRICES

METALLIQUES POUR DES APPLICATIONS EN
AUTOMOBILES

Aluminum matrix nanocomposites for applications in

automobile industry

Par Binghua MA

Thèse de doctorat de Physique et Chimie Matériaux

Présentée et soutenue publiquement le 27 Mai 2021

Devant un jury composé de :

M. Dirrenberger Justin Maître de conférence Rapporteur

(PIMM, CNAM)
Mme. Tréguer-Delapierre Mona Professeure Rapportrice
(ICMCB, Université de Bordeaux)
Mme. Courty Alexa Professeure Examinatrice
(Monaris, Sorbonne Université)
M. Mazerolles Léo Directeur de recherche CNRS Examinateur
(ICMPE)
M. Portehault David Chargé de recherche CNRS Directeur de thèse
(LCMCP, Sorbonne Université)
M. Delalande Stéphane Ingénieur de recherche Co-directeur de thèse
(Stellantis)
M. Sanchez Clément Professeur émérite Co-directeur de thèse
(LCMCP, Collège de France)
Table of contents
General introduction ................................................................................................................... 3
Chapter 1 .................................................................................................................................... 7
1.1 Context ............................................................................................................................. 9
1.1.1 Metal matrix composites ........................................................................................... 9
1.1.2 How can aluminum matrix nanocomposites improve sustainability in the automotive
industry? ........................................................................................................................... 11
1.2 Mechanical property improvement ................................................................................ 12
1.2.1 Strengthening mechanisms ...................................................................................... 12
1.2.2 Strategies to improve mechanical properties........................................................... 15
1.3 Phase change materials for thermal management of Li-ion battery ............................... 22
1.3.1 Importance of thermal management of the Li-ion battery ...................................... 22
1.3.2 Cooling strategies .................................................................................................... 25
1.3.3 Metal matrix nanocomposite with phase change materials for temperature ........... 26
1.4 MMCs fabrication techniques ........................................................................................ 27
1.4.1 Powder metallurgy .................................................................................................. 28
1.4.2 Liquid state process ................................................................................................. 31
1.4.3 Squeeze casting ....................................................................................................... 35
1.4.4 Other techniques ...................................................................................................... 36
1.5 Conclusion and motivation of this work ........................................................................ 37
References ............................................................................................................................ 39
Chapter 2 .................................................................................................................................. 47
2.1 Introduction .................................................................................................................... 49
2.1.1 State of the art on the Bi synthesis .......................................................................... 49
2.1.2 State of the art on the growth of silica shell to encapsulate bismuth particles ........ 54
2.2 Synthesis and characterization of Bi nanoparticles ........................................................ 54
2.3 Synthesis and characterization of Bi@SiO2 nanoparticles............................................. 56
2.4 Conclusion ...................................................................................................................... 61
Chapter 3 .................................................................................................................................. 63
3.1 Introduction .................................................................................................................... 65
3.2 Materials and methods ................................................................................................... 66
3.3 Results and discussion .................................................................................................... 69
3.4 Conclusion ...................................................................................................................... 92
References ............................................................................................................................ 93
Chapter 4 .................................................................................................................................. 95
4.1 Introduction .................................................................................................................... 97
4.2 Experimental section ...................................................................................................... 99
4.3 Results and discussion .................................................................................................. 102
4.3.1 Synthesis and characterization of Bi and Bi@SiO2 nanoparticles ........................ 102
4.3.2 Synthesis and characterization of Bi@SiO2-Al nanocomposite ........................... 103
4.3.3 Synthesis and characterization of Bi-Al nanocomposite ....................................... 110
4.3.2 Heat absorption ability evaluated by DSC measurements .................................... 111
4.4 Conclusions .................................................................................................................. 114
References .......................................................................................................................... 115
General conclusions and perspectives .................................................................................... 117
Appendix ................................................................................................................................ 123
Résumé en Français ................................................................................................................ 143
i
ii
General introduction

3
4
 In the last 20 years, the key importance of energy and environmental challenges has
paved its way to vehicle consumers. They are becoming more and more aware of environmental
issues and are looking for cleaner vehicles. In parallel with the customer's desire, accelerated
regulatory constraints on CO2 emissions and pollutants (such as CO, NO2) for new vehicles
came into force in order to reduce greenhouse gas emissions. In this context, light-weight and
electrical vehicles-related materials are one of the most important trends for the development
of materials in the transportation sector.

This PhD work aimed at exploring novel metal matrix nanocomposites fabricated by
ultrasound-assisted casting techniques to address automotive-related issues: mechanical
enhancement and thermal management, with specific emphasis on the processing parameters
for the dispersion ability and the interaction between nanosized-inclusion and metal matrix.

Aluminum and its alloys have been largely used in many industrial fields because of
their excellent properties, such as lightweight, low cost, good thermal properties and high
ductility. But the poor stiffness, strength, and wear properties still limit their applications as
vehicle materials. Metal matrix composites are a class of metallic materials containing
reinforcements and that can be traced back to the 1970s. Nanoparticles could be applied in a
similar manner to enhance the mechanical properties of aluminum-based matrices by the design
of metal matrix nanocomposites, which will be discussed in chapter 1. In the same chapter, the
mechanisms behind the improved properties of nanocomposites will also be described. Then
the techniques used to fabricate metal matrix nanocomposites will be further discussed.

Chapter 2 will present the synthesis and characterization of building blocks for the
preparation of nanocomposites for thermal management: core-shell Bi@SiO2 nanoparticles
obtained by sol-gel synthesis. Our efforts for decreasing the aggregation state of Bi
nanoparticles in silica shell will be presented, including ligand exchange at the surface of Bi
nanoparticles. These particles will be used as building blocks of composites in chapter 4.

The incorporation of stiffer and harder reinforcement materials leads to aluminum metal
matrix composites combining the properties of both the matrix and the reinforcement phase.
Considerable efforts have been deployed to incorporate commercial nanoparticles into metal

5
matrices. Many recent chemical synthesis methods have widened the range of accessible
nanoparticles, especially for boron-based phases, which are interesting for vehicle applications
due to their light weight and hardness. Their application as metal matrix inclusions is studied
and will be discussed in chapter 3. This chapter will present the fabrication and characterization
of HfB2-Al and SiC-Al nanocomposites. It will present in detail the original set-up we have
designed, based on ultrasound-assisted casting, in order to increase the dispersion of
nanoparticles in aluminum-based matrices. In the same chapter, the microstructure evolution
with different experimental parameters is determined and the mechanical properties of these
composites are evaluated by microindentation and nanoindentation tests.

While electrical vehicles are promising systems and have increased market share since
several years, the research on the thermal management of battery packs is still in need. Among
classic cooling methods, phase change material is a passive strategy relying on heat absorption
through melting. Several recent developments focused on metal matrix nanocomposites with
phase change materials as inclusions, which are of great interest due to their high thermal
conductivity, high mechanical properties and wide working temperature range. Conventional
strategies including powder metallurgy and bottom-up fabrication require high fabrication
costs. At the operational level, we investigated the liquid phase method: metallic nanoparticles
were encapsulated in a shell of SiO2 to prevent destruction of the particles and macroscopic
phase separation. Subsequently, these nanoparticles are incorporated in the molten metal matrix
by the ultrasound-assisted process developed in chapter 3. This whole part will be presented in
chapter 4, which encompasses the synthesis of phase change nanocomposites, the study of the
reactivity occurring at the nanoparticles-matrix interface, and the assessment of the heat
absorption ability of the resulting composites.

Overall, this work opens new perspectives for the application of ultrasound-assisted
casting to the fabrication of mechanically reinforced and phase change metallic materials that
could be potentially applied to fabricate vehicle components. We have also identified specific
difficulties in reaching high dispersion of boron-based nanoparticles in aluminum matrices,
thus identified future tracks to explore in order to achieve metal matrix nanocomposites with
high homogeneity at the nanoscale for a wide range of reinforcement nanomaterials.

6
 Chapter 1

Bibliography and state-of-the-art

7
8
 This first chapter focuses on bibliographic aspects. First, the notion of metal matrix
nanocomposites and their actual/potential applications in the automotive industry are
presented. Then, the interest shifts toward two potential applications, mechanical
reinforcement and heat management, with overviews of the existing techniques and solutions.
Finally, the fabrication methods for nanocomposites are briefly presented.

1.1 Context

Composite materials consist of two or more constituents with different physical and/or
chemical features. They can exhibit properties superior to each of the components and to their
sum, thanks to so-called synergistic effects.1 The major constituent is regarded as the matrix
phase, and the remaining is called the reinforcement phase(s).2 Often, the reinforcement phase
is stronger, harder and stiffer than matrix phase, enabling improvement of the properties of the
matrix. Commonly, composite materials are divided into three main types according to the
matrix phase: polymer matrix composites,3,4 metal matrix composites (MMCs)5,6 and ceramic
matrix composites.7,8 Especially, MMCs have attracted lots of attention due to their superior
properties and wide applications range from aerospace, to automobile industry and to defense.

1.1.1 Metal matrix composites

Metal matrix composites (MMCs) are a class of metallic materials containing reinforcement
phase(s) that can be traced back to the 1970s. Researchers managed to improve properties of
metals/alloys with ceramic reinforcements by stir casting, rheocasting and powder
metallurgy.9,10 For MMCs, three main types of reinforcement materials can be distinguished:
continuous fiber networks and discontinuous reinforcement materials as particles or whiskers11
(Figure 1.1). Compared to continuous reinforcement, the discontinuous one presents
advantages of lower cost and isotropic properties. Generally, there are two main types of
reinforcement phases used for MMCs: (a) ceramic reinforcements, such as Al2O3,12,13 SiC,14,15
TiC,16 B4C,17 and (b) carbon-based materials like graphite, fly ash, etc. To fabricate MMCs,
aluminum,2,18 magnesium,19,20 titanium21 and copper22 are widely used as matrix phases.

9
 Figure. 1.1 Three types of reinforcement in MMCs11

As mentioned above, reinforcement with metallic or ceramic phases can improve

hardness of the composite versus the sole matrix, enabling the material to withstand higher load
but also providing additional specific properties. The incorporated reinforcement objects could
be microsized or nanosized. Micron-scale reinforcement in the metal matrix has been well
studied and demonstrated to provide high hardness, elastic modulus, and strength.23 Reducing
the size of the inclusions to the nanoscale brings additional advantages: reduced load (5 to 20%
for macroscale reinforcements vs. lower than 5% for nanoscale ones) and increased strength
by Orowan hardening and pinning of grain boundaries.24 Hence, metal matrix nanocomposites
(MMNCs) possess in general properties superior to those of composites with microscale
reinforcement. Nanoscaled reinforcement materials for MMNCs include graphene, carbon
nanotubes and nanofibers, and nanoparticles (such as Al2O3, SiO2, SiC, TiO2).5,6

10
1.1.2 How can aluminum matrix nanocomposites improve sustainability in the
automotive industry?

A new form of mobility represented by the ACER concept (Autonomous, Connectivity,

Electrification and Ridesharing) has shifted dramatically the market in the past few years.
Based on the latest legislation and current trends dealing with clean vehicles and carbon
neutrality, automotive industries have integrated sustainability into their strategy. In this
context, the research of light-weight, tough, and functional materials has a growing significance.

The application of nanocomposites for the design of automotive components reflects

the search for improved production rate and thermal stability, weight reduction, and less wear
parts. All these features contribute to a reduction of CO2 emissions and of environmental
impact. Most of them also enable an increase of the cruising time, which is of great importance
in the case of electric vehicles.25

Light (density < 2.7 g·cm-3) and tough (tensile strength > 350 MPa) aluminum MMNCs
are especially interesting for automobile applications. Aluminum and its alloys are already
largely used in many industrial fields because of their light weight, low cost, good thermal
stability and high ductility. But their poor stiffness, strength, and wear properties still limit their
applications.2 These drawbacks can be circumvented by the design of aluminum metal matrix
composites with the addition of reinforcement materials stiffer and harder than aluminum and
its alloys.

Apart from traditional mechanical reinforcement, metal matrix nanocomposites can

also acquire properties specific to the inclusion nanoparticles. A typical application deals with
thermal management of batteries.26 The combination of phase change inclusion particles and
aluminum matrix leads to bifunctional materials with heat management properties and good
thermal dissipation.27 These properties will be detailed in the next chapters.

11
1.2 Mechanical property improvement

Enhanced mechanical properties is one of the main pursuits for MMNCs fabrications. Below,
we present possible strengthening mechanisms and strategies to improve mechanical
properties.

1.2.1 Strengthening mechanisms

Most of the work on metal matrix nanocomposites focuses on the improvement of mechanical
properties: hardness, rigidity and wear resistance.23 As the major component, the metallic
matrix provides ductility. Its properties depend on the polycrystallinity of its microstructure.
Indeed, grain boundaries strengthen the material by acting as barriers to the dislocation
movement.

The enlisted strengthening mechanisms for MMNCs encompass load transfer effect,
Hall-Petch strengthening, Orowan strengthening, coefficient of thermal expansion (CTE) and
elastic modulus (EM) mismatch.24

Load transfer effect corresponds to direct strengthening by transferring of the load from
the matrix to the reinforcement nano-objects. The reinforcement material is harder than the
matrix, thus contributing to strengthening of the overall material. Nardone and Prewo28
proposed a modified Shear lag model to explain the effect of load transfer on mechanical
properties. Since the volume fraction of the reinforcement particles is small, load transfer is
small as well. For isotropic particles, the model can be simplified to yield the equation29 below:

1
∆s𝐿𝑇 = 𝑣𝑃 𝑠𝑚 (1)
2

Where 𝑣𝑃 and 𝑠𝑚 are the particle volume fraction and the yield strength of the un-reinforced
matrix, respectively.

Orowan strengthening. This effect is based on the interaction between nanoparticles

and dislocations. It can be ignored for particles with a diameter larger than 1 µm.30 Dislocations
tend to bend around particles and to form dislocation loops with dislocations on the other side

12
of the particles. These dislocation loops are then pinned onto the nanoparticles, which leads to
enhanced strength. At same volume fraction, smaller particles contribute to the higher
enhancement of final strength. This effect can be expressed as the equation31,32 below:

0.13𝑏𝐺𝑚 𝑑𝑝
∆𝑠𝑂𝑅 = 1 ln (2)
1 3
2𝑏
𝑑𝑝 [(2𝑣 ) −1]
𝑝

where 𝑏, 𝐺𝑚 , 𝑑𝑝 and 𝑣𝑝 are the Burgers vector, shear modulus, particle average diameter and
particle volume fraction, respectively. For example, Cao et al.33 applied ultrasound-assisted
casting to fabricate AlN reinforced magnesium alloy AZ91D that exhibited a yield strength of
144 MPa compared to 100 MPa for the nonreinforced alloy. The author ascribed this 44%
increase to the Orowan strengthening and the effect of the particles decreasing (or refining) the
matrix grain size (see below), contributing to 35.8 MPa and 8.2 MPa, respectively.

Hall-Petch strengthening. The improvement of strength is also related to decreased

grain size since grain boundaries hinder dislocation movements.24 The relationship between
grain size (𝑑) and yield strength (∆𝑠𝐻−𝑃 ) is expressed through the Hall-Petch equation:34

𝑘𝑦
∆𝑠𝐻−𝑃 = (3)
√𝑑

where 𝑘𝑦 is the strengthening coefficient. From equation (3), we can clearly see that the yield
strength of nanocomposite is inversely correlated to the grain size. This is true for particle sizes
above ca. 10 nm. Below, the inverse Hall Petch effect prevails, leading to an opposite
dependence versus the grain size.35 Hence, an optimal grain diameter for Hall Petch hardening
is ca. 10 nm.

The addition of particles results in a decrease of the grain size of the matrix through
localization of the particles at the grain boundaries, which retards or stops matrix grain
growth.24 The Zener equation35 shows the relationship between grain size (𝑑𝑚 ), particle volume
fraction 𝑣𝑝 and particle diameter 𝑑𝑝 :

4𝑎𝑑𝑝
𝑑𝑚 = (4)
3𝑣𝑝

13
where 𝑎 is a constant. Increasing the particle volume fraction and decreasing the particle size
result in reduced grain size, thus leading to enhanced yield strength of metal matrix
nanocomposites.

Coefficient of thermal expansion (CTE) and elastic modulus (EM) mismatching.

After solidification of the composite, the temperature keeps decreasing to room temperature.
Due to the difference of the coefficients of CTEs between the particles and the matrix, stresses
are generated at the interface, leading to geometrically necessary dislocations (GND). GND
can also generate from the difference of elastic modulus between particles and matrix during
straining. The GND density due to CTE and elastic modulus can be calculated from the
following equations:35

𝐴?𝑎?𝑇𝑣𝑝
𝜌𝐶𝑇𝐸 = (5)
𝑏𝑑𝑝 (1−𝑣𝑝 )

6𝑣𝑝
𝜌𝐸𝑀 = 𝜀 (6)
𝑝𝑑𝑝 3

where 𝐴 is a geometric constant, ∆𝑎 and ∆𝑇 are the differences of the CTE and of temperature,
respectively, 𝑒 is the bulk strain of the composite.

These models could be used to estimate the effect of different parameters of the
composites. However, some critical factors are needed to make these models valid: a full
density, good adhesion between the reinforcement particles and the matrix, and a uniform
distribution of the particles. This is where the understanding and control of the process appear
critical.

Plenty of materials have been commercially used as reinforcements. Among them, hard
materials have often been chosen as reinforcement phase: covalent hard materials/diamond-
like structure (C, CNx, BN, SiC, AlN); metallic hard materials/metal like structure (TiC, TiN,
AlB2, WC, Ti2N, TiAlCN…) and ionocovalent hard materials/ionic structure (Al2O3,
ZrO2…).36

14
 Table 1.1 Commercially available nanoparticles used for reinforcement37

1.2.2 Strategies to improve mechanical properties

Several strategies have been used to improve the mechanical properties through the
mechanisms detailed above, such as selecting suitable reinforcement materials and optimizing
particle characteristics (such as size distribution, composition and interfacial bonding), which
will be discussed in the following paragraph.

15
1.2.2.1 Selection of suitable reinforcements

Recent studies on MMNCs focus on the light metal matrix. Among metals, the light alloys of
Al, Ti and Mg are desirable as metal matrix. Their properties depend on the reinforcement
particle features, more precisely, their morphology, size, load fraction, as well as the quality of
the interface with the matrix. Table 1.1 above listed some of the commercial nanoparticles for
MMNCs. They can be categorized into several types:

a) Carbon allotropes

Figure 1.2. Strengthening and stiffening efficiencies of reduced graphene oxide in a reduced
graphene oxide (RGO)/Al nanolaminated composite compared to other reinforcement materials
in Al matrix composites38

Carbon nanotubes (CNTs) are among the most studied strengthening additives for MMC.39
Their strong sp2 bonds lead to high elastic stretching properties. For multi-wall carbon
nanotubes, the theoretical elastic modulus can reach 600 and 1100 GPa while the tensile
strength ranges from 35 to 110 GPa.40 Bakshi and Agarwal41 have reviewed the strengthening
properties of CNT-Al composites. The record yield strength and elastic modulus are 621 MPa
and 110 GPa, respectively.41

16
 Other allotropes like nanodiamond and graphene have also been studied for their load-
transfer ability and high thermal conductivity. Li et al.38 fabricated a bulk reduced graphene
oxide-reinforced Al matrix composites with nanolaminated microstructure. Compared with the
unreinforced Al matrix, these composites showed significantly improved stiffness and tensile
strength, with good yield strength (302 MPa) and elastic modulus (87 GPa) (Figure 1.2).

It should be noticed that for all carbon allotropes in Al matrix, a thin layer of Al4C3
could be formed by interfacial reaction, at the expense of the reinforcement objects. It then
degrades the reinforcement objects but promotes load transfer.40

b) Oxides

Oxides like Al2O3, ZrO2, TiO2, ThO2 are often added to metallic matrices to provide ductility
and fracture toughness from the metallic matrix, creep resistance and high strength from the
additives.11

c) Nitrides

Nitrides like nanoparticles of Si3N442 and microparticles of Fe3N43 have been investigated to
improve mechanical properties. Recent work focused on layered boron nitride. BN nanotubes
embedded in an Al matrix possess excellent elastic modulus (750–1200GPa), strength (61 GPa)
and good chemical inertness (oxidation up to 950 °C). They strongly improve mechanical
properties and thermal conductivity.44 These values varied with the processing technique.

BN nanoparticles and microparticles have also been embedded in an Al matrix. With

powder spark plasma sintering (SPS) and ball-milling, Firestein et al.45 produced a BN-Al
composite, with in situ formation of AlB2 and AlN at the interface. Improved tensile strength
of 170 MPa at 500 °C was achieved compared to a value of only 33 MPa for a pure Al sample
produced by spark plasma sintering. A nanocomposite reinforced with large-sized cubic BN
particles has also been fabricated by cold spray and shows improved abrasive wear resistance.46

17
d) Carbides

Carbides like TiC47, SiC,48,49 TaC, WC, B4C50 have been investigated by many researchers.
SiC is one of the most common reinforcements for MMNCs because of its high strength, low
thermal expansion, high thermal conductivity, high hardness, high elastic modulus, high
chemical inertness, and its quite good wettability by the Al matrix. Li et al.51 firstly prepared
well mixed Al powder and 10 vol% SiC nanoparticles via ball milling, then hot pressure
sintering and hot extrusion were applied sequentially to get an Al/SiC nanocomposite that
showed ultrafine Al grains and intragranular SiC nanoparticles. Compared to pure Al alloy, the
nanocomposite exhibited superior mechanical behavior as shown from the stress-strain curve
(Figure 1.3). The yield strength and tensile strength are improved by 284% and 259%,
respectively, compared to the pure Al alloy.

Figure 1.3. Stress-strain curves of a pure Al alloy and a 10 vol% SiC/Al alloy nanocomposite,
(b) Fractograph of Al alloy, (c) Fractograph of the nanocomposite.51
18
e) Borides

Among all the ceramic particles used as reinforcements, titanium diboride (TiB2) has attracted
the interest of many researchers because of its excellent resistance to mechanical erosion,
higher hardness (960 HV), high melting point (2790 °C) and good thermal stability. Besides,
TiB2 has been used as a grain refiner in Al alloys to suppress the growth of Al grains.52

Nanoparticles of TiB2 about 31 nm diameter were used to reinforce Al by cryomilling

and spark plasma sintering (SPS). The strength of the composite was improved by 20%
compared with equivalent SPS-consolidated 5083 Al alloy.52 Nampoothiri et al.53 re-melted a
TiB2/Al composite prepared by powder metallurgy and treated it with ultrasound process. The
sizes of the TiB2 particles decreased after this treatment, yielding a two times increased yield
strength compared to samples before treatment.

AlB254 and ZrB255 were synthesized by in situ reaction with boron and with a Al-
K2ZrF6-KBF4 system. For example, Tian et al.55 introduced pre-dried salts (K2ZrF6 and KBF4)
wrapped in an aluminum foil into molten 2024 Al alloy to promote the in situ reaction in the
molten salt between the different precursors. The formed ZrB2 nanoparticles of ca. 30 to
100 nm were rectangular and were also uniformly distributed in the final composite. ZrB2-
reinforced nanocomposites exhibited higher tensile strength and yield stress.

f) Intermetallic materials

The review of Ward-Close and Doorbar56 summarized the research on intermetallic matrices.
Intermetallic compounds like NiAl and Al3Ti2457 have been successfully used as reinforcement
phases in MMNCs. The Al–Al3Ti nanocomposite revealed good mechanical behavior at high
temperature, and TiAl–NiAl MMNCs showed low fracture toughness and high hardness.

g) Combination of several reinforcement particles

A combination of several reinforcement particles could produce a synergetic effect. For

example, Sujan et al.49 studied the performance of stir casted Al2O3 and SiC reinforced metal
matrix composite material. The result showed that the composite materials exhibit improved
physical and mechanical properties, such as low coefficient of thermal expansion as low as

19
4.6 10-6·N-1·°C-1, high ultimate tensile strength increased by up to 23.68 %, high impact
strength and hardness.

1.2.2.2 Optimized properties of reinforcement

a) Reinforcement uniformity into matrix phase

Non-uniform distribution of reinforcement objects is one of the main obstacles for enhanced
mechanical properties.58 Several techniques have been explored to circumvent this difficulty.
Ultrasound-assisted casting is effective to reach an homogeneous dispersion of the
reinforcement objects in the matrix phase.14,59,60 Another method consists in enhancing the
wettability of reinforcement objects by the matrix. Pourhosseini et al.61 successfully modified
Al2O3 nanoparticles via electroless deposition of Cu, Ni and Co and subsequently applied these
particles to reinforce A356 alloy by stir casting. The wettability of the coated nanoparticles was
enhanced, resulting in homogeneous distribution in the aluminum matrix. The final mechanical
properties (such as microhardness and compressive strength) were then significantly improved,
especially for composites encompassing Ni-coated Al2O3. Another method to reach dispersion
focuses on the ability of the matrix to carry reinforcement objects. Mousavian et al.62 managed
to reinforce an A356 aluminum matrix with SiC ceramic nanoparticles by combining
techniques of stir casting and semi-solid extrusion. The authors found that the addition of nickel
acted provided a SiC carrier, leading to a uniform dispersion of particles into the aluminum
matrix. In view of the mechanical property, the ultimate tensile strength and elongation of the
final Al/SiC composite in presence of nickel could reach around 304 MPa and 5.8%,
respectively, while the introduction of titanium decreased the elongation ratio.

b) Reinforcement particle size

The size of the reinforcement particles plays a critical role in the mechanical properties of
MMCs. For example, a A356 alloy reinforced with nanoscaled TiB2 possessed higher
elongation and tensile strength than with microscaled particles.63 Another example from
Youssef El-Kady et al.64 deals with alumina-reinforced A356 alloy. Rheocasting and squeeze
casting were used to incorporate 60 and 200 nm Al2O3 nanoparticles into A356 alloy. The
nanocomposite with 60 nm Al2O3 particles showed higher ultimate tensile strength, yield
20
strength and elongation compared to the composite with 200 nm Al2O3, while microhardness
showed the opposite trend, probably because of the agglomeration of nanoparticles.

c) Volume fraction of the reinforcement phase

The content of reinforcement phase in the nanocomposite is also an important parameter to

control the mechanical properties.20,63–67 Reddy et al.67 fabricated aluminum metal matrix
nanocomposites reinforced by different contents of SiC nanoparticles of 55 nm diameter (0,
0.3, 0.5, 1.0 and 1.5 vol%) via microwave sintering and hot extrusion techniques. Increasing
the SiC content from 0.3 to 1.5 vol% enhanced the final yield strength from 66 to 114 MPa,
and the ultimate compressive strength from 313 to 392 MPa, respectively. A365/Al2O3 metal
matrix nanocomposites also showed improved ultimate tensile strength, yield strength and
elongation with particle size of 60 nm. However, the excessive incorporation of nanoparticles
could be detrimental to mechanical properties. 64 Indeed, Akbari et al.63 found that increasing
the content of TiB2 nanoparticles into A356 nanocomposites from 0.5 to 1.5 wt% significantly
enhanced the final yield stress and ultimate yield strength, while an opposite trend appeared
when the nanoparticle content kept increasing to 3.0 wt%. This behavior could be ascribed to
increasing particle aggregation.

d) Interfacial bonding

A strong interface between the metal matrix and reinforcement objects is necessary to ensure
effective load transfer.68 Wang et al.69 utilized reduced graphene oxide (RGO) to reinforce a
Cu alloy (in presence of Ti) and found that a TiC layer was formed at the interface of
RGO/CuTi composite. The TiC layer led to enhanced interfacial bonding enabling yield
strength and tensile strength of 242 MPa and 307 MPa, 24 and 16% higher than those of 2 vol%
RGO/Cu composite, respectively. Another example relates to the interface reaction between
44
BN and Al that leads to the formation of AlN and AlB2. Their formation in Al-based
composites is not as deleterious as compared with carbides (Al4C3).70 AlN reinforcement
improves the hardness and flexural strength of Al matrix. AlB2 is also less detrimental to the
fracture toughness of Al.44

21
 As for the previously described parameters, a tradeoff value exists for interfacial
bonding. Indeed, excessive interfacial bonding may deteriorate the mechanical properties of
the composite. Zhang et al.21 applied spark plasma sintering to fabricate Ti/nanodiamond
nanocomposites and found the formation of the new phase TiC derived from the reaction
between Ti and nanodiamond at the grain boundaries. Small amounts of in situ formed TiC
was regarded as one of the main strengthening mechanisms since it is likely to induce
dislocation pinning. However, larger amounts of TiC (≥ 0.5 wt% nanodiamonds) tended to
decrease the mechanical properties of the final nanocomposite.

1.3 Phase change materials for thermal management of Li-ion battery

As mentioned above, MMNCs are potential light-weight materials for thermal control, which
is essential for vehicle electrification based on lithium batteries technology. Below, we are
presenting major strategies for thermal control of batteries together with a brief overview of
phase change materials.

1.3.1 Importance of thermal management of the Li-ion battery

Figure 1.4 Energy volumetric and mass densities of different battery systems.71

22
Electrical vehicles (EVs) including hybrid electric vehicles, plug-in hybrid electric vehicles
and battery electric vehicles focus lots of attention because of their ability to reduce greenhouse
gas emissions.72,73 One of the main challenges to develop EVs lies in the energy storage system.
Lithium-ion batteries possess some advantages compared to other types of batteries, such as
nickel–cadmium and nickel-metal hybrid batteries: a) higher energy density (Figure 1.4), b)
higher specific power, c) lighter weight, d) lower self-discharge rates, e) higher recyclability
and f) longer life cycle.71,74 A full battery pack consists in several connected electrochemical
cells (in series and/or in parallel).71

A traditional cell contains an anode (graphite), a cathode (LiCoO2 layer), a separator

and an electrolyte (Figure 1.5). On charging, the cobaltite is oxidized, leaving Li+, which goes
through the separator and inserts into the interlayer space of graphite, which then is reduced.
The overall process implies an electric current and an electron transfer through the external
circuit.75 This reaction is reversed during discharge. However, the working temperature affects
greatly the kinetics of these reactions, thus the performances, life and safety of Li-ion batteries.
Furthermore, temperature changes of Li-ion battery due to environmental conditions and heat
release from several cycles of charging and discharging are inevitable.73 In Tesla’s Model 3
car, the battery pack could release more energy (in form of heat) than 6000 iPhone handsets.76
Therefore, battery thermal management is necessary. We focus below on management of
overheating.

23
 Figure 1.5 Scheme of a traditional Li-ion battery.77

Heat generation within Li-ion battery is derived from resistive dissipation, relaxation
of the concentration gradient in the cell and chemical reaction.27,73 It can be simply described
as follows:

𝑄 =𝐴−𝐵

𝐴 = 𝐼(𝑈 − 𝑉)

𝑑𝑈
𝐵 = 𝐼(𝑇 )
𝑑𝑇

Q, A, B, I, U and V are the heat generation rate, resistance loss in the cell, reversible entropic
heat of the electrochemical reaction, electric current traveling through the cell, open-circuit
voltage and the cell voltage of the Li-ion batteries, respectively.

24
1.3.2 Cooling strategies

The major role of battery thermal management is to maintain a suitable temperature range and
evenly distributed temperature within battery during charging and discharging, even at extreme
environmental conditions.78 The most effective working temperature of Li-ion battery ranges
from 15 to 40 ℃.79 That means that the battery should on one hand reduce its temperature when
operating temperature increases or environmental temperature is high, on the other hand
increase temperature when the environmental temperature is low. The battery thermal
management system itself should be compact, light, cost-saving, reliable and easy to operate,
to pack and to maintain for its application in electronic vehicles.80,81 Battery thermal
management system could be categorized according to a) medium type, b) power consumption
and c) whether it directly contacts with battery or not.73 Here, we are going to discuss medium-
based cooling systems: air cooling, liquid cooling and phase change materials.

Air cooling. The simplest way to cool a battery is to blow air on the battery walls
through a fan or a wind tunnel. In this case, during design of batteries, several parameters need
to be taken into account, such as the battery layout, airflow or wind speed to minimize the
energy consumption and costs. For example, Xu and He82 improved heat dissipation by simply
changing the longitudinal battery pack to a horizontal battery pack, which shortened the airflow
path. Furthermore, they found that a double “U”-type duct enhanced heat dissipation
performance rather than double “1” type. Around the battery, the air flow takes time to blow
the whole system, leading to unevenly distributed thermal gradient. To solve this issue, bi-
directional or reciprocating airflow has been implemented, which relies on cell temperature
monitoring to actively adjust the airflow direction.

Liquid cooling. Although air cooling could achieve battery thermal management as
described above, its efficiency is not sufficient to dissipate heat from an electric vehicle battery
pack under a high charge/discharge rate at high working temperature, mostly because of the
low thermal conductivity of air.83,84,79 Thus researchers introduced liquid cooling that involves
either direct immersion, or indirect circulation of thee liquid surrounding battery pack. Liquid
cooling utilizes coolants such as water, acetone, glycol and oil, and has been proved 3500 times
more efficient than air cooling.85 A cold plate with built-in channels for liquid flowing could

25
be arranged surrounding battery cells and remove excess heat. Researchers also point out that
channel design and geometries should be adjusted depending on the target applications. Cold
plate with rectangular channels achieved 5% more heat transfer than one plate with elliptical
channel design.86

Phase change materials. For air cooling and liquid cooling, an additional electrical
source is needed and other accessories such as pump, blower and cold plate are necessary as
well, leading to high costs. In this context, researchers have explored alternatives for battery
thermal management: phase change materials. The phase change, mostly solid-to-liquid, of
organic/inorganic/eutectic phase change materials could extract heat from batteries87,88 and
thus regulate the working temperature to a reasonable range. However, phase change materials
(without fixing matrix) have low thermal conductivity and thus have slower regeneration time,
which is not effective enough for fast charging-discharging cycles. Researchers have then
explored different strategies to improve heat transfer: a) using fixed and non-moving surfaces,
b) using phase change material composites, c) impregnating porous materials, and d) adding
particles with high conductivity.79 Li et al.89 applied a sandwiched cooling element composed
of copper metal foam saturated with phase change materials (paraffin) as thermal management
for prismatic power batteries. As a control, pure paraffin enabled a dramatic reduction of
battery surface temperature and maintained a suitable range of working temperature because
of its latent heat absorption and of the natural convection of melted paraffin during melting
event. A copper-paraffin composite could further reduce the surface temperature and enhance
the uniformity of the distribution of temperature due to the improvement of thermal
conductivity.

1.3.3 Metal matrix nanocomposite with phase change materials for temperature

Commonly used phase change materials – organics and salt hydrates – are restrained to
applications at relatively low temperature and suffer from limited mechanical resistance, low
stability upon cycling and low thermal conductivity. Researchers have incorporated thermally
conductive fillers (such as graphite, metallic nanoparticles) and foams (metals and graphite, for
instance) to produce phase change composites with improved thermal conductivity. The role
of fillers is limited due to a discontinuous structure. Thermal interface resistance between phase

26
change materials and fillers still prevail. On the opposite, a continuous structure with high
thermal conductivity could be obtained after incorporating phase change materials into
foams.90 Due to the difficulties of infiltration process and to corrosion, foams are difficult to
implement for phase change materials with high melting temperature.91 Metallic materials have
been found applicable as phase change materials for their high thermal conductivity (~10 to
400 W m-1K-1), high melting temperature (-40 to 300 ℃) and large enthalpy of fusion.92
Metallic phase change materials are then good candidates for thermal management.

To our knowledge, the first documented work on metal matrix nanocomposites with
phase change inclusions was published in the 1990s. Goswami et al.93 have incorporated Bi
into glassy Al-Fe-Si matrix by mixing Bi with the matrix through induction melt spinning.
Nanocomposites containing Bi nanoparticles between 5 to 50 nm were achieved. This work
was revisited in 2015, with the aim of improving the control of the nanoparticle size. Liu et
al.92 embedded Bi nanoparticles into an Ag matrix by directly adding the desired amount of Bi
nanoparticles to molten Ag. The Ag matrix was found efficient to prevent coalescence of Bi
nanoparticles and the phase change properties were stable even after 100 melt-freeze cycles.
Compared to common organic phase change materials, the incorporation of Bi nanoparticles
resulted in 50-100% improvement of thermal energy density. In addition, by tuning
nanoparticle size from 8 to 15 nm, the melting temperature of the silver matrix was shifted
from 236 to 252 ℃. To convert into a less expensive aluminium metallic matrix, Abraham et
al.26 ball-milled a mixture of Al and Bi2O3. Subsequently, this pellet was heated and held over
300 °C to promote aluminothermic reduction of Bi2O3 and form an Al2O3 layer surrounding Bi
particles within an Al matrix. The matrix encapsulating Bi inclusions could store thermal
energy upon heat-freeze cycles and retained its structure. In the control sample with non-
encapsulated Bi inclusions, molten Bi tended to separate from the Al matrix and to accumulate
on the surface of materials.

1.4 MMCs fabrication techniques

Most manufacturing techniques for the synthesis of bulk metal matrix nanocomposites have
already been reviewed in the literature among which we can cite the review of Bakshi et al.39
as well as the latest review of Casati and Vedani24. Metal matrix composites/nanocomposites
27
can be produced by many techniques (Table 1.2) mainly composed of solid-state, liquid-state
and semisolid-state processes.2 In the following paragraphs, friction stirring, spark plasma
sintering, ultrasound-assisted casting, vortex casting, squeeze casting bottom-up fabrication
and electrodeposition are going to be discussed.

Table 1.2 Class and process techniques of powder metallurgy10

1.4.1 Powder metallurgy

Powder metallurgy includes all the production processes using metal powders as matrix
precursor. It consists of three main steps: mixing metal matrix powders with the reinforcement

28
phase in the form of powders, consolidating the powders mixture, post-treating the mixture to
trigger e.g. sintering.94–96 The aluminum and/or reinforcement powders are obtained by ball
milling. Mechanical alloying, a repeated process of cold welding, fracturing and re-welding of
powder particles in a high-energy ball mill, is a fundamental technique for the preparation of
MMNCs.11 Several techniques have been used to perform the compaction of composite
powders: hot pressing, cold pressing, cold isostatic pressing (CIP), hot isostatic pressing (HIP),
equal channel angular pressing (ECAP) followed by a sintering treatment.

Powder metallurgy produces well-dispersed metal matrix composites with uniform

microstructure. It can be used to produce high melting point metals like Ti with less difficulty
than liquid metallurgy. However, many treatment steps are necessary, which increases the price
of fabrication. The presence of clusters of nanoparticles is also difficult to prevent. The
products also show a limited range of shapes that is detrimental to vehicle industry. Therefore,
the focus is put in this PhD work on liquid processing. Nonetheless, we describe below some
powder metallurgy processes that are the main approaches used up to now to reach MMNCs.

Friction stir processing is one of the powder metallurgy approaches. The introduction
of reinforcement objects into aluminum matrix can also be performed by multi-pass friction
stir process. The principle is based on three major steps: firstly, fixing the matrix sample;
secondly, drilling holes according to a predesigned pattern with pins into the matrix phase;
thirdly, inserting nanoparticles into the holes.97 Multiple passes are necessary for a uniform
distribution of reinforcement particles. The friction stir process enables the introduction of
more than one reinforcement. For example, Jalilvand et al.97 applied a friction stir processing.
48 holes were drilled on the Al356 alloy surface. 14 were filled with 80 nm Al2O3 nanoparticles
and the other 14 were filled with a powder of 155 nm SiO2 nanoparticles to fabricate the final
Al356/Al2O3+SiO2 nanocomposites (Figure 1.6). Microstructural analysis indicated a uniform
distribution of nanoparticles inside the passed zone. Compared to the A356 Al alloy and the
friction stir-processed A356, enhanced mechanical and corrosion properties were observed for
nanoparticles-reinforced A356 alloy. Several processing parameters can be tuned: rotational
and velocity, number of passes and tilting angle.97–100 The major disadvantage lies in the
agglomeration of reinforcement particles caused by the inhomogeneous heating and the
asymmetric flow of the material during processing.101
29
 Figure 1.6. Friction stir process for fabricating SiO2 and Al2O3 reinforced A356
nanocomposite.97

Spark plasma sintering is a method to consolidate powders to fabricate dense

composites. The setup for this technique encompasses a uniaxial press, punch electrodes and a
DC pulse current generator (Figure 1.7). The process general consists in four stages: a) creating
a vacuum environment, b) exerting pressure, c) heating by DC pulse discharge that goes
through the sample, and d) cooling and pressure release stage. This technique possesses several
advantages, such as high heating rate, ability to reach enhanced densification with limited grain
growth while promoting diffusion mechanisms. Saheb et al.13 firstly prepared an homogeneous
powder made of Al with 10 vol.% Al2O3 nanoparticles by ball milling for 24 hours and then
processed the mixed powder in a spark plasma sintering setup with sintering temperature of
550 ℃, holding time of 20 min, compaction pressure of 50 MPa and heating rate of 200 ℃ min-
1
. The final Al/10Al2O3 nanocomposite exhibited hardness of 1309 MPa and yield strength of
311 MPa, significantly higher than those of bare aluminum. The Al/10Al2O3 MMNC also
showed lower thermal expansion. The main mechanisms of SPS sintering is Joules heating of
the sample, but further experiments should be performed to elucidate the mechanisms and
figure out whether other phenomena such as plasma, sparks or electromigration exist or not.102

30
 Figure 1.7 Scheme of a spark plasma sintering device.102

1.4.2 Liquid state process

In liquid metallurgy techniques, the reinforcement phase is added to the molten metal
matrix in order to attain a desirable composition. There are some significant challenges to
incorporate nanoparticles in molten metals. Firstly, the melt temperature is relatively high,
some nanoparticles are not stable enough in these conditions; second, nanoparticles tend to
float on the surface of the molten metal due to the poor wettability of the nanoparticles with
the molten metal; third, nanoparticles agglomerate due to attractive Van der Waals force and
to the lack of a repulsive force in molten metals; and fourth, nanoparticles are normally pushed
to the grain boundary by solidification fronts during solidification.103

Three main liquid state processes are discussed here: stir casting, vortex casting,
squeeze casting.

31
1.4.2.1 Ultrasound-assisted casting

Figure 1.8. Schematic diagram of ultrasonic assisted casting.103

Stir casting has been widely applied as a commercial technique to fabricate MMCs. In
conventional stir casting, the metal matrix is firstly de-moistured and heated above its melting
temperature. Then the de-moistured reinforcement phase is added into the molten metal,
followed by stirring with a stirrer to homogenize the reinforcement/metal matrix mixture. After
cooling down in a mold, the final MMCs with desired shape could be obtained. But two main
challenges prevail for this technique: one is related to the quantitative incorporation of the
objects in the melt, the other one is the homogeneous distribution of the nano-reinforcements
because of the high surface-to-volume ratio of the objects and of their poor wettability by the
molten matrix, resulting in particle segregation and floating due to the density difference
between the matrix and the nano-reinforcement objects.104,105
32
 In order to solve these challenges, ultrasonication has been developed to disperse
nanoparticles in molten metals (Figure 1.8). Eskin106 has explained the application of
ultrasound treatment to light alloy melts. In melts, the mechanism of cavitation is similar to
other liquids. High intensity ultrasonic waves are introduced to generate transient cavitation
and acoustic streaming, which are used to refine the microstructure, degas the melt and finally
homogenize the distribution of reinforcement nano-objects.107 During this process, the
ultrasonic cavitation produces a strong implosive impact that could on one hand help to well
disperse the nano-reinforcements and on the other hand enhance the wettability of the
reinforcements by the molten metal.108

Table 1.3: Sound propagation rate in different metals and water106

The varied material-dependent propagation sound speed (see Table 1.3), thermal
conductivity and viscosity influence the acoustic impedance and the attenuation factor. The
surface tension of metal melts is significantly higher than that of water (0.86 N m-1 for Al at
700 °C vs. 0.079 N m-1 for water at 20 °C), which decreases the dispersion efficiency. In order
to achieve a good dispersion and distribution of nano-reinforcements, this other difficulty must
be overcome when fabricating MMNCs.106

33
 The production of composite materials with particles via ultrasound-assisted casting
comprises several steps.

Step 1: Preparation of the particles with suitable surface treatment to decrease the
surface tension and increase wetting of the reinforcement phase particles by the molten metal.
A coating with metal like Cu or Ni, oxidation (like Al2O3), degassing and drying is useful to
reach an homogeneous dispersion.24

Step 2: Preparation of the metal matrix. Additives for surface tension control are useful,
especially for the liquid Al melt, which has a high surface tension. A lower surface tension
metal like Bi, Pb, Sb, Li, Ca, Mg and Sn can be used to decrease the surface tension of Al.109

Step 3: Incorporation of the particles. The simplest technique consists in spraying the
particles onto the surface of the melt. This technique works relatively well for magnesium
MMCs19 but has limitations in for MMCs due to the thick oxide film at the surface of the melt
and to the high surface tension of the melt. In some cases, mechanical mixing using an impeller
under protective atmosphere are combined with ultrasonic mixing.110 Besides, the ultrasound
processing parameters, the melt temperature, duration of processing, size, morphology and
content of reinforcement particles can be adjusted to improve the dispersion.24 One way to
incorporate small particles while preventing flotation and macroscopic phase separation is to
add the particles to the matrix in the semiliquid state (the temperature of the metal matrix is put
between its liquidus and solidus temperature). For instance, SiC nanoparticles were added to a
slurry of Mg-18% Zn alloy and dispersed with mechanical stirring. Then the composite
material was heated to the liquid state and ultrasonicated to get a satisfying dispersion.111

Step 4: Casting and solidification. Ultrasonication in light alloy melts may affect
nucleation and release of dissolved gas from the liquid phase. In semiliquid mixture, it
influences the structure formation by fragmentation of solid crystals and their distribution.

In order to well understand the effect of ultrasonic stirring on the structure and
mechanical properties of MMNCs, Jia et al.112 applied ultrasonic treatment to
incorporate 1 wt.% SiC or Al2O3 nanoparticles to aluminum alloy A356. The authors firstly
melted the alloy in an induction furnace and then added nanoparticles into molten liquid under
34
ultrasonic stirring with power of 1.75 kW, frequency of 18 kHz. Subsequent solidification was
performed to obtain aluminum metal matrix nanocomposite. The ultrasonic treatment could
change the grain structure from dendritic to globular shape and also contribute to reasonable
dispersion of nanoparticles into the final composite although some agglomeration still existed.
Furthermore, the tensile strength of the nanoparticle-reinforced composites was higher with
ultrasonic treatment.

The ultrasound assisted casting is highly dependent on the stability of nanoparticles and
wettability of the inclusions by the matrix. Several combinations1 are proven to be efficient,
especially for the SiC/Mg6Zn couple. Research of novel nanoparticles/metal matrix is still in
need to achieve better mechanical properties or other atypic properties.

1.4.2.2 Vortex casting

Vortex casting has been proven useful to uniformly distribute reinforcement objects into a
metal matrix. In a typical experiment, the milled reinforcement phase and matrix powders are
continuously added to molten aluminum or aluminum alloy under stirring. Another stirring is
performed to homogenize the mixtures after several minutes’ crystallization. The final MMCs
are obtained by shaping in a mold and an extrusion process. However, this method is difficult
to be used for the incorporation of nanosized particles as it is difficult to reach high dispersion.

1.4.3 Squeeze casting

Squeeze casting is also a casting technique that involves two main steps: a) reinforcement
objects addition into the molten metal matrix under mechanical stirring; b) solidification under
high squeeze pressure. This technique can reduce porosity and shrinkage, improve casting yield
and enhance properties of the final MMCs.113 Lü et al.60 firstly fabricated a millimeter-sized
granule composed of 8 wt% SiC nanoparticles in aluminum by dry high-energy ball milling,
then remelted and diluted this granule in molten A356 alloy to get final 1 wt% nanoparticles
composites under ultrasonication. The molten mixture was poured into a die cavity that was
disposed into an environment with 200 MPa squeeze pressure for 30 s until the end of
35
solidification (Figure 1.9). As a result, the nanoparticles were uniformly distributed into the
nanocomposite and tensile strength of squeeze casted MMNCs was 2169 MPa, 25% higher
than the non-reinforced alloy matrix.

Figure 1.9. Flow chart of the fabrication of squeeze casted A356/SiC nanocomposites.60

1.4.4 Other techniques

1.4.4.1 Bottom-up fabrication

Luechinger et al.114 have fabricated a bismuth and cobalt alloy by combining two immiscible
metals at the nanoscale: bismuth and cobalt. Nanoparticles are mixed and pressed into pellets,
then sintered to decrease the porosity. This bottom-up approach allows a perfect dispersion of
the nanocrystalline phase, which could increase hardness and strength. However, porosity still
exists and the fabrication needs pre-synthesized nanoparticles without passivating oxide layer,
which is difficult to achieve for highly reductive metals like magnesium and aluminum.

36
1.4.4.2 Electrodeposition

The group of Akbulut presented a new hybrid nanocomposite thermal interface material (TIMs)
relying on copper or silver, boron nitride nanosheets (BNNSs) and organic ligands. The
composite was prepared by chemisorption-coupled electrodeposition approach.115 BN
nanosheets were exfoliated and functionalized by bifunctional ligands, which were supposed
to link the metal and the BNNSs together through Lewis acid-base interactions. The co-
electrodeposition/chemisorption process was then applied, which coupled the electrolytic
reduction of metal ions and the chemisorption of functionalized BN nanosheets on the cathode.
The thermal conductivity of the final composite was 236–258 W m-1 K-1 and 306–321 W m-
1
K-1 for physically and chemically conjugated BNNS-silver-ligands materials, respectively,
while their hardness varied from 400–495 MPa and from 240 to 360 MPa, respectively. The
electrodeposition method is suitable to fabricate coating materials, but is hard to be applied to
bulk materials.

1.5 Conclusion and motivation of this work

Faced with energy and environmental challenges and in line with the consumer demand, the
Stellantis has been investing for several years in manufacturing vehicles that are more envir
onmental friendly. To execute this strategy, the group wishes to invent and fabricate new
materials that are light-weight and low cost with high ductility and also with good thermal
conductivity. In this context, this PhD work aims to investigate MMNCs with improved
mechanical properties and thermal properties. In particular, the fabrication processes must be
designed to meet the needs of the current casting process of alloy.

Among the four types of fabrication techniques shown in this chapter, ultrasonic-
assisted casting can meet these requirements. The use of ultrasonic treatment is a good way to
improve the dispersion of nanoparticles during the casting process of aluminum alloy. This
process also offers advantages such as low cost, easy molding process and adaptability to large-
scale fabrication.

37
 The literature review of current MMNCs opens many intriguing opportunities for the
development of new materials. Regarding the mechanical properties of MMNCs, commercial
inorganic nanoparticles have been proven useful to efficiently improve strength and
functionality of the metal matrix. The reinforcement nanoparticles must be designed to satisfy
the strengthening mechanisms and harsh casting conditions. Boride nanoparticles synthesized
by molten salts are good candidates as they are small in size, light, hard and stable at high
temperature. At the time when we started this project, there was only two papers available in
the literature that used molten salt synthesized nanoparticles (particles size between 10 to 100
nm) as ex situ reinforcement.63 Other related works were published recent years, especially
from the group of Li116 who worked on the incorporation of molten-salt synthesized TiB2 and
ZrB2 nanoparticles in aluminum. A good dispersion of nanoparticles is essential for the
mechanical properties and is a clear bottleneck in the facile fabrication of MMNCs with various
reinforcement phases. It requires thorough control of compositional parameters (wettability of
the matrix, nature of the compounds) and of processing parameters (heating process, ultrasound
treatment power, heating/cooling rate). This part of the work will be explored in chapter 3.

Phase change materials have limited applications in vehicle’s battery thermal control
because of their low thermal conductivity and low resistance against vibration. Composites
made by the incorporation of phase change materials in metal matrices are an interesting
candidate for the automobile industry due to their high thermal conductivity, good mechanical
properties and wide working temperature range. Conventional strategies for MMNCs design
include powder metallurgy and bottom-up fabrication, which require high fabrication costs.
Additionally, the working temperature is also limited: phase change only occurs at the melting
temperature of pure metal, which is too high in most cases. To overcome these issues, we
propose a liquid phase method: metallic nanoparticles (pure metal or alloy) will be
encapsulated in an inorganic shell to prevent macroscopic phase separation in the melt.
Subsequently, these nanoparticles will be incorporated in molten metal matrix by ultrasound-
assisted process. This work will be presented in two parts: chapter 2 will deal with the synthesis
of the building block core-shell nanoparticles, chapter 4 will deal with the fabrication of the
nanocomposites and their thermal properties.

38
References

1. Ramanathan, A., Krishnan, P. K. & Muraliraja, R. A review on the production of metal

matrix composites through stir casting – Furnace design, properties, challenges, and
research opportunities. J. Manuf. Process. 42, 213–245 (2019).
2. Mondal, S. Aluminum or Its Alloy Matrix Hybrid Nanocomposites. Met. Mater. Int.
(2020) doi:10.1007/s12540-020-00750-5.
3. Petukhova, Y. V. et al. Polymer composites containing dispersed VO2 of various
polymorphs: Effects of polymer matrix on functional properties. Mater. Chem. Phys.
235, (2019).
4. Antil, P., Singh, S. & Manna, A. Analysis on effect of electroless coated SiCp on
mechanical properties of polymer matrix composites. Part. Sci. Technol. 37, 787–794
(2019).
5. Balamurugan, P., Uthayakumar, M. & Niemczewska-Wójcik, M. Wear studies of
copper-fly ash composite under dry sliding conditions. Mater. Res. Express 6, (2019).
6. Song, Y. et al. Effect of carbon-fibre powder on friction and wear properties of copper-
matrix composites. Mater. Sci. Technol. (United Kingdom) 36, 92–99 (2020).
7. Li, M. et al. Microstructure evolution and properties of graphene nanoplatelets
reinforced aluminum matrix composites. Mater. Charact. 140, 172–178 (2018).
8. Kim, J. Tensile fracture behavior and characterization of ceramic matrix composites.
Materials (Basel). 12, (2019).
9. Ceschini, L.; Dahle, A.; Gupta, M.; Jarfors, A.E.W.; Jayalakshmi, S.; Morri, A.;
Rotundo, F.; Toschi, S.; Singh, R. . Aluminum and Magnesium Metal Matrix
Nanocomposites. (Springer, New York, NY, 2017).
10. Lloyd, D. J. Particle reinforced aluminium and magnesium matrix composites. Int.
Mater. Rev. 39, 1–23 (1994).
11. Pardeep, S., Gulshan, C. & Neeraj, S. Production of AMC by Stir Casting – An
Overview. Int. J. Contemp. Pract. 2, 23–46 (2013).
12. Ardakani, M. R. K., Amirkhanlou, S. & Khorsand, S. Cross accumulative roll bonding-
A novel mechanical technique for significant improvement of stir-cast Al/Al2O3
nanocomposite properties. Mater. Sci. Eng. A 591, 144–149 (2014).
13. Saheb, N. & Khan, M. S. Compressive strength and thermal properties of spark plasma
sintered Al-Al2O3 nanocomposite. Sci. Sinter. 50, 1–14 (2018).
14. Kandemir, S. Microstructure and mechanical properties of A357/SiC nanocomposites
fabricated by ultrasonic cavitation-based dispersion of ball-milled nanoparticles. J.
Compos. Mater. 51, 395–404 (2017).
15. Law, E., Pang, S. D. & Quek, S. T. Discrete dislocation analysis of the mechanical
response of silicon carbide reinforced aluminum nanocomposites. Compos. Part B Eng.
42, 92–98 (2011).
16. Cao, C., Ling, H., Murali, N. & Li, X. In-situ molten salt reaction and incorporation of
small (10 nm) TiC nanoparticles into Al. Materialia 7, (2019).
17. Dhandapani, S., Rajmohan, T., Palanikumar, K. & Charan, M. Synthesis and
39
 characterization of dual particle (MWCT+B4C) reinforced sintered hybrid aluminum
matrix composites. Part. Sci. Technol. 34, 255–262 (2016).
18. Sharma, R., P, S. J., Kakkar, K., Kamboj, K. & Sharma, P. A Review of the Aluminium
Metal Matrix Composite and its Properties. Int. Res. J. Eng. Technol. 4, 832–842 (2017).
19. Tjong, S. C. Novel nanoparticle-reinforced metal matrix composites with enhanced
mechanical properties. Adv. Eng. Mater. 9, 639–652 (2007).
20. Zhong, X. L., Wong, W. L. E. & Gupta, M. Enhancing strength and ductility of
magnesium by integrating it with aluminum nanoparticles. Acta Mater. 55, 6338–6344
(2007).
21. Zhang, F., Liu, S., Zhao, P., Liu, T. & Sun, J. Titanium/nanodiamond nanocomposites:
Effect of nanodiamond on microstructure and mechanical properties of titanium. Mater.
Des. 131, 144–155 (2017).
22. Kim, K. T., Cha, S. Il, Gemming, T., Eckert, J. & Hong, S. H. The role of interfacial
oxygen atoms in the enhanced mechanical properties of carbon-nanotube-reinforced
metal matrix nanocomposites. Small 4, 1936–1940 (2008).
23. Ibrahim, I. A., Mohamed, F. A. & Lavernia, E. J. Particulate reinforced metal matrix
composites - a review. J. Mater. Sci. 26, 1137–1156 (1991).
24. Casati, R. & Vedani, M. Metal Matrix Composites Reinforced by Nano-Particles—A
Review. 65–83 (2014)
25. Veličković, S., Stojanović, B., Ivanović, L., Miladinović, S. & Milojević, S. Application
of Nanocomposites in the Automotive Industry. Mobil. Veh. Mech. 45, 51–64 (2019).
26. Abraham, A., Schoenitz, M. & Dreizin, E. L. Energy storage materials with oxide-
encapsulated inclusions of low melting metal. Acta Mater. 107, 254–260 (2016).
27. Thomas, K. E. & Newman, J. Thermal Modeling of Porous Insertion Electrodes. J.
Electrochem. Soc. 150, A176 (2003).
28. Nardone, V. C. & Prewo, K. M. On the strength of discontinuous silicon carbide
reinforced aluminum composites. Scr. Metall. 20, 43–48 (1986).
29. Ramakrishnan, N. An analytical study on strengthening of particulate reinforced metal
matrix composites. Acta Mater. 44, 69–77 (1996).
30. Miller, W. S. & Humphreys, F. J. Strengthening mechanisms in particulate metal matrix
composites. Scr. Metall. Mater. 25, 33–38 (1991).
31. Zhang, Z. & Chen, D. L. Consideration of Orowan strengthening effect in particulate-
reinforced metal matrix nanocomposites: A model for predicting their yield strength.
Scr. Mater. 54, 1321–1326 (2006).
32. Zhang, Q. & Chen, D. L. A model for predicting the particle size dependence of the low
cycle fatigue life in discontinuously reinforced MMCs. Scr. Mater. 51, 863–867 (2004).
33. Cao, G., Choi, H., Oportus, J., Konishi, H. & Li, X. Study on tensile properties and
microstructure of cast AZ91D/AlN nanocomposites. Mater. Sci. Eng. A 494, 127–131
(2008).
34. Hull, D. & Bacon, D. J. Introduction to dislocations. (Butterworth Einemann: Oxford,
UK, 2001).
35. Sanaty-Zadeh, A. Comparison between current models for the strength of particulate-

40
 reinforced metal matrix nanocomposites with emphasis on consideration of Hall-Petch
effect. Mater. Sci. Eng. A 531, 112–118 (2012).
36. Zhang, S., Sun, D., Fu, Y. & Du, H. Recent advances of superhard nanocomposite
coatings: A review. Surf. Coatings Technol. 167, 113–119 (2003).
37. Malaki, M., Xu, W., Kasar, A. K., Menezes, P. L. & Dieringa, H. Advanced Metal
Matrix Nanocomposites. (2013).
38. Li, Z. et al. Enhanced Mechanical Properties of Graphene (Reduced Graphene
Oxide)/Aluminum Composites with a Bioinspired Nanolaminated Structure. Nano Lett.
15, 8077–8083 (2015).
39. Bakshi, S. R., Lahiri, D. & Agarwal, A. Carbon nanotube reinforced metal matrix
composites - a review. Int. Mater. Rev. 55, 41–64 (2010).
40. Peng B, Locascio M, Zapol P, Shuyou L, Mielke SL, Schatz GC, et al. Measurements
of near-ultimate strength for multiwalled carbon nanotubes and irradiation-induced
crosslinking improvements. Nat Nanotechnol 3, 626–31 (2008).
41. Choi H, Shin J, Min B, Park J, B. D. Reinforcing effects of carbon nanotubes in structural
aluminum matrix nanocomposites. J Mater Res 24, 2610–6 (2009).
42. Ma, Z. Y., Lia, Y. L., Zheng, Y. L. F., Bp, J. & Tjong, S. C. Nanometric Si3N 4
particulate-reinforced aluminum composite. 219, 229–231 (1996).
43. Chaira, D. et al. Synthesis of Aluminium – Cementite Metal Matrix Composite by
Mechanical Alloying Synthesis of Aluminium – Cementite Metal Matrix Composite.
6914, (2016).
44. Lahiri, D. et al. Insight into reactions and interface between boron nitride nanotube and
aluminum. J. Mater. Res. 27, 2760–2770 (2012).
45. Firestein, K. L. et al. High-strength aluminum-based composites reinforced with BN,
AlB2 and AlN particles fabricated via reactive spark plasma sintering of Al-BN powder
mixtures. Mater. Sci. Eng. A 681, 1–9 (2017).
46. Luo, X. & Li, C. Materials & Design Large sized cubic BN reinforced nanocomposite
with improved abrasive wear resistance deposited by cold spray. Mater. Des. 83, 249–
256 (2015).
47. Kennedy, A. R., Weston, D. P. & Jones, M. I. Reaction in Al-TiC metal matrix
composites. Mater. Sci. Eng. A 316, 32–38 (2001).
48. Gopalakannan, S. & Senthilvelan, T. Application of response surface method on
machining of Al-SiC nano-composites. Meas. J. Int. Meas. Confed. 46, 2705–2715
(2013).
49. D. Sujan, Z. Oo, M. E. Rahman, M. A. Maleque, C. K. T. Physio-mechanical Properties
of Aluminium Metal Matrix Composites Reinforced with Al2O3 and SiC. 6, 678–681
(2012).
50. Miracle, D. B. Metal matrix composites - From science to technological significance.
Compos. Sci. Technol. 65, 2526–2540 (2005).
51. Li, A. B. et al. Enhanced combination of strength and ductility in ultrafine-grained
aluminum composites reinforced with high content intragranular nanoparticles. Mater.
Sci. Eng. A 745, 10–19 (2019).
52. Nampoothiri, J., Harini, R. S., Nayak, S. K., Raj, B. & Ravi, K. R. Post in-situ reaction
41
 ultrasonic treatment for generation Al–4.4Cu/TiB2 nanocomposite: A route to enhance
the strength of metal matrix nanocomposites Jayakrishnan. J. Alloys Compd. (2016)
53. Li, M. et al. Synthesis and mechanical behavior of nanostructured al 5083/ n-TiB2 metal
matrix composites. Mater. Sci. Eng. A (2016) doi:10.1016/j.msea.2016.01.031.
54. Yuan, L., Han, J., Liu, J. & Jiang, Z. Mechanical properties and tribological behavior of
aluminum matrix composites reinforced with in situ AlB2 particles. Tribiology Int.
(2016)
55. Tian, K. et al. Effects of in situ generated ZrB2 nano-particles on microstructure and
tensile properties of 2024Al matrix composites. J. Alloys Compd. 594, 1–6 (2014).
56. Ward-Close, C. M., Minor, R. & Doorbar, P. J. Intermetallic-matrix composites—a
review. Intermetallics 4, 217–229 (1996).
57. Hsu, C. J., Chang, C. Y., Kao, P. W., Ho, N. J. & Chang, C. P. Al-Al3Ti nanocomposites
produced in situ by friction stir processing. Acta Mater. 54, 5241–5249 (2006).
58. Xiu, Z. et al. Microstructure and Mechanical Properties of 45 vol.% SiCp/7075Al
Composite. J. Mater. Sci. Technol. 31, 930–934 (2015).
59. Gao, Q. et al. Effects of ultrasonic vibration treatment on particles distribution of TiB2
particles reinforced aluminum composites. Mater. Sci. Eng. A 680, 437–443 (2017).
60. Lü, S., Xiao, P., Yuan, D., Hu, K. & Wu, S. Preparation of Al matrix nanocomposites
by diluting the composite granules containing nano-SiCp under ultrasonic vibaration. J.
Mater. Sci. Technol. 34, 1609–1617 (2018).
61. Pourhosseini, S., Beygi, H. & Sajjadi, S. A. Effect of metal coating of reinforcements
on the microstructure and mechanical properties of Al-Al2O3 nanocomposites. Mater.
Sci. Technol. (United Kingdom) 34, 145–152 (2018).
62. Taherzadeh Mousavian, R. et al. Incorporation of SiC Ceramic Nanoparticles into the
Aluminum Matrix by a Novel Method: Production of a Metal Matrix Composite. Met.
Mater. Int. (2020) doi:10.1007/s12540-019-00604-9.
63. Karbalaei Akbari, M., Baharvandi, H. R. & Shirvanimoghaddam, K. Tensile and fracture
behavior of nano/micro TiB2 particle reinforced casting A356 aluminum alloy
composites. Mater. Des. 66, 150–161 (2015).
64. El-Kady, E.-S. Y., Mahmoud, T. S. & Sayed, M. A.-A. Elevated Temperatures Tensile
Characteristics of Cast A356/Al&lt;sub&gt;2&lt;/sub&gt;O&lt;sub&gt;3&lt;/sub&gt;
Nanocomposites Fabricated Using a Combination of Rheocasting and Squeeze Casting
Techniques. Mater. Sci. Appl. 02, 390–398 (2011).
65. Vorozhtsov, S. et al. The Influence of ScF3 Nanoparticles on the Physical and
Mechanical Properties of New Metal Matrix Composites Based on A356 Aluminum
Alloy. JOM 68, 3101–3106 (2016).
66. Reddy, M. P. et al. Enhancing thermal and mechanical response of aluminum using
nanolength scale TiC ceramic reinforcement. Ceram. Int. 44, 9247–9254 (2018).
67. Reddy, M. P. et al. Enhanced performance of nano-sized SiC reinforced Al metal matrix
nanocomposites synthesized through microwave sintering and hot extrusion techniques.
Prog. Nat. Sci. Mater. Int. 27, 606–614 (2017).
68. Hwang, J. et al. Enhanced mechanical properties of graphene/copper nanocomposites
using a molecular-level mixing process. Adv. Mater. 25, 6724–6729 (2013).
42
69. Wang, F., Chu, K., Li, Y. biao & Wang, X. hu. Enhanced Interfacial Bonding and
Mechanical Properties of Graphene/Cu Composites: A Matrix-Alloying Method. Phys.
Status Solidi Appl. Mater. Sci. 215, (2018).
70. Dorri Moghadam, A., Omrani, E., Menezes, P. L. & Rohatgi, P. K. Mechanical and
tribological properties of self-lubricating metal matrix nanocomposites reinforced by
carbon nanotubes (CNTs) and graphene - A review. Compos. Part B Eng. 77, 402–420
(2015).
71. Tarascon, J.-M. & Armand, M. Issues and challenges facing rechargeable lithium
batteries. in Materials For Sustainable Energy 171–179 (A Collection of Peer-Reviewed
Research and Review Articles from Nature Publishing Group, 2011).
72. Oberthür, S. The New Climate Policies of the European Union: Internal Legislation and
Climate Diplomacy.
73. Kim, J., Oh, J. & Lee, H. Review on battery thermal management system for electric
vehicles. Applied Thermal Engineering vol. 149 192–212 (2019).
74. Bukhari, S. M. A. S., Maqsood, J., Baig, M. Q., Ashraf, S. & Khan, T. A. Comparison
of Characteristics-Lead Acid, Nickel Based, Lead Crystal and Lithium Based Batteries.
in Proceedings UKSim 2015 444–450 (Institute of Electrical and Electronics Engineers
Inc., 2015).
75. Lowe, M., Tokuoka, S., Trigg, T. & Gereffi, G. Lithium-ion Batteries for Electric
Vehicles. (Tech. Rep, 2010).
76. Offer, G., Patel, Y., Hales, A., Bravo Diaz, L. & Marzook, M. Cool metric for lithium-
ion batteries could spur progress. Nature 582, 485–487 (2020).
77. Yuan, X., Liu, H. & Zhang, J. Lithium-ion batteries : advanced materials and
technologies. (CRC Press Taylor & Francis Group, 2012).
78. Adair, D., Ismailov, K. & Bakenov, Z. Thermal Management of Lithium-ion Battery
Packs. (2014).
79. Arora, S. Selection of thermal management system for modular battery packs of electric
vehicles: A review of existing and emerging technologies. J. Power Sources 400, 621–
640 (2018).
80. Khan, M. R., Swierczynski, M. J. & Kær, S. K. Towards an ultimate battery thermal
management system: A review. Batteries vol. 3 (2017).
81. Pesaran, A. A. An Approach for Designing Thermal Management Systems for Electric
and Hybrid Vehicle Battery Packs Multi-scale mechanical-electrochemical-thermal
coupled modeling framework for lithium-ion battery under mechanical abuse View
project. (1999).
82. Xu, X. M. & He, R. Research on the heat dissipation performance of battery pack based
on forced air cooling. J. Power Sources 240, 33–41 (2013).
83. Wu, M. S., Liu, K. H., Wang, Y. Y. & Wan, C. C. Heat dissipation design for lithium-
ion batteries. J. Power Sources 109, 160–166 (2002).
84. Nelson, P., Dees, D., Amine, K. & Henriksen, G. Modeling thermal management of
lithium-ion PNGV batteries. J. Power Sources 110, 349–356 (2002).
85. Xia, G., Cao, L. & Bi, G. A review on battery thermal management in electric vehicle
application. Journal of Power Sources vol. 367 90–105 (2017).
43
86. Fisher, T. S., Torrance, K. E. & Torrance, K. E. Optimal Shapes of Fully Embedded
Channels for Conjugate Cooling. IEEE TRANSACTIONS ON ADVANCED
PACKAGING vol. 24 (2001).
87. Al Hallaj, S. & Selman, J. R. A Novel Thermal Management System for Electric Vehicle
Batteries Using Phase-Change Material. Journal of The Electrochemical Society vol.
147 (2000).
88. Jaguemont, J., Omar, N., Van den Bossche, P. & Mierlo, J. Phase-change materials
(PCM) for automotive applications: A review. Applied Thermal Engineering vol. 132
308–320 (2018).
89. Li, W. Q., Qu, Z. G., He, Y. L. & Tao, Y. B. Experimental study of a passive thermal
management system for high-powered lithium ion batteries using porous metal foam
saturated with phase change materials. J. Power Sources 255, 9–15 (2014).
90. Ji, H. et al. Enhanced thermal conductivity of phase change materials with ultrathin-
graphite foams for thermal energy storage. Energy Environ. Sci. 7, 1185–1192 (2014).
91. Zhao, C. Y. & Wu, Z. G. Heat transfer enhancement of high temperature thermal energy
storage using metal foams and expanded graphite. Sol. Energy Mater. Sol. Cells 95, 636–
643 (2011).
92. Liu, M., Ma, Y., Wu, H. & Wang, R. Y. Metal matrix-metal nanoparticle composites
with tunable melting temperature and high thermal conductivity for phase-ch1. Liu M,
Ma Y, Wu H, Wang RY. Metal matrix-metal nanoparticle composites with tunable
melting temperature and high thermal conductivity . ACS Nano 9, 1341–1351 (2015).
93. Goswami, R. & Chattopadhyay, K. Depression of melting point of multidomained
bismuth in aluminum based metallic glass nanocomposites. Appl. Phys. Lett. 69, 910–
912 (1996).
94. Sharifi, E. M. & Karimzadeh, F. Wear behavior of aluminum matrix hybrid
nanocomposites fabricated by powder metallurgy. Wear 271, 1072–1079 (2011).
95. Arif, S., Alam, T., Ansari, A. H. & Shaikh, M. B. N. Morphological characterization,
statistical modelling and tribological behaviour of aluminum hybrid nanocomposites
reinforced with micro-nano-silicon carbide. J. Asian Ceram. Soc. 7, 434–448 (2019).
96. Ahmadi, M. & Siadati, M. H. Synthesis, mechanical properties and wear behavior of
hybrid Al/(TiO2 + CuO) nanocomposites. J. Alloys Compd. 769, 713–724 (2018).
97. Jalilvand, M. M., Mazaheri, Y., Heidarpour, A. & Roknian, M. Development of
A356/Al2O3 + SiO2 surface hybrid nanocomposite by friction stir processing. Surf.
Coatings Technol. 360, 121–132 (2019).
98. Nazari, M., Eskandari, H. & Khodabakhshi, F. Production and characterization of an
advanced AA6061-Graphene-TiB2 hybrid surface nanocomposite by multi-pass friction
stir processing. Surf. Coatings Technol. 377, (2019).
99. Ostovan, F., Amanollah, S., Toozandehjani, M. & Shafiei, E. Fabrication of Al5083
surface hybrid nanocomposite reinforced by CNTs and Al 2 O 3 nanoparticles using
friction stir processing . J. Compos. Mater. 54, 1107–1117 (2020).
100. Heidarpour, A., Ahmadifard, S. & Kazemi, S. On the Al5083–Al2O3–TiO2 Hybrid
Surface Nanocomposite Produced by Friction Stir Processing. Prot. Met. Phys. Chem.
Surfaces 54, 409–415 (2018).

44
101. Rathee, S., Maheshwari, S., Siddiquee, A. N. & Srivastava, M. A Review of Recent
Progress in Solid State Fabrication of Composites and Functionally Graded Systems Via
Friction Stir Processing. Critical Reviews in Solid State and Materials Sciences vol. 43
334–366 (2018).
102. Saheb, N. et al. Spark plasma sintering of metals and metal matrix nanocomposites: A
review. J. Nanomater. 2012, 1–13 (2012).
103. Wang, F. et al. A synchrotron X-radiography study of the fragmentation and refinement
of primary intermetallic particles in an Al-35 Cu alloy induced by ultrasonic melt
processing. Acta Mater. 141, 142–153 (2017).
104. Su, H., Gao, W., Feng, Z. & Lu, Z. Processing, microstructure and tensile properties of
nano-sized Al2O3 particle reinforced aluminum matrix composites. Mater. Des. 36,
590–596 (2012).
105. Garg, P. et al. Advance research progresses in aluminium matrixcomposites:
Manufacturing & applications. Journal of Materials Research and Technology (2019)
106. Eskin, G. I. E. D. G. Ultrasonic treatment of light alloy melts, second ed. (CRC Press,
2015).
107. Eskin, G. I. Cavitation mechanism of ultrasonic melt degassing. Ultrason. -
Sonochemistry 2, 20–28 (1995).
108. Cao, G. et al. Mg-6Zn/1.5%SiC nanocomposites fabricated by ultrasonic cavitation-
based solidification processing. J. Mater. Sci. 43, 5521–5526 (2008).
109. Mohanty, P., Mahapatra, R., Padhi, P., Ramana, C. H. V. V. & Mishra, D. K. Ultrasonic
cavitation: An approach to synthesize uniformly dispersed metal matrix
nanocomposites—A review. Nano-Structures and Nano-Objects 23, 100475 (2020).
110. Khandelwal, A., Mani, K., Srivastava, N., Gupta, R. & Chaudhari, G. P. Mechanical
behavior of AZ31/Al2O3 magnesium alloy nanocomposites prepared using ultrasound
assisted stir casting. Compos. Part B Eng. 123, 64–73 (2017).
111. Chen, L. Y., Peng, J. Y., Xu, J. Q., Choi, H. & Li, X. C. Achieving uniform distribution
and dispersion of a high percentage of nanoparticles in metal matrix nanocomposites by
solidification processing. Scr. Mater. 69, 634–637 (2013).
112. Jia, S., Zhang, D., Xuan, Y. & Nastac, L. An experimental and modeling investigation
of aluminum-based alloys and nanocomposites processed by ultrasonic cavitation
processing. Appl. Acoust. 103, 226–231 (2016).
113. Vijayaram, T. R., Sulaiman, S., Hamouda, A. M. S. & Ahmad, M. H. M. Fabrication of
fiber reinforced metal matrix composites by squeeze casting technology. J. Mater.
Process. Technol. 178, 34–38 (2006).
114. Luechinger, N. A., Grass, R. N., Athanassiou, E. K. & Stark, W. J. Bottom-up
fabrication of metal/metal nanocomposites from nanoparticles of immiscible metals.
Chem. Mater. 22, 155–160 (2010).
115. N. Nagabandi, C. Yegin, X. Feng, C. King, J.K. Oh, E.A. Scholar, S. N. & Akbulut, M.
Chemically Linked Metal-Matrix Nanocomposites of Boron Nitride Nanosheets and
Silver as Thermal Interface Materials. (2018)
116. Javadi, A., Pan, S. & Li, X. Fabrication of High Strength Al Nanocomposites with
Populous TiB2 Nanoparticles. Procedia Manuf. 26, 629–632 (2018).
45
46
 Chapter 2
Synthesis of low melting point
Bi@SiO2 nanoparticles

47
48
2.1 Introduction

As described in chapter 1, metal matrix phase change nanocomposites are promising

candidates for the thermal control of battery packs. To design such nanocomposites, we have
selected Bi nanoparticles as proof-of-concept phase change material for their relatively low
melting temperature. We focus on an aluminium matrix to target light weight composites. A
Bi@Al2O3-Al nanocomposite has been successfully prepared previously by powder
metallurgy.1 In their process, a mixture of Al and Bi2O3 was ball milled then the pellet was
heated and held above 300 °C to promote aluminothermic reduction to form an Al2O3 layer
surrounding Bi particles within the Al matrix. The Main limitation of its use on battery pack is
its high fabrication cost and low adaptability of the temperature range. Therefore, we
investigate herein the ability of liquid metallurgy processes to design metal matrix phase
change nanocomposites. To do so, we must first elaborate an approach that will avoid the
coalescence of molten Bi droplets in the molten aluminium matrix. For this purpose, we seek
to protect the Bi nanoparticles against coalescence by a silica shell to prevent nanoparticles
coalescence. The objective of this chapter is to present the synthesis and characterization of
these building blocks: bismuth nanoparticles encapsulated in a silica shell.

2.1.1 State of the art on the Bi synthesis

Bismuth nanoparticles are of interest for several applications such as catalysis2 and
biomedical contrast agents in computed tomography scan or X-Ray imaging.3 In the latter case,
they are especially good candidates due to their low toxicity and high biocompatibility.4 The
synthesis of bismuth nanoparticles has been performed by using the sol-gel process,
solvothermal and hydrothermal methods as well as in microemulsions. Selected examples of
syntheses are summarized in Table 2.1.

Back in 2000, Foos et al.5 applied the inverse micelle method to produce Bi
nanoparticles with a high yield and a diameter below 10 nm. The reduction of BiOClO4 occurs
in an aqueous solution inside of AOT (dioctyl sulfosuccinate, sodium salt) reverse micelles. A
second micelle solution containing inverse micelles was composed of NaBH4 as reducing agent
dissolved in aqueous solution with the same [H2O]/[AOT] ratio. Those two micellar systems

49
were then mixed under argon under stirring, yielding Bi particles. After washing, 50 mg of
nanocrystalline Bi powder could be obtained. The particle size could be adjusted down to
ca. 3 nm by changing the [H2O]/[AOT] ratio, but the samples showed high polydispersity.

One of the most prominent colloidal synthesis pathways to monodispersed Bi

nanoparticle samples has been proposed by the group of Buhro’s group4,5. Thermal
decomposition of Bi[N(SiMe3)2]3, prepared by the reaction between Na[N(SiMe3)2]3 and Bi
salts, was performed in the presence of a polymer surfactant poly(1-hexadecene-co-1-
vinylpyrrolidinone) (PHD-co-PVP) and NaN(SiMe3)2 at 190 ℃ for 20 h. This synthesis
resulted in a dark Bi colloidal dispersion. The particle size was restricted to around 21 and
24 nm.4 In order to expand the range of diameter size, the authors modified the synthesis
protocol by tuning concentration ratios between Bi[N(SiMe3)2]3, PHD-co-PVP and
NaN(SiMe3)2, leading to a tuning of the particle size over a wide range between 3-115 nm and
with various morphologies such as Bi nanorods and nanoplates.4 The authors also applied a hot
injection method to reach a diameter ranging from 30 to 45 nm. But the synthesis procedure
still shows some limitations: 1) the necessity of an additional step to prepare the Bi[N(SiMe3)2]3
precursor, and 2) the unavailability of the stabilizing polymer PHD-co-PVP.7 BiCl3 can also be
used as Bi precursor to obtain Bi nanoparticles with size range of 4-29 nm. These particles
were proved as good catalysts for solution-liquid-solid growth of high-quality CdSe quantum
wires.5

A simpler method has been developed by Scheele et al.,8 by reducing Bi(OAc)3 with
oleylamine (OA) or trioctylphosphine (TOP) in the presence of thiols as stabilizing ligands at
lower temperature (60 °C). The resulting Bi nanoparticles were obtained as monodisperse
samples, especially for a diameter of 3.3 ± 0.5 nm. The size of the nanoparticles could be
increased from 7-9 nm to 10-14 nm by tuning the ratio between thiol and oleylamine, and could
be further increased by replacing oleylamine with TOP to get nanoparticles of 40 nm.

A stable suspension of Bi nanoparticles without toxic organic residue is one of the main
challenges for the application of Bi nanoparticles in the field of biomedicine. Solvothermal
synthesis and hydrothermal synthesis have been developed afterwards. Wang et al.,9 applied
solvothermal synthesis approach for Bi nanoparticles with different shapes. NaBiO3 was

50
reduced by ethylene glycol in presence of the capping polymer of poly(vinyl pyrrolidone)
(PVP) or a trace amount of Fe3+ at 200 ℃. By tuning the ratio of PVP and Bi, also the amount
of Fe3+, the authors could achieve different shapes for the crystals such as spheres when the
PVP/Bi ratio was 0, and nanocube when the ratio was 1.6. In presence of a trace amount of
Fe3+, the PVP/Bi ratio of 1.6 resulted in nanobelted crystals. Microwave heating could also
assist this approach.

Theoretically, hydrothermal synthesis is difficult to be applied to fabricate oxidatively

prone elements such as bismuth, which is prone to get oxidized in water. Brown et al.10
developed an aqueous synthesis of Bi nanoparticles based on a pH-dependent solubility
equilibrium. An aqueous solution of Bi(NO3)3·5H2O was reduced by NaBH4 at pH 7.4 in the
presence of dextran as a stabilizer. The nanoparticles were stable between pH 7 and pH 12.
Increasing the synthesis pH inside this range could slow down the reaction rate and lead to
more dispersed Bi nanocrystals. However, the shape of nanoparticles was not well controlled.

The main reported methods for the synthesis of Bi nanoparticles have been presented
above. In our work, we were particularly interested in obtaining bismuth nanoparticles smaller
than 40 nm, since these particles exhibit a depression of the melting point around 250 °C, lower
than the melting point of bulk Bi at 271 °C,11 which could enable a decrease of the phase change
temperature in the final metal matrix nanocomposites. The quantity of the product in each batch
is also a critical factor because these nanoparticles will serve as precursors for the synthesis of
nanocomposites, which requires more than 100 mg of nanoparticle for each lab-scale (see
chapter 3) synthesis. Solvothermal and hydrothermal methods are attractive in this respect, but
the samples are highly polydisperse and these syntheses often result in large particle size over
40 nm. We then selected the method developed by Scheele et al.8 as it could satisfy the
requirements described above.

51
 Table 2.1 Representative examples of synthesis methods for Bi nanoparticles.4

Method Reductant Stabilizing agent shape Particle size range Reference

Hydrothermal (80 ℃) N2H4·H2O Na oleate nanospheres 25-40 nm Wang, Y, 2009.12
Hydrothermal (160 ℃) N2H4·H2O PSS nanospheres 60-90 nm Yang, H., 2013.13
Hydrothermal (90 ℃) N2H4·H2O PEG nanobelts 10-50 nm Ma, D., 2015.14
Hydrothermal (90 ℃) NaH2PO2 Tartaric acid nanobelts 10-100 nm Ma, D., 2012.15
Hydrothermal (RT) NaBH4 starch nanocrystals 10-20 nm Xia, F., 2014.2
Chemical reduction in DMSO NaBH4 Sodium citrate nanocrystals 3.3 nm Velasco-Arias, D.,
(RT) 2012.16
Solvothermal (185 ℃) EG PVP, DMSO nanospheres 200-400 nm Li, J., 2009.17
Solvothermal (200-300 ℃) TEG PVP nanospheres 60 nm Kim, H.J., 2015. 18
Solvothermal (198 ℃) EG PVP, urea Microspheres 400-800 nm Chen, G., 2009. 19
nanospheres
Microwave-assisted solvothermal DEG; urea Sodium citrate nanospheres 100-600 nm Wu, J., 2010. 20
Microwave-assisted solvothermal EG EG Microspheres 45 nm-10 mm Estrada-Flores, M.,
nanospheres 2016. 21
Thermal reduction (140-180 ℃) - - nanospheres 170 nm Carotenuto, G., 2009.22
Colloidal synthesis (315 ℃) OA OA nanospheres 10-70 nm Wang, Z., 2014. 23
Colloidal synthesis (180 ℃) NaN(SiMe3)2 PHD-co-PVP in DIPB nanodots 3-115 nm Wang, F., 2008.6
Photochemical synthesis in toluene - TMSC amorphous nanoparticles 20 nm Breitwieser, D., 2015. 24
Photochemical synthesis in OA OA, n-heptane nanoparticles 10 nm Brütsch, L., 2017. 25
propylene carbonate (unspecified shape)
One-pot cementation method Fe (powder) - nanowires 60 nm width Wang, Q., 2007. 26
(150 ℃)
One-pot cementation method (RT) Zn (powder) - nanotube arrays 3-5 nm width Yang, B., 2003. 27

>1 mm length
Two-pot cementation method (RT) Zn, Al (compact MD, MA, SA nanoparticles 10-120 nm Reverberi, A., 2018.4
form) (unspecified shape)

52
 Colloidal synthesis (200 ℃) Reduction of PHD-co-PVP or OA nanoparticles 4-29 nm Wang, F., 2010. 7
Bi[N(SiMe3)2]3
by TOP
Colloidal synthesis (178 ℃) - 1-dodecanethiol nanoparticles ~22 nm Wei, B. 2016 3
Colloidal synthesis (40-60 ℃) Reduction of 1-dodecanethiol nanoparticles 7-9 nm Scheele, M. 2009 8
Bi(OAc)3 by
oleylamine/TOP
Colloidal synthesis (T = 80 ℃) TOP 1-dodecanethiol nanoparticles 6-27 nm Son, J. S., 2011 28
Colloidal synthesis (100 ℃) - HAD nanoparticles 10-24 nm Yarema, M., 201029
Colloidal synthesis (RT) NaBH4 1-decanethiol and EDTA nanoparticles 2-10 nm Marchak, D. 201330
Colloidal synthesis (315 ℃) OA OA nanoparticles 30-60 nm Wang, Z., 201331

Abbreviations:

PSS: Poly(sodium 4-styrene-sulfonate) MD: Maltodextrin

PEG: Polyethylene glycol MA: D-Mannitol

EG: Ethylene glycol SA: Saponin

PVP: Polyvinylpyrrolidone TOP: trioctylphosphine

DMSO: Dimethyl sulfoxide OA: Oleylamine

DEG: diethylene glycol (DEG) HAD: hexadecylamine

PHD-co-PVP: poly(1-hexadecene-co-1-vinylpyrrolidinone EDTA: Ethylenediaminetetraacetic acid

53
2.1.2 State of the art on the growth of silica shell to encapsulate bismuth particles

The synthesis of silica coated metallic nanoparticles has been well studied for
nanoparticles such as silver and gold.32 Among the numerous applications of metal@silica core-
shell nanoparticles, most biological applications of nanoparticles require a good control of the
surface state of nanoparticles. The implementation of a silica shell helps to graft specific
reactive functions such as antibodies or fluorophores.32 Among all synthesis methods, we
should cite two main processes: a) the Stöber process, a sol-gel process where the Si precursor
(typically tetraethyl orthosilicate (TEOS)) is hydrolyzed and condensed in an alcohol-
ammonia-water solution;33 b) the use of a reverse microemulsion before condensation.34

The growth of a silica shell on Bi nanoparticles has been less studied in the literature.
Lee et al.35 developed Bi nanotube@SiO2 core-shell structures by atomic layer deposition. Chen
et al.36 reported Bi@SiO2 nanospheres of 100 nm via in situ reduction of Bi2O3@SiO2 by
NaBH4. Luo et al.37 synthesized around 80 nm Bi@SiO2 with 5 to 20 aggregates encapsulated
in the same silica shell. The synthesis method was not described in the article. Li et al.38
synthesized 40 - 60 nm Bi nanoparticles in acidic solution at 80 °C, then the authors grew a 2 –
5 nm SiO2 layer on the Bi nanoparticles by the Stöber process. Li et al.39 succeeded to achieve
monodisperse core-shell objects. Spherical Bi nanoparticles (70–75 nm) were synthesized by
thermal reduction of Bi(NO3)3 in dodecanethiol at 178 °C. Then, a 40 nm-thick SiO2 shell was
grown by employing the reverse-microemulsion method: Bi nanoparticles were dispersed in
cyclohexane and treated with ultrasonication, then polyoxyethylene, nonylphenylether and
ammonia solution were added and stirred for 24 h.

In view of the facility of the synthesis and of the quantities of product achievable, the
Stöber process has been applied in our work.

2.2 Synthesis and characterization of Bi nanoparticles

Firstly, we applied the protocol developed by Scheele et al.8 for the synthesis of Bi
nanoparticles. This protocol consists in reducing bismuth acetate in oleylamine in the presence
of 1-dodecanethiol as ligand at 60 °C following a hot injection protocol, the experimental details
are described herein:

54
 Bismuth acetate (1 mmol) was mixed with 1- dodecanethiol (11.1 mL) and heated to
45 °C for 45 min under vacuum. The reaction flask was then purged with N2 and heated to
60 °C. Oleylamine (22.2 mL) was quickly added under stirring and the reaction mixture was
maintained at 60 °C for 24 h. Finally, Bi nanoparticles were washed twice by an
ethanol/chloroform mixture (20/1 vol./vol.) and then re-dispersed in ethanol.

The powder X-ray diffraction (XRD) patterns confirmed that Bi nanoparticles were
obtained (Figure 2.1a). TEM images (Figure 2.1b) showed that the sample consists in bismuth
nanoparticles with diameter around 25 nm with a relatively narrow size distribution. Contrary
to the literature in which nanoparticles of 12-14 nm were obtained by this process,8 the
nanoparticles we synthesized are larger and the sample is more polydisperse. We then evaluated
the impact of a range of parameters to understand this difference. In brief, the particle size
distribution did not exhibit significant dependence on the origin and purity of the reagents, on
the synthesis temperature (60 ± 2 °C), neither on the reaction duration from 20 to 24 and 28 h.
When the reaction time was decreased to 6 h, the nanoparticles (figure 2.1c) were more faceted.
The only remaining synthesis parameter is the injection speed and/or the heating ramp after the
injection. These parameters have not been described in the original article of Scheele et al.8
Then, we have attempted to preheat oleylamine at 55 °C before injection. The resulting
nanoparticles shown in figure 2.1d exhibit smaller particle size ranging from 4 to 24 nm with a
wide range of morphologies compared with the sample prepared by hot injection of room-
temperature oleylamine (figure 2.1b). Because the final nanoparticles are dedicated to the
design of phase change materials relying on the melting temperature decrease of nano-sized
bismuth, nanoparticles smaller than 30 nm are sufficient for this study as this size range
corresponds to a noticeable decrease of the melting temperature.40 Thus, we selected
nanoparticles of 25 nm diameter (figure 2.1b) from room temperature oleylamine injection for
the following steps.

55
Figure 2.1. (a) Powder XRD pattern and (b-d) TEM images of Bi nanoparticles synthesized (b)
at 60 °C for 24h from room-temperature oleylamine injection, (c) at 60 °C for 6h from room-
temperature oleylamine injection, the low contrast part around nanoparticles are the organic
residue. (d) at 60 °C for 24h from 55 °C oleylamine injection.

2.3 Synthesis and characterization of Bi@SiO2 nanoparticles

In a second step, the bismuth nanoparticles washed with 2 centrifugation-dispersion

cycles with ethanol and chloroform and were coated by a thick silica shell via the Stöber
method.33 The protocol is as follows:

Tetraethyl orthosilicate (TEOS) was dissolved in absolute ethanol at a concentration of

0.38 mol L-1. 10 μL of this diluted TEOS solution and 100 μL of ammonia (30%) were
subsequently added to 1 mL ethanolic bismuth suspension at [Bi] = 6.10-4 mol.L-1 under
vigorous stirring. The mixture was heated to 40 °C for 3 hours. After cooling down, it was
washed two times with ethanol.

56
 The silica shell obtained is clearly observed on TEM images, with a lower contrast than
that for the Bi core (Figure 2.2a). Its thickness was around 7 nm. TEM (Figure 2.2a) shows that
the silica shell actually embeds several particles, forming aggregates with size up to 160 nm.
Since the shape of aggregates is irregular, here we characterize its size with the area distribution
(Figure 2.2c). The area varies from ~1500 nm2 to ~20000 nm2. The area of a single core
nanoparticle is ~ 1500 nm2, and the area of the largest object observed correspond to the
14 times the size of single object. Since nanoparticles partially overlap in TEM image, the
largest objects possess ca. 20 nanoparticles inside the same silica shell.

Figure 2.2. (a, b) TEM images of Bi@SiO2 nanoparticles with silica shells prepared by the
Stöber method. The Bi particles in (a) were washed 2 times before encapsulation. The
corresponding size distribution is shown in (c). The Bi particles in (b) were washed 5 times
before encapsulation.

57
 This aggregation phenomenon is usual for the encapsulation by the Stöber process of
poorly stable ethanolic suspensions of small nanoparticles.34 The number of washing cycles of
the initial particles strongly influences the size of the final objects as well. With subsequent
centrifugation and washing steps, nanoparticles are more aggregated and yield Bi@silica
aggregates larger than 1 micrometre after 5 washing steps (figure 2.2b). One explanation could
be the removal of the stabilizing ligands on the surface of nanoparticles after washing. This
phenomenon has been well described in the literature for gold nanoparticles,41 in which repeated
washing removed the ligands, yielding irreversible aggregation. As repeated washing results in
aggregation of nanoparticles, no more than two washing steps were applied in order to preserve
the suspension’s stability in ethanol. We then attribute the formation of Bi@SiO2 aggregates
during silica growth to the low dispersibility of Bi nanoparticles in ethanol. By taking
inspiration from the case of gold nanoparticles,42,43,32 we applied ligand exchange to incorporate
a functionalized thiol molecule that would ensure higher dispersibility of the particles in ethanol
in the ligand monolayer. The ligand exchange protocol is described below:

After the synthesis of Bi nanoparticles and two washing cycles with ethanol and
chloroform, the particles were redispersed in ethanol. A given quantity of ligand, varying from
0 to 5 ligand per Bi surface unit (nm2), was solubilized in ethanol and was added into the
reaction medium, followed by heating to 40 °C for 1 h in a closed vial in an oven.

At the end of their synthesis, the pristine Bi nanoparticles were covered by 1-

dodecanethiol as ligand. Mercaptoundecanoic acid (MUA) and mercaptohexadecanoic acid
(MHA) (Table 2.2) were assessed to perform ligand exchange. The thiol group is expected to
graft on the surface of the nanoparticles and the acid group is expected to improve the dispersion
ability in ethanol via hydrogen bonding. Different lengths of the spacer chain between MUA
and MHA were used to tune the hydrophobicity of the ligand and of the nanoparticles.

Table 2.2 Molecular structure of MUA and MHA

MUA MHA

58
 The quantity of ligands is also inspired from the case of gold nanoparticles. The total
coverage ratio of thiol on gold surface was 5 thiol·nm−2.43 Series of experiences have been
performed to study the influence of the exchanged ligand on the aggregation state of Bi cores
within the silica shell/matrix. Figure 2.3 shows the similarity of the dispersibility of samples
with or without MUA. Therefore, MUA does not appear as an efficient molecule to trigger
ligand exchange and/or dispersion of the Bi particles.

Figure 2.3. TEM images of Bi@SiO2 nanoparticles from Bi nanoparticles washed during 2
cycles before further silica coating: (a) without ligand exchange; (b,c) ligand exchange with
1.5 MUA nm-2 (d) ligand exchange with 3 MUA nm-2 (e) ligand exchange with 5 MUA nm-2.

In order to verify the efficiency of ligand exchange before growth of the silica shell, we
observed the bismuth nanoparticles via TEM just after ligand exchange (in ethanol) and then
after redispersion in an apolar solvent (hexane). Here, we should notice that the TEM grids
were prepared by drop casting of nanoparticles’ suspensions onto carbon-coated copper grids,
which triggers further aggregation during evaporation. Nonetheless, we could not detect
significant differences in terms of dispersibility between bare particles and ligand exchanged
ones (figure 2.4a-c): aggregates of around 15 nanoparticles down to single nanoparticles were
both observed in each sample. The effect of ligand exchange was not significant in ethanolic

59
suspensions. After redispersion in hexane, we observed aggregates around 200 nm, 400 nm and
even larger than 800 nm for the samples without ligand exchange, exchanged with MUA and
exchanged with MHA, respectively (figure 2.4d-f). The increased aggregation state with
exchange of ligands indicated that ligand exchange actually occurred, but the increase of the
hydrophobicity of the ligand didn’t improve the dispersion in hexane, which could be ascribed
as the influence of organic residuals as observed in TEM images. The aggregation state
observed from the pristine ligand-exchanged Bi nanoparticles was maintained after growth of
the silica shell (figure 2.5). The difficulty to reach a stable suspension of Bi nanoparticles is
then the bottleneck that hinders isolation of single Bi@SiO2 nanoparticles. This could arise
from the reaction medium’s composition (e.g. side product residuals from the Bi nanoparticles
synthesis) and from intermolecular interactions between thiols, thiolates and Bi-thiolate
complexes, which probably show differences in complexing abilities on different metal surface.

Figure 2.4 Influence of ligand exchange on the dispersibility of bismuth nanoparticles. (a-c)
TEM images of Bi nanoparticles after two washing cycles and redispersion in ethanol: (a)
without ligand exchange; after ligand exchange with (b) MUA and (c) MHA. (d-f) TEM images
of corresponding nanoparticles shown in (a-c), centrifuged at 2000 rpm and redispersed in
hexane: (d) without ligand exchange; after ligand exchange with (e) MUA and (f) MHA before
redispersion. Low contrast layers around nanoparticles correspond to the organic residuals
from the synthesis.
60
Figure 2.5 TEM images of Bi@SiO2 samples for initial Bi nanoparticles after two washing
cycles and coated with silica (a) without ligand exchange, with ligand exchange by (b) MUA
and (c) MHA before silica encapsulation.

2.4 Conclusion

In this work, we succeeded in synthesizing core-shell Bi@SiO2 nanoparticles with

25 nm Bi core and 7 nm silica shell. Clusters of several particles are embedded in the silica,
forming aggregates of size up to 160 nm. By tuning the number of washing cycles of the pristine
Bi nanoparticles, the most dispersed samples contain 1 to 15 nanoparticles that are encapsulated
in the same silica shell. Single Bi nanoparticles encapsulated in silica shells have already been
achieved for particles larger than 40 nm by the reverse-microemulsion method39 and by acidic
solution synthesis followed by Stöber method.38 The encapsulation of Bi particles smaller than
30 nm by silica is still a challenge. Compared to previous results on silica-encapsulated Bi
nanoparticles with particle size below 30 nm, which showed aggregates of 5 to 25
nanoparticles37 and more than 20 nanoparticles36 encapsulated in the same silica shell, our
protocol shows enhanced dispersion of the Bi cores, although we could not obtain only single
cores. Nevertheless, these nanoparticles meet the requirements described in the introduction:
the particle size is small enough to trigger a decrease in the melting temperature and the
synthesis was successfully scaled-up to 80 mg of product per batch. Another unexpected
advantage is that the large size of the silica aggregates could facilitate the identification of the
reaction between silica and aluminium. This part will be discussed in detail in chapter 4.

61
 We have developed a two-step synthesis combining colloidal synthesis of Bi and Stöber
method. The ligand exchange with two well-known ligands for gold system is not efficient in
the case of bismuth particles. The dispersibility after the exchange of ligands is a complex
phenomenon, which depends greatly on the experimental conditions. Several mechanisms are
in competition between surface binding species like thiols, thiolates, Bi-thiolate complexes.
External parameters such as solvent or atmosphere also influence the exchange, as well as the
terminal functional group and the chain length. In the future, simulations could be carried out
to further understand surface binding and then identify more suitable ligands. The reverse
microemulsion method could also be a promising direction. The aqueous droplets in oil are
uniform in size and usually smaller than 100 nm. They could act as nanoreactors and produce
narrow-size distributions of core-shell nanoparticles.34

62
 Chapter 3
Metal matrix nanocomposites with
improved mechanical properties

63
64
3.1 Introduction

One of the key challenges in fabricating metal matrix nanocomposites by liquid metallurgy
processes lies in the harsh conditions of the nanoparticles feeding and dispersion process at high
temperature (over solidus or liquidus of aluminum alloys, generally over 550 °C). Indeed, the
stability of nanoparticles at high temperature should always be kept in mind in order to obtain
the final material. In the past decades, classic aluminum matrix nanocomposites have relied on
commercial nanoparticles with a limited choice.1 Recent progress in the in situ growth of oxide,
boride and carbide nanoparticles has shown some promises in the increase of nanoparticle
loading and in the interfacial bonding between nanoparticles and metal matrix.2,3,4 In particular,
the reaction of K2TiF6 amd KBF4 in molten mixtures of aluminum and fluoride salts enabled
the in situ formation of TiB2 nanoparticles in different aluminum alloys.4 However, two main
concerns still exist: one is related to the difficulty to control the particle size and the other one
lies in the presence of impurities in the final products.4,5,6,7,8 The continuous development of
nanoparticle synthesis approaches, especially the emergence of molten salts synthesis offers an
alternative solution.9 Several classes of nanoparticles including borides, nitrides, carbides and
silicides have been synthesized using molten salt solvents.10 The use of inorganic molten salts
as high temperature solvents (300 to 900 °C) could lead to nanoparticles that are stable at the
processing temperature of liquid metallurgy. As-synthesized nanoparticles possess well-
controlled particle size and they could be easily dispersed in water or alcohols after several
washing. All these developments taken together, nanoparticles prepared in molten salts appear
as interesting potential precursors to create new metal matrix nanocomposites. Nevertheless,
there are still several issues to overcome to reach this aim, such as the demixation of
nanoparticles powders from the molten matrix during their incorporation and during
solidification.

In this work, we aimed at designing boride-aluminum nanocomposites in the presence

of 99.9% pure aluminum and of an aluminum alloy AlSi7MgCu (Table 3.1) by synthesizing ex
situ the nanoparticles. Boride nanoparticles are hard and light-weight and are promising
candidates to further reinforce light-weight alloys. In this chapter, we attempt to establish the
suitability of boride nanoparticles synthesized in molten salts as a potential reinforcement for
aluminum-based matrices. HfB2 was selected as a target because it is a superhard material, with

65
a predicted Vickers hardness of 30.5 GPa,11 It also contains hafnium, which is expected to
facilitate the electronic microscopy observations and chemical detection in the final composites
and will then be a useful probe to reach a proof-of-concept material. The dispersion behaviour
of the nanoparticles in molten aluminum and its effect on the enhancement of mechanical
properties are evaluated. The results are then compared with those obtained for reference SiC-
based nanocomposites, a well studied aluminum-based matrix system.

Noticeably, at the time when we started this project, there were only two articles
available in the literature that used molten salt synthesized nanoparticles (particles size between
10 to 100 nm) as external reinforcement.12 But other related works have been published
recently, especially from Li Group13 who worked on the incorporation of molten-salt
synthesized TiB2 and ZrB2 nanoparticles in aluminum. However, none of them have applied
hafnium boride nanoparticles as reinforcement so far.

Alloying elements of alloys for cast parts

Alloy Composition tolerances (contents in %)
Si Fe Cu Mn Mg Ni Zn Ti Other
elements
AlSi7MgCu 6.5 <0.2 0.4 <0.1 0.28 < 0.05 < 0.1 0.08 < 0.1

7.5 0.6 0.4 0.20

Table 3.1 Aluminum alloy AlSi7Mg0.3Cu0.5 composition14

3.2 Materials and methods

3.2.1 Preparation of nanoparticles

HfB2 nanoparticles were prepared in molten salts as previously reported.15 0.5 mmol of hafnium
(IV) chloride, 1 mmol of sodium borohydride and 2.5 g of eutectic mixture LiCl/KCl powders
were ground together with a Retsch MM400 ball miller for 2 minutes at 20 Hz. The reaction
medium was heated in a carbon crucible under argon flow at 900 °C for 4 h. After cooling down
to room temperature, the nanoparticles were washed with 12 cycles of centrifugation-
redispersion in water.

66
 The synthesis of amorphous boron nanoparticles was similar to HfB2 nanoparticles. 6
mmol of sodium borohydride was grounded with 5 g of eutectic LiI/KI. The medium was heated
at 800 °C for 1 h. SiC nanoparticles with a diameter of 35 nm were purchased from Nanomaker,
France.

3.2.2 Preparation of nanocomposites

HfB2-Al and SiC-Al nanocomposites were prepared by ultrasound-assisted casting of a mixture

composed of nanoparticles and aluminum. Two sets of experiments were designed. The first
approach involved a direct introduction of nanoparticles into the molten metal. The metal matrix
was heated at the desired temperature for 5 minutes. The desired amount of nanoparticles
powder was packed in an aluminum foil and dipped into the molten metal at 720 °C. A steel
spatula was used to stir the melt for 2 minutes. Then, a high-power ultrasonic probe made of
niobium with tip diameter of 4 mm (QSonica sonicator Q500A-220, USA) was applied to
disperse the nanoparticles for 5 minutes, with an interval of 5 s on / 5 s off at 30% of the nominal
power. Finally, the sample was cast in an alumina crucible. For characterization, the middle
part of the sample was cut and analyzed.

In the second approach, 2 g of aluminum powders (325 mesh, 99.99%) were added into
an suspension containing the desired quantity of nanoparticles powders into 2 mL of ethanol,
dispersed by sonication at 80 °C as setpoint for 1 hour. Then the mixture was dried in a vacuum
oven at 40 °C for 1 h, followed by pressing into a pellet under 8 tons for 2 minutes.
Subsequently, the pellet and 1 g of pure Al beads were heated by an induction oven at 720 °C
under argon atmosphere in an alumina crucible. After dwelling for 1 min, ultrasounds were
applied as described above to disperse nanoparticles for 5 minutes, with an interval of 5 s on /
5 s off. The final composites were obtained by cooling down the melt in the crucible. They were
then cut by diamond wire saw for further analyses.

3.2.3 X-ray diffraction

X-ray diffraction was performed with a D8 Bruker diffractometer operating with CuKɑ
radiation in Bragg-Brentano configuration. The ICSD reference card 30422 was used for the
indexation of the HfB2 structure.

67
3.2.4 Transmission electron microscopy

For transmission electron microscopy (TEM), the nanoparticles were dispersed in ethanol, then
one drop was evaporated on carbon-coated copper grids. For the prepared composites, TEM
thin foils were prepared by a rectangle of 3*3*0.5 mm cut in the middle of the sample. The
polishing step was then performed by diamond disc to get a 200 μm thick sample and the
thickness of the centre was further reduced to 50 μm by using a precision dimpling instrument
(Model 515, South Bay Technology). Ion milling was performed for the final thinning. TEM
images were recorded on a Tecnai spirit G2 microscope operating at 120kV. The STEM images
and STEM-EDX mapping were recorded with a JEM 2100Plus UHR microscope, and were
both performed at 200 kV. The size distribution of nanoparticles was determined by counting
150 to 200 nanoparticles from TEM images.

3.2.5 Scanning electron microscopy

The microstructure of the nanocomposites was characterized by scanning electron microscopy

(SEM). The samples were mounted in polyfast resin (Struers), followed by a series of polishing
steps through 1000, 2400, 4800 grit papers (Buehler) then with 3 µm and 1 µm diamond
suspensions. For SEM observations, the samples were directly attached on a SEM holder using
carbon tape. They were then observed under scanning electron microscopy (SEM, Hitachi S-
3400N and Hitachi SU-70 microscope) equipped with an Oxford X-Max 50 mm2 EDS detector.
Image analysis was performed with the Fiji software. The grain size distribution was measured
by ImageJ from optical microscope images and SEM images.

3.2.7 Dynamic Light Scattering

The average hydrodynamic diameter of HfB2 aggregates dispersed in water was determined by
dynamic light scattering (DLS) in the Zetasizer Nano ZS90 (Malvern Panalytical).

3.2.7 Microhardness tests and nanoindentation tests

The Vickers hardness was measured with a load of 0.05 kg, dwell time 10 s using Wilson
VH1102 microhardness testers.

68
 The nanoindentation measurements were realized using a Nano Indenter XP® equipped
with a Berkovich tip. The hardness was measured at the unloading process using Oliver and
Pharr method. 400 measurements were achieved on a matrix of 21*19 with a distance of 10 µm
at a depth of 1 µm to obtain a mapping of mechanical properties. The distance between two
scratches was 10 µm. A matrix of 6*5 with a depth of 3 µm was performed to obtain an average
measurement of mechanical properties.

3.3 Results and discussion

3.3.1 Preparation and characterizations of nanoparticles

HfB2 nanoparticles were prepared by molten salts synthesis.15 Different contrasts observed by
TEM (figure 3.1a) confirm the core-shell structure of the nanoparticles. TEM shows a diameter
of the contrasted core of 4 to 14 nm. This core shows high contrast, which is ascribed to a
crystalline character and to a higher electron density, due to the Hf content. The powder XRD
pattern (figure 3.1b) is fully indexed along the hexagonal HfB2 structure. The crystallite size of
HfB2 calculated according to the Scherrer formula is 8 nm, in agreement with TEM, thus
indicating that the core is single crystalline and made of HfB2. A less contrasted layer of ca.
3 nm shown covers the surface of HfB2 nanoparticles, corresponding to a layer of amorphous
boron as described in a previous study.15 The diameter measured by DLS on an aqueous
dispersion is about 400 nm, which is much larger than the TEM data, indicating the formation
of aggregates. These objects are clearly observed by TEM, which indicates that the amorphous
shell embeds several particles. According to X-ray photoelectron spectroscopy (XPS) published
in a previous work,15 the boron surface is partially oxidized for the nanoparticles washed with
water.

69
Figure 3.1 TEM images (left) and powder XRD patterns (right) of (a,b) HfB2 nanoparticles
coated with an amorphous boron-rich layer, as shown between two arrows; (c,d) of commercial
SiC nanoparticles, respectively.

Commercial SiC nanoparticles were purchased from Nanomakers. The particle size
ranges from 20 to 50 nm, with an average diameter of 35 nm. Defects can be observed inside
the commercial nanoparticles (figure 3.1-c). The powder XRD pattern (figure 3.1d) confirms
that SiC is the only crystalline phase.

3.3.2 Preparation of metal matrix nanocomposites

To overcome the difficulties of directly loading the nanoparticles powders in the melt (see
Appendix S3.1), which are mainly ascribed to the poor wettability and high volume of
nanoparticles in powder form, we used a two-step process to introduce nanoparticles in the melt:
a) dispersing the nanoparticles in ethanol in the presence of 325 mesh aluminum micro-
powder, then pressing the slurry into a pellet;
b) diluting the pellet in molten aluminum under ultrasonication.
Pressing of the pellet at room temperature without heating prevented the coalescence of the
nanoparticles and excluded the potential reactivity between Al micrograins and the boron shell
of the nanoparticles before the dispersion process.12
70
 The experimental setup is shown in figure 3.2a. Ultrasonic waves generate cavitation
and acoustic streaming, which is most likely to refine the microstructure, break aggregates and
homogenize the distribution of nanoparticles.16 However, most commercial ultrasound
microprobes for < 5mL volume are made by Ti alloy. This material is not applicable in our case
because of the erosion of the Ti rod in molten aluminum17 and the reactivity with the
nanoparticles, especially their boron shell. A niobium tip of 4 mm diameter has been specially
designed by the supplier company to limit the contamination of the samples (figure 3.2b). We
further designed the shape of the crucible, as shown in figure 3.2c, to increase the dipping depth
of the ultrasonic tip. The tip was dipped in 2 mL of molten aluminum at a depth of 3 mm,
corresponding to 60% of the depth of liquid Al. This set-up is suitable for preliminary studies
without the need to scale up the synthesis of nanoparticles.

Figure 3.2 Experimental setup for incorporating nanoparticles into a molten metal by coupling
induction heating and high temperature ultrasonication. a) reaction system, b) sonicator with
Nb probe, c) alumina crucible.
71
 Furthermore, the ultrasound treatment power and intervals have also been tuned.
Previous works have shown that the mechanical properties of composites prepared by
discontinuous ultrasonic treatments were better enhanced than with the continuous ultrasonic
treatments due to enhanced dispersion.16 In our study, an interval of 5s on/ 5s off was applied
for all experiments. Considering the liquid splash at high power and pore formation at
intermediate power, we chose to work at an intermediate ultrasound power at 30% of the
maximum power followed by 30 s degassing with a low power of 10%. These conditions were
then set as a model for the following experiments.

More details for the improvement of the process can be found in appendix S3.1.

3.3.3 Microscopy analysis of HfB2 and SiC nanocomposites

HfB2-Al nanocomposites

Several experiments were performed by dispersing 0.8 vol.% HfB2-Al nanoparticles into
aluminum using conditions defined in 3.3.2. The experimental parameters are presented in
table 3.2 (P83). Broad peaks from XRD patterns (Appendix 3.5) confirmed that the phase was
maintained. Low intensity peaks at 28° and 41° might be indexed along the AlB12 phase, which
may arise from the reaction between molten aluminum and the boron-rich shell of the
nanoparticles.

These samples were further analyzed by electron microscopy coupled to EDS.

Figures 3.3a,b show SEM images of HfB2-Al nanocomposites prepared in a resistive furnace,
with ultrasound treatment through a Nb tip for 5 minutes, followed by cooling down in liquid
nitrogen. Aggregation of HfB2 particles was observed along the Al grain boundaries of the final
metal matrix rather than a uniform distribution within the matrix. EDS analysis measurements
(Figure 3.3c and d) yield a heterogeneous composition between the boundaries and the grains,
which indicated that HfB2 nanoparticles were segregated towards grain boundaries.

72
Figure 3.3. Characterization of a 0.8 vol% HfB2-Al nanocomposite prepared with ultrasound
treatment by Nb tip for 5 minutes and cooled down in the liquid nitrogen. (a)Optical image and
size distribution of Al grains. SEM image (b) and EDS analysis of a grain boundary (c,
corresponding to the point of spectrum 1 in b) and inside the Al grain (d, corresponding to the
point of spectrum 2 in b). Dark and grey contrasts correspond to aluminum and to aggregates
of nanoparticles, respectively.

TEM images (figure 3.4) also showed that nanoparticles formed into sub-micron
aggregates at the Al grain boundaries and no separated nanoparticles could be found in Al grain.
Compared to the initial nanoparticle precursors, HfB2 nanoparticles could maintain their size
after the harsh melting process at high temperature (figure 3.4a) although their morphology
slightly changed from sphere-like particles to faceted particles (figure 3.4b). The HfB2 crystal
structure was confirmed according to the lattice spacing observed by HRTEM, also indicating
the thermal stability of HfB2 nanoparticles.

73
Figure 3.4. (a,b) TEM images of a 0.8 vol% HfB2-Al nanocomposite prepared by 5 min
sonication in a resistive furnace followed by liquid N2 quenching. Dark and light parts
correspond to HfB2 nanoparticles and to the aluminum matrix, respectively. (c-e) STEM-EDS
mapping showing the overlap of Hf (d) and (e) B elements in the aggregates

The state of the interface between aluminum and nanoparticles could reveal the reactions
and events during nanoparticle composite synthesis. We have focused on interfaces between
crystalline HfB2 nanoparticles and the Al matrix as was observed in figure 3.4b. Three
relationships have been observed at interface between HfB2 nanoparticles and Al:
a) (010)HfB2//(111)Al as shown by blue lines. In fact, these plane presents the best match
(d=0.26 nm) with (111)Al, which should promote the epitaxial crystallization of the Al grain.
b) A distortion of lattice fringes extending up to 1 nm from the interfaces (green arrow).
The mismatch between other HfB2 facets and Al lattice planes, except for the plane couple
described above, could disturb crystalline order of Al near the interface, and thus cause Al
lattice distortions near the interface.

74
 The combination of possible epitaxial relationship (a) and lattice distortion (b) implies
that the HfB2-Al interface could be semi-coherent, which is expected to favour the mechanical
properties.18 The modification of the shape of the nanoparticles implies also that the particles
undergo facet reconstruction, favouriting (010) facets epitaxially oriented versus (100)Al planes.
c) Amorphous-like defects. (yellow arrow, within yellow dashes) This could be
attributed to the reactivity of the matrix with the ca. 2 nm thick amorphous boron-rich layer
exhibited by the initial nanoparticles. The size range of this area at the interface between HfB2
and Al in figure 3.4b is consistent with the thickness of the pristine layer. According to the Al-
B phase diagram (Appendix 3.6),19 AlB2 and AlB12 could form above 500°C. Although the
kinetics of the reaction between nanoscaled boron (in the form of the nanoparticles’ shell) and
molten Al are unknown, a previous study showed that AlB2 AlB10 and AlN form at the interface
between BN nanotubes and aluminum after 10 minutes at 650°C,20 which is corresponding to
conditions similar to those used in our experiment. The Gibbs free energy for the formation of
aluminum boride from Al and boron is similar to the one for the reaction described in the
literature,20 so the presence of crystalline aluminum boride is highly possible in our sample. No
fringe is observed, probably because this phase is not well oriented.

The identification of the nature of the boron-rich shell is necessary to differentiate boron
from the Al matrix. It is a strong challenge as boron cannot show contrast differences in the
presence of Al. We performed high resolution TEM in a region where the Al matrix was not
well oriented: Al fringes cannot be observed (Appendix 3.7). A crystalline layer around 2 nm
shows lower contrast compared with the 4 nm dark core. The HRTEM micrograph cannot be
indexed satisfactorily along structures such as Al, Al2O3, Hf, HfO2, HfB2, HfB, AlB2, AlB10,
but could be attributed to AlB12, in agreement with XRD (Appendix 3.5).

SiC-Al nanocomposites

A 0.8 vol% SiC-Al nanocomposite was obtained via the approach detailed in the previous
section. It showed results similar to HfB2-Al nanocomposites: nanoparticles mostly segregated
at the grain boundaries according to SEM images (figure 3.5a-e). Figure 3.5f shows a TEM
image of a SiC nanoparticle in a 200 nm Al grain. This indicates that part of the SiC particles

75
are located inside the grains, witch is according to the previous literature,16 and contrary to our
TEM observations for HfB2-Al composites.

In fact, one of the main challenges for the preparation of metal matrix nanocomposites
is that nanoparticles tend to segregate towards the grain boundaries at the solidification fronts
upon cooling down. This phenomenon can be explained by several reasons: a) in both SiC and
HfB2 samples, nanoparticles form aggregates in the solid state due to attractive Van der Waals
forces, while the applied ultrasound cavitation and acoustic streaming in molten metal is not
sufficient to disperse aggregates into single particles. HfB2 are also initially located into a
boron-rich layer that embeds irreversibly several particles together. b) during solidification,
attractive Van der Waals forces and interfacial forces overcome the Brownian motion and
viscous drag force,21 leading to particle aggregates, c) SiC and Al have a rather large
equilibrium lattice mismatch,22 which is detrimental to interfacial stability and energetically
disfavors an homogeneous dispersion, d) The presence of a passivation oxide layer onto Al, as
shown by EDS mapping, may also impact the wetting behaviour of Al on SiC.

Previous works have been performed to improve the dispersion of nanoparticles within
SiC-Al alloy nanocomposites.23,24,25 A better dispersion has been achieved by Lu et al., where
diluted granules are prepared by powder metallurgy and dispersed into a molten Al alloy by
ultrasonication. Although micro-sized clusters were observed, nanoparticles were not totally
pushed to the grain boundaries.21 An even better dispersed nanocomposite could be achieved
by changing the composition of the matrix to tune the wettability and the rheology, such as for
the dispersion of SiC in Mg-Zn alloys.26

76
Figure 3.5. Microscopic images of a 0.8 vol% SiC-Al nanocomposite prepared with ultrasound
treatment by Nb tip for 5 minutes and cooled down in liquid nitrogen. (a) SEM micrograph, (b-
d) SEM-EDS mapping, (e) Optical microscopy image of the composite (inset shows the grain
size distribution), (f) TEM image of a SiC nanocrystal in the composite.

Compared to HfB2/Al composites, SiC-Al composites exhibited a larger Al grain size

(figure 3.3a and 3.5e). This result suggests that HfB2 creates more effectively heterogeneous
nucleation sites of aluminum grains during cooling and solidification, which could be ascribed

77
to the local structural rearrangements caused by the semi-coherence of the interface between
HfB2 and Al and to the AlB12 interphase compared with the misfit between Al and SiC.
Nanoparticles in HfB2-Al composites are nonetheless more segregated than that in the SiC-Al
composites, which could be ascribed to the strongly aggregated initial state of HfB2
nanoparticles.

HfB2-AlSi7Mg0.3Cu0.5 nanocomposites

We have further assessed the possibility to prepare light weight matrix composites with the
AlSi7Mg0.3Cu0.5 foundry alloy. Considering the difficulties to decipher Si arising from the alloy
from the one that may arise from incorporated SiC nanoparticles, only HfB2-based
nanocomposites were realized. A pellet consisting of pure Al and nanoparticles was prepared
and incorporated in molten AlSi7Mg0.3Cu0.5 at 600 °C. The melting of the pellet took around
5 minutes. Then, ultrasound treatment was applied for 5 minutes with an Nb tip. The
solidification of the material was carried out using liquid nitrogen quenching.

The optical microscopy images (figure 3.6a) highlight the Si phase in dark grey colour.
SEM-EDS mapping (figure 3.6b-g) suggests that hafnium is located in the grain boundaries
rich in Si phase. According to the precedent results, it is not surprising that nanoparticles were
segregated to the grain boundaries.

We also investigated the effects of the nanoparticles on the structure of the metal matrix
during solidification. A pure aluminum pellet was added to AlSi7Mg0.3Cu0.5 alloy and prepared
by the same method (Appendix 3.9). Comparing the sample without treatment (Appendix 3.10)
and the sample treated by ultrasounds (Appendix 3.9), we observed globular grain structures
after ultrasound treatment. This phenomenon is in accordance with previous reports.1,16,27 When
the nanoparticles were incorporated to the alloy, the microstructure of the alloy was slightly
refined, with Al grains of an average size of 90 μm. The globular shape was also modified.
However, this behavior is less marked than with HfB2-Al samples. As a matter of fact, the solute
present in the alloy is responsible of constitutional undercooling during solidification, which
plays a primary role in decreasing the size of the grains. The grain size in the alloy is then
dominated by the composition of the alloy.

78
Figure 3.6. Microscopy observations for a 0.8 vol% HfB2-AlSi7Mg0.3Cu0.5 nanocomposite
prepared with ultrasound treatment by Nb tip for 5 minutes and cooled down in liquid nitrogen.
(a) Optical microscopy image and corresponding size distribution of Al grains, (b) SEM
micrograph and (c-g) corresponding SEM-EDS maps

3.3.4 Decrease of the matrix grain size

Influence of different cooling rates on the microstructure of HfB2 and SiC nanocomposites

As described earlier, wettability and viscous drag force are the key factors that may maintain
the nanoparticles dispersed in the metal grains and enable them to resist the solidification
front.28 However, the control of the freezing conditions has also an influence on the
microstructure. Increasing the cooling rate can form a refined network of nanoparticles by
quenching the samples, thus leading to high modulus for the metal matrix grains.29 Therefore,
the microstructure was assessed for 0.8 vol% nanoparticles-Al nanocomposites obtained by
three different cooling methods:

79
 a) For samples heated in the resistive furnace, metallic samples are always inside an
alumina crucible during cooling. Relatively slow cooling is achieved by air cooling of the quartz
tube. The solidification time is 2 minutes from 720°C to 660°C. The cooling from operating
temperature to room temperature takes 2 hours (around 6°C min-1).

b) For samples heated in the induction furnace, only the metallic melts are heated, the
crucible is only heated by contact with the Al melt. As the heating time is short (10 minutes),
we expect that the crucible is colder than in the case of the resistive furnace and the cooling rate
is then higher. The molten metal returned to the solid state in around 10 s after heating. The
cooling from operating temperature to room temperature takes approximately 40 minutes
(around 18°C min-1).

c) Another series of samples was obtained by resistive heating and cooling relatively
quickly by immersion in a liquid nitrogen bath. The sample was cooled to the temperature of
liquid nitrogen (no more bubble appears) in 10 minutes, with a cooling rate of around
100 °C min-1.

Firstly, the homogeneity of the microstructure was investigated. In conventional air

freezing, the microstructure of the sample on the top, middle and bottom section was
homogeneous. Nucleation occurs homogeneously inside the sample and Al grains grew in an
anisotropic fashion. For the samples cooled by liquid nitrogen, presented in Appendix 3.11, the
microstructure at the core of the sample remains homogeneous, but the top and the bottom areas
show completely different morphologies. The homogeneous core might be due to the presence
of an important temperature gradient between the core and the surface of the sample. It is
accepted that liquid on the surface solidifies almost immediately when the sample is plunged
in liquid nitrogen. Top view images showed large grains around 60 μm. In this region, the liquid
metal was in contact with the alumina crucible, which is a low heat conductor. Fewer nucleation
sites are formed at the low cooling rate region. The nucleated grains grow into a macroscopic
grain during solidification. Thus, the 100 μm layer on the surface and in contact with the
crucible are not considered for further analysis.

Analysis of the average grain size reveals a refinement of the microstructure with the
increase of the heating rate. Indeed, 0.8 vol% HfB2-Al samples cooled at different rates showed

80
an average grain size of 120, 78 and 20 µm for cooling rates 6, 18 and 100°C min-1, respectively.
(figure 3.7) SiC-Al nanocomposites also follow the same trend. Liquid nitrogen cooling is
therefore proven to be the most efficient among the three methods assessed to refine the
microstructure of MMnCs.

Figure 3.7. 0.8 vol% HfB2/Al prepared with a cooling rate of (a) 6, (b)18 and (c) 100 °C min-1

A common cooling method in the industry is water quenching. The difference between
water quenching and liquid N2 quenching has been studied in the course of this PhD work
(appendix 3.12). Regarding the average grain size for different cooling methods, liquid nitrogen
quenching showed similar grain refinement behavior (82 µm with natural cooling, 48 µm for
water cooling and 41 µm for liquid nitrogen cooling). Considering the wide distribution of grain
size, the difference between these two cooling methods was not significant. Even though the
temperature of liquid nitrogen is much lower than that of water, the formation of N2 bubbles at
the surface of hot metal inhibits heat transfer and then decrease the cooling ability of liquid
nitrogen, thus explaining the similar microstructure of water and N2-cooled samples. Water
cooling and liquid nitrogen cooling could both be applied with variable sets of parameters.

Influence of the inclusions load on the microstructure of nanocomposites

Aiming at improving the dispersion of nanoparticles, we compared a series of composites

prepared at identical conditions (5 minutes ultrasound treatment by Nb tip, cooling by liquid
nitrogen) but with different inclusion loads.

Optical microscopy images (Figure 3.8) indicated a surprising increase of grain size with
the inclusion load. In line with previous SiC-Al nanocomposites without extra dispersion
methods,16 a volume ratio of 1.5 % is hard to achieve. This phenomenon might be explained by
81
macro-segregation observed e.g. in the samples at 2.4 vol% of inclusions. This phenomenon
could relate to the viscosity and its impact on the dynamics of the solidification front during the
solidification process. Therefore, a ratio of 0.8 vol% was kept as a model.

Figure 3.8. Optical microscopy images of 2.4 vol% SiC-Al (a) and 0.8 vol% SiC-Al

3.3.5 Mechanical tests

Microhardness measurements

The mechanical properties were firstly investigated by microhardness measurements and then
compared to the pure reference samples prepared in the same way (table 3.2). For all
experiments, the hardness at grain boundaries (rich in nanoparticles) was much higher than the
value within the grain, as expected from the analysis of microstructure. Considering the
inhomogeneity, an average of measurements on grain boundaries and inside the grains was also
calculated.

82
 Table 3.2. Microhardness tests for aluminum nanocomposites

Sample Ultrasound Cooling Average Average Average standard

treatment method hardness Al grains grain deviation
time (HV (HV 0.05) boundaries
0.05) (HV 0.05) (HV
0.05)
Al 5 minutes Air, in 24.9 / / 1.3
crucible
0.8vol% HfB2-Al 5 minutes Air, in 40.4 25.7 69.3 18.1
crucible
2.4vol% SiC-Al 5 minutes Air, in 45.9 32.9 78.9 22.2
crucible
0.8vol% HfB2-Al 5 minutes Liquid 34.9 29.2 39.4 5.0
N2
2.4vol% SiC-Al 5 minutes Liquid 43.3 36.1 52.0 5.4
N2
0.8vol% HfB2-Al 30 minutes Liquid 33.0 31.0 34.4 1.0
N2
0.8vol% SiC-Al 30 minutes Liquid 40.9 33.2 52.7 4.9
N2
Al- 30 minutes Water 67.2 48.1 80.7 9.8
AlSi7Mg0.5Cu0.5

0.8vol% HfB2-Al- 30 minutes Water 71.7 51.7 91.6 12.8

AlSi7Mg0.5Cu0.5

AlSi7Mg0.5Cu0.5 97.9 80.7 126.8 16.5

Al 5 minutes Liquid
N2
Al 5 minutes Liquid
AlSi7Mg0.5Cu0.5 N2
0.8vol% HfB2-Al 5 minutes Liquid
AlSi7Mg0.5Cu0.5 N2
*HV 0.05: micro vickers hardness measured with a load of 0.05 kgf.

For nanocomposites containing 0.8 vol% of nanoparticles (SiC and HfB2) sonicated for
30 min and quenched by liquid nitrogen, the measured hardness (table 3.2) for nanocomposites
was notably greater than the hardness of the base material: average HvSiC = 40 HV and HvHfB2
= 33 HV compared to 24 HV in pure Al. The increase of the hardness of > 40% is likely due to
the presence of hard nanoparticles at the grain boundaries, which enhances the hardness of the
final composites. In order to understand in more depth the mechanisms underlying hardness
increase, it is worth characterizing average hardness values inside Al grain. The average Hv
inside the Al grains for the sample 0.5vol% SiC-Al is 32.2 HV, higher than the value for pure
Al (24 HV), this notable increase is more likely to be attributed to the presence of SiC
83
nanoparticles trapped inside the grains, as observed by TEM and discussed above. The average
hardness of HfB2-Al inside the grains has no significant improvement compared with that of
pure Al, thus confirming full segregation of HfB2 particles at grain boundaries. At fixed
concentration, the average hardness for 0.8 vol% SiC nanocomposites was slightly greater than
the hardness of HfB2 nanocomposites. This is contrary to the Vickers hardness of bulk SiC and
HfB2, of 25.5 and 30.5 GPa, respectively.11 The difference might be attributed to the following
reasons: a) enhanced hardness of the grains due to SiC incorporation in the grains, b) inverse
Hall-Petch occurring occurred for HfB2 particles smaller than 10 nm.

When varying the concentration of SiC from 0.8 to 2.4 vol%, the average hardness of
SiC-Al increases from 40 HV to 43 HV. The large standard deviations for these values must
however be underlined. This may be explained by macro-segregation observed at 2.4 vol%.
Samples processed with different ultrasound treatment period showed no significant change in
hardness for 0.8 and 2.4 vol% SiC-Al samples. This result can be confirmed by their similar
microstructure according to the microscopic images.

HfB2 nanoparticles were dispersed in the AlSi7Mg0.3Cu0.5 foundry alloy as well. When
the two-step incorporation method was used, the foundry alloy and pure Al were mixed with a
ratio of 1:1. As expected, the hardness of the reference sample prepared by this method without
nanoparticles was lower than the hardness of the initial alloy (67.2 HV for the mixed alloy and
97.9 HV for the foundry alloy). The sample with nanoparticles shows similar hardness
compared to the reference sample, with a hardness of 71.7 HV. Considering the high standard
deviation around 10 HV, this difference is not significant. This result is in accordance with
microscopy observations, where the Si-based solute of the alloy is responsible for the hardness,
rather than incorporated nanoparticles.

Nanoindentation tests

We then sought deeper understanding of structure-property relationships in the nanocomposites

by using nanoindentation to assess the mechanical properties of the matrix and of reinforcement
nanoparticles. Based on the the microindentation results, 4 samples (tables 3.3 and 3.4) were
analyzed to study the influence of the heterogeneous composition on the hardness and the elastic

84
modulus of nanocomposites at the sub-microscopic level. The results were compared with
reference samples prepared in the same way in the absence of nanoparticles.

Table 3.3 Hardness from nanoindentation (at a depth of 3 µm).

Matrix NPs UT Cooling Hardness (GPa) Hardness of central

(min) method population (GPa)
Average SD Median Average SD Median
Al (Ref) 5 Liquid N2 0.43 0.03 0.42 0.43 0.02 0.42
Al HfB2, 5 Liquid N2 0.53 0.07 0.54 0.49 0.06 0.49
0.8 vol%
(3 wt%)
Al SiC, 30 Liquid N2 0.52 0.07 0.52 0.52 0.03 0.52
0.8 vol%
(1 wt%)
Al SiC, 5 Liquid N2 0.80 0.36 0.67 0.68 0.13 0.66
2.4 vol%
(3 wt%)
Al+ 5 Liquid N2 0.77 0.14 0.71 0.74 0.06 0.72
AlSi7Mg0.3Cu0.5
Al+ HfB2, 30 Liquid N2 0.80 0.20 0.74 0.75 0.09 0.74
AlSi7Mg0.3Cu0.5 0.8 vol%
(3 wt %)
SD stands for standard deviation
The average of central population eliminates the 25% lowest values and the 25% highest values

85
 Table 3.4. Elastic modulus from nanoindentation (at a depth of 3 µm).

Matrix NPs UT Cooling Elastic modulus (GPa) Elastic modulus of

(min) method central population (GPa)
Average SD Median Average SD Median
Al (Ref) 5 Liquid N2 79 2 79 79 0.5 79
Al HfB2, 5 Liquid N2 80 4 80 78 3 79
0.8 vol%
(3 wt%)
Al SiC, 30 Liquid N2 79 2 78 76 1 78
0.8 vol%
(1 wt%)
Al SiC, 5 Liquid N2 95 10 91 91 4 89
2.4 vol%
(3 wt%)
Al+ 5 Liquid N2 88 3 88 88 1 88
AlSi7Mg0.3Cu0.5
Al+ HfB2, 30 Liquid N2 86 4 85 86 1 85
AlSi7Mg0.3Cu0.5 0.8 vol%
(3 wt %)
SD stands for standard deviation
The average of central population eliminates the 25% lowest values and the 25% highest values

As described in the previous section, SEM images and microindentation showed that
the ultrasound treatment time has no significant effect on the microstructure and the hardness
of the final material for both SiC and HfB2 nanocomposites, and the hardness is largely related
to the location of aggregates of nanoparticles. This allows us to compare samples prepared with
different ultrasound treatment times. The mapping of the mechanical properties correlated with
the optical image of the samples is helpful to distinguish the interior of the grains and grain
boundaries. 400 measurements were achieved on a matrix of 21*19 with a distance of 10 µm
at a depth of 1 µm (figure 3.9). The hardness of several areas could not be reliably measured
(black pixels in the hardness maps and values equal to 0) due to the existence of micropores,
especially for the sample at 2.4 vol% SiC-Al. These defects could be ascribed to the scratching
of nanoparticles during polishing steps, or the presence of micropores in the sample. In fact,
with the increase of the load, Al melt tends to be more viscous and micropores are more difficult
to be eliminated, further investigations on SEM should be performed to confirm this. Figure 3.9
86
presents the mapping of mechanical properties of the 0.8% HfB2-Al nanocomposite. A
heterogeneous hardness distribution was observed: a population with a relatively homogeneous
hardness corresponding to the grains and a second population significantly higher for the grain
boundaries. Similar observations were made for other nanocomposites: SiC-Al and HfB2-
AlSi7Cu0.5Mg0.3 (Appendix 3.14). The high hardness/modulus regions (grain boundaries) and
low hardness/modulus regions (grains) will be discussed separately in the following sections.

Figure 3.9 (a) Optical microscopy image, corresponding (b) hardness and (c) elastic modulus
of a 0.8% HfB2-Al nanocomposite.

Effect of the nature of nanoparticles on the hardness of pure Al. The hardness was
recorded on nanocomposites, pure Al and Al alloy in a matrix of 6*5 with a depth of 3 µm
(Table 3.3). The 0.8 vol% SiC-Al composite exhibited higher hardness than the 0.8 vol% HfB2-
Al sample, with values 0.52 and 0.49 GPa, respectively (table 3.3 and distribution figure 3.10a).
These results were in agreement with microhardness tests, where nanocomposites containing
the same volume percentage of SiC nanoparticles were harder than that of HfB2. Measurements
at a depth of 1 µm were performed 400 times to reach a statistical assessment of mechanical
properties. The cumulative distribution curve (figure 3.10b) for the 0.8 vol% SiC-Al composite
is above the curve for the 0.8 vol% HfB2-Al sample, indicating that the reinforcement effect of
SiC is more pronounced than HfB2 on the hardness of the final nanocomposite. Nonetheless,
HfB2 still shows reinforcement of hardness properties compared to bare aluminum (figure
3.10b).

87
Figure 3.10. Statistic tests for hardness measured at (a) 3 µm and (b) 1 µm for reference Al,
0.8 vol.% HfB2-Al and 0.8 vol.% SiC-Al

The grain matrix and the intergranular regions were then compared. Measurements at
1 µm depth revealed the reinforcement effect at the extreme surface. In low hardness regions,
the middle of the plateau for 0.8 vol% SiC was higher than that of pure Al (0.6 GPa vs 0.43
GPa) (figures 3.10b and 3.11b). Hardness increases in the low hardness regions, correspond the
matrix grain boundaries, implies trapping of part of the SiC nanoparticles inside Al grains,
which is in accordance with TEM observations described above. Conversely, the hardness
distribution for the 0.8 vol% HfB2-Al at 1 µm depth (figure 3.10b) shows low-hardness regions
with hardness values similar to pure Al around 0.45 GPa, confirming that most HfB2
nanoparticles were segregated towards grain boundaries. The distribution curve for HfB2-Al
(figure 3.10b) deviates towards higher hardness than pure aluminum for more than 50% of
measurements (from 200 to 400), which corresponds to reinforced area at grain boundaries
according to mapping (figure 3.9). The average of this population is 0.66 GPa (calculated from
the highest 200 measurements), clearly higher than pure Al. These measurements indicated that
the reinforcement effect of HfB2 nanoparticles occurs only at grain boundaries.

88
Figure 3.11. Statistical evaluation of hardness measured at 3 µm (a) and 1 µm (b) for 0.8 vol.%
SiC-Al, 2.4 vol.% SiC and reference Al

Influence of the content of nanoparticles on the hardness of pure Al. The hardness
measured at an indentation depth of 3 µm for the 2.4 vol% SiC-Al composite was on average
0.68 GPa (Table 3.3), significantly higher than for the 0.8 vol% SiC-Al sample. The hardness
of 2.4 vol% SiC nanocomposites has been increased by 68% compared to the reference
aluminum matrix. However, we should also notice that with the increase of inclusion content,
the standard deviation is also higher. The extremely high hardness regions up to 3 GPa were
found to be aggregates of SiC. This indicates that the microstructure is more heterogeneous, as
observed in microscopy images (figure 3.8a).

Effect of nanoparticles on the hardness of foundry Al alloy. In the case of the

AlSi7Mg0.3Cu0.5 foundry alloy, a similar hardness has been observed: 0.79 GPa on average with
0.8 vol% HfB2 nanoparticles, compared with 0.77 GPa for the reference (table 3.3). The large
standard deviation of 0.2 GPa was mainly related to the different hardness values of Al grains
and of Si-rich regions in the alloy. The distribution curves (figure 3.12) were drawn from 30
measurements at 3 µm (figure 3.12a) and 400 measurements at 1 µm (figure 3.12n). The latter
show an overlapping between the bare matrix and the HfB2-containing one in the 240 lowest
hardness values. On the opposite, a deviation towards higher hardness values is observed for
the highest 160 values, a similar fashion as for the Al matrix composites. The tangential
midpoint of the distribution curves between the 250th and 400th measurements are 1.8 GPa for
the sample with nanoparticles and 1.4 GPa for the reference matrix. Therefore, HfB2
89
nanoparticles enhances the hardness, again at the grain boundaries (Appendix 3.14), consistent
with the observations for the Al matrix composites.

Figure 3.12. Statistic evaluation of hardness measured at 3 µm (a) and 1 µm (b) for reference
Al-AlSi7Mg0.3Cu0.5 and for the 0.8 vol.% HfB2-Al-AlSi7Mg0.3Cu0.5 nanocomposite

Effect of nanoparticles on the elastic modulus. The elastic modulus has also been
measured by nanoindentation (Table 3.4). The mapping of the elastic modulus (figure 3.9)
follows the same trend as the hardness mapping. Measurements at 1 µm depth (figure 3.13)
reveal that the incorporation of 0.8% nanoparticles of HfB2 and SiC could both improve the
average elastic modulus by 13% and 16%, respectively. With the increase of SiC nanoparticles
to 2.4%, segregated areas at grain boundaries showed high elastic modulus. On contrary, for
measurements at 3 µm, samples containing 0.8% of nanoparticles showed similar elastic
modulus with Al reference. The less obvious reinforcement effects at 3 µm could be attributed
to the increased analysis volume, where the grain boundary effects are averaged out over the
matrix grains. As the nanoindentation measurements (Figure 3.13 and 3.14) only focused on
the local modulus, these results presented a high standard deviation and could only serve as a
rough estimation of the property of final nanocomposite. Mechanical traction measurements
are required to further evaluate the macroscopic behavior of the nanocomposites.

90
Figure 3.13. Elastic modulus measured at 1 µm

Figure 3.14. Elastic modulus measured at 3 µm

91
3.4 Conclusion

Herein we have incorporated silicon carbide nanocrystals and HfB2 nanoparticles synthesized
in molten salts in aluminum-based matrices by ultrasounds-assisted liquid metallurgy. Overall,
the nanoparticles are stable after heating and aggregated at the grain boundaries. This
segregation contributes to a refinement of the microstructure by decreasing the matrix grain
size. Mechanical characterization by micro- and nano-indentation delineated the hardness and
strength profiles of these composites. The hardness is heterogeneous at the sub-microscale,
exhibiting a population with relatively homogeneous hardness corresponding to the matrix
grains and a second population with significantly higher hardness for the intergranular zones.
The reinforcement effect is less pronounced for HfB2-based nanocomposites than for SiC-based
samples, which could be ascribed to larger segregation of the boride particles at the grain
boundaries. Indeed, inside the matrix grains, SiC nanocomposites are harder than pure
aluminum, while the HfB2-based nanocomposites show mechanical properties similar to the Al
reference. This further confirms the poor dispersion of HfB2 particles into aluminum, which is
probably due to the initial aggregation state of the starting nanoparticles. Quenching the sample
in liquid nitrogen allows refining the microstructure to limit the growth of the aluminum grains.
Other quenching methods, previously developed to fabricate amorphous alloys30 should be
explored in the future to obtain better dispersed HfB2-Al nanocomposite, as well as modified
molten salt strategies to design less aggregated boride nanoparticles.

92
References

1. Casati, R. & Vedani, M. Metal Matrix Composites Reinforced by Nano-Particles—A

Review. 65–83 (2014) doi:10.3390/met4010065.
2. Mohanavel, V., Suresh Kumar, S., Sathish, T. & Anand, K. T. Effect of ZrB2 content on
mechanical and microstructural characterization of AA6063 aluminum matrix
composites. Mater. Today Proc. 5, 13601–13605 (2018).
3. Wang, F., Chu, K., Li, Y. biao & Wang, X. hu. Enhanced Interfacial Bonding and
Mechanical Properties of Graphene/Cu Composites: A Matrix-Alloying Method. Phys.
Status Solidi Appl. Mater. Sci. 215, (2018).
4. Liu, X., Liu, Y., Huang, D., Han, Q. & Wang, X. Tailoring in-situ TiB2 particulates in
aluminum matrix composites. Mater. Sci. Eng. A 705, 55–61 (2017).
5. Liu, W., Cao, C., Xu, J., Wang, X. & Li, X. Molten salt assisted solidification
nanoprocessing of Al-TiC nanocomposites. Mater. Lett. 185, 392–395 (2016).
6. Cao, C., Ling, H., Murali, N. & Li, X. In-situ molten salt reaction and incorporation of
small (10 nm) TiC nanoparticles into Al. Materialia 7, (2019).
7. Nazari, M., Eskandari, H. & Khodabakhshi, F. Production and characterization of an
advanced AA6061-Graphene-TiB2 hybrid surface nanocomposite by multi-pass friction
stir processing. Surf. Coatings Technol. 377, (2019).
8. Pan, S. et al. Effect of electron concentration on electrical conductivity in in situ Al-TiB2
nanocomposites. Appl. Phys. Lett. 116, (2020).
9. Portehault, D., Delacroix, S., Gouget, G., Grosjean, R. & Chan-Chang, T. H. C. Beyond
the Compositional Threshold of Nanoparticle-Based Materials. Acc. Chem. Res. 51, 930–
939 (2018).
10. Gupta, S. K. & Mao, Y. Recent Developments on Molten Salt Synthesis of Inorganic
Nanomaterials : A Review. (2021) doi:10.1021/acs.jpcc.0c10981.
11. Xie, C. et al. Stable and hard hafnium borides: A first-principles study. J. Appl. Phys.
125, (2019).
12. Estruga, M., Chen, L., Choi, H., Li, X. & Jin, S. Ultrasonic-assisted synthesis of surface-
clean TiB2 nanoparticles and their improved dispersion and capture in al-matrix
nanocomposites. ACS Appl. Mater. Interfaces 5, 8813–8819 (2013).
13. Javadi, A., Pan, S. & Li, X. Fabrication of High Strength Al Nanocomposites with
Populous TiB2 Nanoparticles. Procedia Manuf. 26, 629–632 (2018).
14. Internal reference from PSA Groupe.
15. Portehault, D. et al. A General Solution Route toward Metal Boride Nanocrystals.
Angew. Chemie 123, 3320–3323 (2011).
16. Prasad Reddy, A., Vamsi Krishna, P., Narasimha Rao, R. & Murthy, N. V. Silicon
Carbide Reinforced Aluminium Metal Matrix Nano Composites-A Review. Mater.
Today Proc. 4, 3959–3971 (2017).
17. Komarov, S. & Kuznetsov, D. Erosion resistance and performance characteristics of
niobium ultrasonic sonotrodes in molten aluminum. Int. J. Refract. Met. Hard Mater. 35,
76–83 (2012).
18. Luo, Z. P. Crystallography of SiC/MgAl2O4/Al interfaces in a pre-oxidized SiC
reinforced SiC/Al composite. Acta Mater. 54, 47–58 (2006).
19. Wang, X. The Formation of AlB2 in an Al-B Master Alloy. J. Alloy. Comp. 403, 283–
287 (2005).
20. Nautiyal, P., Gupta, A., Seal, S., Boesl, B. & Agarwal, A. Reactive wetting and filling
93
 of boron nitride nanotubes by molten aluminum during equilibrium solidification. Acta
Mater. 126, 124–131 (2017).
21. Lü, S., Xiao, P., Yuan, D., Hu, K. & Wu, S. Preparation of Al matrix nanocomposites by
diluting the composite granules containing nano-SiCp under ultrasonic vibaration. J.
Mater. Sci. Technol. 34, 1609–1617 (2018).
22. Hoekstra, J. & Kohyama, M. Ab initio calculations of the β-SiC(001)/Al interface. Phys.
Rev. B - Condens. Matter Mater. Phys. 57, 2334–2341 (1998).
23. Dehnavi, M. R., Niroumand, B., Ashrafizadeh, F. & Rohatgi, P. K. Effects of continuous
and discontinuous ultrasonic treatments on mechanical properties and microstructural
characteristics of cast Al413-SiCnp nanocomposite. Mater. Sci. Eng. A 617, 73–83
(2014).
24. Lü, S., Xiao, P., Yuan, D., Hu, K. & Wu, S. Preparation of Al matrix nanocomposites by
diluting the composite granules containing nano-SiCp under ultrasonic vibaration. J.
Mater. Sci. Technol. 34, 1609–1617 (2018).
25. Jia, S., Zhang, D., Xuan, Y. & Nastac, L. An experimental and modeling investigation
of aluminum-based alloys and nanocomposites processed by ultrasonic cavitation
processing. Appl. Acoust. 103, 226–231 (2016).
26. Chen, L. Y., Peng, J. Y., Xu, J. Q., Choi, H. & Li, X. C. Achieving uniform distribution
and dispersion of a high percentage of nanoparticles in metal matrix nanocomposites by
solidification processing. Scr. Mater. 69, 634–637 (2013).
27. Liang, G., Ali, Y., You, G. & Zhang, M. X. Effect of cooling rate on grain refinement of
cast aluminium alloys. Materialia 3, 113–121 (2018).
28. Chen, L. Y., Peng, J. Y., Xu, J. Q., Choi, H. & Li, X. C. Achieving uniform distribution
and dispersion of a high percentage of nanoparticles in metal matrix nanocomposites by
solidification processing. Scr. Mater. 69, 634–637 (2013).
29. Xuan, Y. & Nastac, L. The role of ultrasonic cavitation in refining the microstructure of
aluminum based nanocomposites during the solidification process. Ultrasonics 83, 94–
102 (2018).
30. Klement W, Willens RH, D. P. No TNon-crystalline structure in solidified gold–silicon
alloys. itle. Nature 187(4740), 869–70 (1960).
31. Lei, W., Portehault, D., Dimova, R. & Antonietti, M. Boron carbon nitride
nanostructures from salt melts: Tunable water-soluble phosphors. J. Am. Chem. Soc. 133,
7121–7127 (2011).
32. Firestein, K. L. et al. High-strength aluminum-based composites reinforced with BN,
AlB2 and AlN particles fabricated via reactive spark plasma sintering of Al-BN powder
mixtures. Mater. Sci. Eng. A 681, 1–9 (2017).

94
 Chapter 4
Metal matrix nanocomposites with
improved thermal properties

95
96
ABSTRACT

Metal matrix nanocomposites with low melting point metal inclusions exhibit specific
mechanical behavior and unique properties for thermal regulation. In this work, we designed
aluminum nanocomposites consisting of phase change nanoparticles dispersed in an aluminum
matrix via liquid-state processing techniques. To fabricate these composites, we prepared firstly
core-shell Bi@SiO2 nanoparticles by colloidal methods, as described in chapter 2. The
nanoparticles were subsequently dispersed in liquid aluminum and the final composites
obtained by quenching were characterized by XRD, SEM, TEM and XPS. We demonstrate that
the silica shell acted as an efficient barrier against extensive coalescence of particles during the
casting process, although the size distribution of bismuth nano-objects widened, which
demonstrates mass transfer from one particle to the other. According to heat absorption
properties measured by DSC, Bi@SiO2-Al nanocomposites are able to store heat by melting
through a wider range of melting temperatures centred at lower temperature than for the
composite obtained without silica. These properties are stable upon cycling. The casting
pathway based on core-shell metal-silica nanoparticles was then efficient to design a heat
absorption material with low fabrication cost, thus paving the way to new thermal regulation
materials.

4.1 Introduction

Phase change materials (PCMs) display a promising range of applications for the thermal
control of electronic devices, since they can absorb/release latent heat from the solid-liquid
phase transition.1 The application for battery packs in electric vehicles is especially promising.
Indeed, the temperature of Li-ion batteries should be controlled within a certain temperature
range to ensure performance, cycling ability and safety. Phase change materials, compared to
air/liquid cooling or heat pipe, use less power and they require less complex thermal system
design.2 Jaguemont et al.3 have reviewed various PCM systems for battery applications and
concluded that PCM could be a low-cost, efficient thermal solution if the conductivity of these
materials could be improved. Thus, many studies have been performed by mixing phase change
materials with metal matrix, possessing better thermal conductivity compared with common
polymer matrices. Metallic materials can also provide better flame-retardant properties than
organic phase change materials. For instance, metal (Al/Cu) foam coupled with organic PCM
97
improved cooling efficiency and decreased local heat accumulation.4,5,6 Abraham et al.7 have
further tried to solve the problem of weakened structural strength when the inclusions melt, by
creating an Al matrix nanocomposites with Bi inclusions encapsulated with Al2O3. The
composite with inclusions encapsulated in alumina shell presents higher yield strength
compared to the non-encapsulated sample, and is capable of retaining its structure during
cycling. Such encapsulation strategy would then be efficient to increase the mechanical
resistance of phase change materials. However, this strategy relies on the reduction of metal
oxide (Bi2O3) by aluminum, which shows severe limitations. First, the working temperature is
limited, as phase change only occurs at the melting temperature of bulk low melting point metal
like Bi, Ga, In, Sn. Second, the composite has been fabricated by powder metallurgy which
requires long preparation time and high fabrication costs. Further investigations are then needed
to develop an alternative encapsulation method and to move the fabrication process to liquid
metallurgy in order to reduce the cost.

To alleviate those constraints, we propose to fabricate metal matrix nanocomposites

with phase change inclusions by ultrasonic-assisted melt processing: casting the metal matrix
over its melting temperature, then dispersing the encapsulated low melting-point metal in the
liquid metallic matrix by ultrasonic-assisted treatment.8 The dispersion of core-shell Bi@SiO2
in pure Al is studied herein as a proof of concept to assess the feasibility of the strategy and to
study the reactions occurring between the nanoparticles core, shell and metal matrix under these
harsh conditions. Here, aluminum was selected as a matrix due to its light weight, high thermal
conductivity and high mechanical properties. Pure Al was used instead of industrial alloys in
order to facilitate the understanding of the reactivity between the metal matrix and the
inclusions. As Bi nanoparticles (melting point around 250 °C) are molten during the
incorporation into the molten metal matrix (melting point 660 °C) and then show a high risk of
coalescence, they were encapsulated in a silica shell prior to the composite preparation to
prevent coalescence. The choice of silica as protective shell is motivated by polymer-based
phase change materials, in which nanoencapsulation of phase change materials by silica has
been proven to be an easy and efficient way to control the size of inclusion particles.9 The
synthesis of Bi@SiO2-Al nanocomposites consists in three steps: synthesis of nanoparticles,
growth of a silica shell and incorporation in molten metal matrix by ultrasonic-assisted process,

98
as schematically illustrated in fig. 4.1. The results were compared with a reference sample
prepared from bare Bi nanoparticles into molten Al, without a protective shell.

Figure 4.1 scheme of synthesis procedures

4.2 Experimental section

Materials

Bismuth acetate (99.99%), Oleylamine (technical grade, 70%), 1-dodenanethiol (98%),

tetraethyl orthosilicate (TEOS, 99.99%), ammonia aqueous solution (28–30%) and aluminum
powder (99.99%, 320 mesh) were all purchased from Sigma-Aldrich. Aluminum beads (99.9%,
around 3mm diameter) were obtained from Alfa Aesar. The reagents were used as received.

Synthesis of bismuth nanoparticles and silica-coated bismuth nanoparticles

Bismuth nanoparticles were prepared by colloidal synthesis in a mixture of organic solvents

based on a previous published protocol and already described in chapter 2.10 Bismuth acetate
(1 mmol) was mixed with 1- dodecanethiol (11.1 mL) and heated to 45 °C for 45 min under
vacuum. The reaction flask was purged with N2 and heated to 60 °C. Oleylamine (22.2 mL) at
room temperature was quickly added under stirring and the reaction mixture was further heated
for 24 h. Finally, Bi nanoparticles were washed two times by ethanol/chloroform

99
(20/1 vol./vol.) and re-dispersed in ethanol. The silica-coated bismuth nanoparticles were
prepared by the classic Stöber method as follows11: tetraethyl orthosilicate (TEOS) was
dissolved in absolute ethanol at a concentration of 0.38 M. 10 μL of TEOS solution and 100 μL
of ammonia 30% were subsequently added to 1 mL alcoholic bismuth suspension [Bi] = 6.10-4
mol.L-1 under vigorous agitation. The mixture was heated at 40 °C for 3 hours followed by
washing by two cycles of centrifugation-redispersion in ethanol. The final powder was dried in
a vacuum oven at 40 °C for 16 hours. In order to evaluate the thermal stability, 20 mg of
Bi@SiO2 nanoparticles were annealed in a Naberthem chamber furnace under air condition for
1 h dwell time with a heating rate of 10 °C/min.

Synthesis of Bi-Al nanocomposites

Bi-Al nanocomposites were prepared by ultrasound-assisted casting of a mixture composed of

nanoparticles and Al in an induction oven. Briefly, 2 g of aluminum powders and alcoholic
bismuth suspension were mixed by sonication at the boiling point for 1 hour to get
stoichiometry ratio of 0.3 at. % between Bi and Al. Then the mixture was dried in a vacuum
oven at 40 °C for 1 hour, followed by pressing into a 13 mm pellet under 8 tons for 2 minutes.
Subsequently, the pellet and 0.4 g of pure Al beads were heated into an alumina crucible by an
induction oven (CEIA, power cube 90 serie 200) at 740 °C under argon atmosphere. After
dwelling for 1 minute, an ultrasound power of 80 watts (QSonica, Q500A-220, USA) was
applied through a sonication horn (Nb tip with 4 mm diameter) to disperse the nanoparticles for
5 minutes. The final composite was obtained by cooling down the melt in the crucible. It was
then cut with a diamond wire saw for further analyses.

X-ray diffraction

X-ray diffraction was performed with a D8 Bruker diffractometer operating at the CuKα
radiation in the Bragg Brentano configuration. The ICSD reference 64709 was used for the
indexation of bismuth nanoparticles.

Electron microscopy

The cut pieces of the samples were observed with scanning electron microscopy (SEM, Hitachi
SU-70 microscope) equipped with an Oxford X-Max 50 mm2 EDS detector. For preparation,
100
the samples were mounted in polyfast resin (Struers), followed by a series of polishing steps
through 1000, 2400, 4800 grit papers (Buchler), then with 3 μm and 1 μm diamond suspensions.

For the observation of nanoparticles in transmission electron microscopy (TEM), the

nanoparticles were dispersed in ethanol, then a drop was evaporated on a carbon-coated copper
grid. For the composites, TEM thin films were prepared by a parallelepiped rectangle of
3*3*0.5 mm cut in the middle of the sample. Polishing was then performed with a diamond
disk to get a 200 μm thick sample. The thickness of the centre of the slab was further reduced
to 50 μm by using a precision dimpling instrument (Model 515, South Bay Technology). Ion
milling was performed for the final thinning. TEM images were recorded on a Tecnai spirit G2
microscope operating at 120 kV. The STEM images and STEM-EDX mapping were recorded
with a JEM 2100Plus UHR microscope and a Tecnai F20 that were both operating at 200 kV.
The size distribution of precursor nanoparticles was determined by counting 150 to 200
nanoparticles from TEM images. The size distribution of nanoparticles in the final
nanocomposites was determined by counting 70 to 120 particles, from SEM images and TEM
images, respectively.

Chemical analysis by Wavelength dispersive X-ray fluorescence spectrometry

(WDXRF)

Wavelength dispersive X-ray fluorescence (WDXRF) spectroscopy was performed on the cut
pieces of the samples with a Bruker S8 Tiger spectrometer equipped with a Rh tube.

Differential Scanning Calorimetry (DSC)

The DSC analysis was performed in a TA Q20 DSC instrument. In a typical experiment,
Around 10 mg of nanoparticles were sealed in an aluminum crucible. In the case of
nanocomposites, top, middle and bottom slices of the samples (around 40 mg) were analysed
separately to assess the homogeneity of the casted samples. To study the phase transition, the
samples were heated to 320 °C and then cooled to 20 °C with a slope of 10 °C·min-1 followed
by 3 minutes of stabilisation at 20 °C between each cycle. For each sample, At least 3 cycles
were performed.

101
4.3 Results and discussion

4.3.1 Synthesis and characterization of Bi and Bi@SiO2 nanoparticles

The first step of the composite preparation consists in the synthesis of size controlled bismuth
nanoparticles (Figure 4.1), which was performed according the colloidal synthesis previously
reported by Scheele et al.10 The powder X-ray diffraction (XRD) patterns confirmed that Bi
nanoparticles were obtained (Figure 4.2a). The TEM images (Figure 4.2b) showed that the
sample consists in bismuth nanoparticles around 25 nm with a relatively narrow size
distribution.

Figure 4.2. (a) Powder XRD pattern and (b) TEM images of Bi nanoparticles, (c) Bi@SiO2
nanoparticles and (d) Bi@SiO2 heated at 350 °C for 1 h. Insets in (b) and (d) show the
corresponding size distributions of Bi core.

In a second step, the bismuth nanoparticles were coated with a 7 nm thick silica shell by
the Stöber method,11 as demonstrated by TEM, with a lower contrast for the shell than for the
102
Bi core (Figure 4.2c). TEM (Figure 4.2c) shows that the silica shell actually embeds clusters of
several particles, forming aggregates of size up to 160 nm. This aggregation is usual for the
encapsulation of small nanoparticles by Stöber process, as suspensions of nanoparticles smaller
than 100 nm are poorly stable and show a strong tendency to aggregate in an alcoholic
solution.12 To demonstrate the stabilizing action of the silica shell against coalescence of the
bismuth cores, we have heated the core-shell particles and sole bismuth nanoparticles without
inorganic shells above the melting point of bulk Bi (271 °C). The bare Bi particles are not stable
when heated at 320 °C and undergo extensive coalescence and growth. (Appendix 4.2) In the
opposite, the silica shell preserves the Bi cores from coalescence and ensures that the Bi core
size distribution is maintained (Figure 4.2d). Although Bi nanoparticles are aggregated inside
the silica shell, their size has been preserved when heated over the melting point, thus allowing
us to use them in further steps to study the stability of Bi nanoparticles in molten aluminum.

4.3.2 Synthesis and characterization of Bi@SiO2-Al nanocomposite

Nanocomposites made of an aluminum matrix embedding Bi@SiO2 (Bi@SiO2-Al) and

Bi (Bi-Al) nanoparticles were achieved by ultrasound-assisted casting. To ensure full
incorporation of the nanoparticles in molten aluminum, aluminum and nanoparticles powders
were firstly dispersed in ethanol by sonication. The mixture was then dried and pressed into a
pellet with Bi loading of 0.25 at.%. The particles were then dispersed in aluminium at 740 °C
by ultrasound treatment in an induction furnace that ensures efficient heating and melting of
the aluminum matrix. The resulting samples showed metallic aluminum appearance (Appendix
4.1). The Bi/Al ratio was evaluated by WDXRF on the top, middle and bottom parts of each
sample. It is about 0.3 at. % in all pieces, which is consistent with the initial precursor ratio and
indicates that the nanoparticles are homogeneously distributed into the whole volume of the
samples.

SEM images (Figure 4.3a) highlights white areas corresponding to electron-rich

bismuth, with a size up to 120 nm. TEM (Figure 4.3b-d) evidences the existence of 10 nm
(Figure 4.3b, c) and 60 nm (Figure 4.3d) nanoparticles of bismuth, as demonstrated by high
resolution TEM (HRTEM, Figure 4.3c) and selected area electron diffraction (SAED) (Figure
4.3d). Bigger nanoparticles around 120 nm were also observed in TEM-HAADF images
presented in Appendix 4.3. The statistical analysis of SEM and TEM images allows the
103
estimation of the size distributions at two different length scales. The TEM investigation of
small nanoparticles displayed a distribution with a mean diameter of ca. 8 nm, while SEM
investigation showed a wide distribution with an average of 60 nm. Particles below 10 nm
showed spherical shape, while the shape of particles above 30 nm was irregular. Therefore,
SEM and TEM evidence a widening of the size distribution from initial 25 nm Bi nanoparticles
towards both smaller and larger particles. This behaviour is indicative the Ostwald ripening
mechanism during the incorporation process, which indicates partial dissolution of bismuth into
molten aluminum as expected from the Bi-Al binary phase diagram, which indicate a ca.
1.5 wt.% solubility of Bi in Al at 700 °C.13 Note that sonication may also exacerbate Ostwald
ripening, as already described on polymer composites.14,15 In addition to the bismuth particles
dispersed in the aluminum grains, we have identified thin Bi regions around 1 nm (Appendix
4.4 e,g) between aluminum grains and conformal to grain boundaries. We attribute this feature
to bismuth segregation at the Al grains’ solidification front during cooling down.

No silicon was detected by using STEM-EDS analysis on a single bismuth nanoparticle

of diameter below 10 nm (Appendix 4.4f). Therefore, bismuth and/or silicon could migrate
away from each other in molten aluminum.

104
Figure 4.3 SEM and TEM characterization of a Bi@SiO2-Al nanocomposite. (a) backscattered
electron imaging-mode SEM image, bright and dark areas correspond to Bi and Al regions,
respectively. (b)-(d) TEM images. Dark and light contrast correspond to Bi and Al, respectively.
Insets in (a) and (b) show the corresponding size distributions. The inset in (d) shows the
selected-area electron diffraction (SAED) pattern corresponding to image (d) and indexed
along the structure of bismuth.

EDS mapping of an area containing a large coalesced bismuth object (figure 4.4a-f)
showed that Bi was surrounded by an O-rich area also containing Al, but exempt of Si.
Therefore, bismuth is surrounded by alumina. HRTEM (Figure 4.5a, orange square) next to the
Bi particle in the Al/O-rich area indicates an Al/O rich rod-like region with non-uniform
contrast which highlights crystal defects. The interplanar distance measured by HRTEM
(Figure 4.5a,b) is consistent with Al2O3 (100) planes. Alumina should arise from
aluminothermic reduction of the silica shell of the particles. Therefore, low oxidation state
silicon species are expected to form concomitantly.

105
Figure 4.4. (a) STEM-HAADF, (b) STEM images and corresponding STEM-EDS maps of
Bi@SiO2-Al nanocomposite (c-f),

106
Figure 4.5. TEM images of the analyzed area. Image (a) was taken in the white rectangle in
figure 4.4(a,b). The upper-left area corresponds to Al-O rich area and the bottom-right area
corresponds to Si rich area. (b) shows the Fast Fourier Transforms (FFTs) corresponding to
the orange square in (a) and indexed along the structure of Al2O3. (c) presents the FFT of the
green square in (a). Image (d) was taken in the blue rectangle in figure 4.4(a,b). Image (e) was
taken in the blue square in figure 4.4(b). Figure (e) shows the FFT corresponding to the dashed
area in (d) and indexed along the structure of Al2SiO5.

The Si map (figure 4.4f) confirms that Si was no longer around Bi nanoparticles: Si
migrated out of Bi nanoparticles towards a low contrast area (figure 4.4b and 4.5a,d) with no
bismuth. The low overlap between Si and O-rich regions observed in the Si and O maps (figure
4.4f and 4.4e) is consistent with the formation of low oxidation state species of silicon from
107
aluminothermic reduction. We observed two regions in the Si-rich area: an amorphous-like area
(figure 4.5a, green square), with corresponding FFT (figure 4.5c) showing no evidence of
crystalline order; and a crystalline region (figure 4.5d, blue square). The corresponding FFT
(figure 4.5e) corresponding FFT (Figure 4.5e) is fully consistent with the structure of sillimanite
Al2SiO5.

EDS performed on a coalesced area (Appendix 4.4 a-d) indicates a Bi:Si atomic ratio of
0.2, well below the initial ratio of 2.0 in the pristine Bi@SiO2 nanoparticles. This is in
agreement with Bi and Si elements migrating away from each other during the incorporation of
the particles.

In order to further understand redox processes involving the aluminum matrix and the
silica shell, the surface states of a cut piece were analysed by XPS in the Si 2s (figure 4.6a) and
Si 2p (figure 4.6b) regions. The spectrum of pure aluminum was recorded as a reference (figure
4.6). The Al plasmonic peak overlaps with the Si 2p binding energy regions between 99 eV and
102 eV (figure 4.6b). The Si 2s area was then chosen to identify the Si chemical states
(figure 4.6a). Although a wide Al plasmon peak is also observed in the Si 2s region at 147.4 eV,
it does not significantly overlap with the Si 2s signal at about 150-155 eV. For the Bi@SiO2-
Al composite (red line, figure 4.6a), the most intense peak at 155.1 eV is ascribed to the Si 2s
binding energy for Si-O bonds of SiO2.16 We observe an additional contribution between
154 eV and 150 eV that is not observed for pure aluminum. This contribution is suggests the
presence of other silicon species at lower binding energy than for SiO2, which is consistent with
low oxidation state Si species and/or an aluminosilicate.17 Note that oxidation of low oxidation
state Si species could also occur during the preparation of the sample for XPS measurements
and exposure to the air. Hence, we can hypothesize that silicon may be more reduced in the
pristine sample than indicated by XPS.

108
Figure 4.6. XPS analysis of Bi@SiO2-Al nanocomposites (red) and pure Al (black). (a) shows
the Si 2s area; (b) displays the Si 2p area.

The reduction process of SiO2 (about 30 μm) by molten aluminum has been studied by
Wang et al.18 In the Al-SiO2 system, the reduction of silica by aluminum is thermodynamically
favourable (reaction 1 below). Once alumina is formed, it can react with silica and then form
mullite or other aluminosilicates (reaction 2), evolving finally towards alumina and silicon
(reaction 3).

4 Al + 3 SiO2 → 2 Al2O3 + 3 Si (1)

3 Al2O3 + 2 SiO2 → Al6Si2O13 (2)

8Al + 3Al6Si2O13 → 13Al2O3 + 6 Si (3)

Wang et al.18 also demonstrated that after 30 min of reaction at 700 °C after a 5 °C·min-
1
heating rate, a 5 μm-thick shell of alumina was found surrounding unreacted silica particles,
while silicon was within the aluminum matrix and formed rod-like Si inclusions. In our study,
the silica component consisted in silica shell with a thickness of 7 nm, three orders of magnitude
lower that the silica particle size investigated by Wang et al.,18 which supports full reaction of
the silica shell at a similar temperature. Nonetheless, aluminothermic reduction in our case is
not complete as Si(IV) is not fully reduced in Si(0). On the opposite, the reaction is stopped at
an intermediate stage, thus delivering an intermediate aluminosilicate. This behaviour can be
attributed to shorter reaction times, as our experiments rely on induction heating at a rate of
350 °C·min-1 and reaction time of 10 min. We achieved sillimanite as intermediate phase.
Generally, sillimanite forms above 776 °C,19 40 °C over the experimental heating temperature.
109
This high-temperature phase may be induced by intense local heating provided by ultrasound
treatment.20

The formation of an aluminosilicate should be accompanied with the silicon, although

no crystallized silicon was observed by TEM. The amorphous-like area we have highlighted
(green square in STEM-EDS mapping in figure 4.4a,b and figure 4.5a) is then attributed to
amorphous silicon. Note that this area could also correspond to crystalline silicon with an
orientation away from diffraction conditions. Further analysis needs to be performed to confirm
this.

Overall, we propose the following reaction mechanism: the pristine silica shell is
reduced by aluminum that is concomitantly oxidized and reacts with remaining silica to form
an aluminosilicate intermediate. During ultrasound-assisted heating for the nanocomposite
preparation, the silica shell separates from the Bi core during, leading to leaching of Bi droplets
which then undergo Ostwald ripening and develop into two populations.

4.3.3 Synthesis and characterization of Bi-Al nanocomposite

A comparative study of uncoated Bi nanoparticles dispersed in molten aluminum was

performed. Figure 4.7 displays the HAADF and TEM images of such a Bi-Al composite.
Similar to the observations made for the Bi@SiO2-Al composite, the particle size evolved
towards two populations due to Ostwald ripening: coalesced particles around several
micrometres and detached particles around 5 to 30 nanometres. Thus, the particle size was no
longer maintained at the nanoscale.

110
 Figure 4.7. HAADF (a) andTEM image(b) of a Bi-Al nanocomposite and size
distribution.

The comparison between Bi-Al and Bi@SiO2-Al samples show that although the silica
shell underwent chemical reaction and local mobilisation, it acted efficiently against
coalescence of the cores occurring in the sample without protective shell. Consequently, a
nanocomposite with wide size distribution at the nanoscale was generated.

4.3.2 Heat absorption ability evaluated by DSC measurements

DSC analysis for nanoparticles

To assess the heat absorption ability, DSC was performed for both precursor nanoparticles and
final nanocomposites. Figure 4.8 displays the melting behaviour of the 3rd heating cycle for
pristine Bi and Bi@SiO2 nanoparticles. Bi@SiO2 showed an endothermic peak at ∼242 °C,
lower than the melting temperature of bulk Bi at 271.3°C, which is in agreement with the
depression of the melting point with particle size decrease.21 The melting point measured for
25 nm silica coated Bi nanoparticle is in accordance with the results of Bi-carbon
nanocomposites.22 Contrary to the similar melting behaviour for Bi@SiO2 nanoparticles upon
cycling (figure 4.8b), the bare bismuth particles displayed a small exothermic peak at ∼245 °C,
followed by two sharp endothermic peaks at ∼263 °C and 271 °C after 3 cycles, the latter
corresponding to bulk Bi melting. The first two endothermic peaks are attributed to the melting
temperature depression with the particle size.21 The appearance of the peaks at higher
temperature indicates that the nanoparticles coalesced and large particles up to bulk were
111
formed. The results for both samples are consistent with TEM observations after analysis.
(Appendix 4.2)

Figure 4.8. Differential Scanning Calorimetry (DSC) analysis of Bi (a) and Bi@SiO2 (b)
nanoparticles

DSC analysis for nanocomposites with phase change inclusions

The DSC curves for the composites (Figure 4.9a) also show endothermic peaks during heating
and an exothermic peak during cooling indicating solidification of undercooled bismuth. The
high temperature regions of DSC curves (Figure 4.9b) show that the melting peaks significant
differ from those for the initial particles described above. The Bi@SiO2-Al nanocomposite
exhibits two endothermic peaks with an onset temperature at 251°C: a broad melting peak at
263 °C corresponding to nanoparticles’ melting and a low intensity sharp peak at 271 °C
corresponding to bulk bismuth. The shift of these peaks to higher temperatures compared to
unreacted nanoparticles confirms that some of the nanoparticles coalesced during the
incorporation of the particles in the melt. The enthalpy ratio for the peaks at 263 °C and at
271 °C is 9:1, indicating that around 90% of Bi element are still nanoscaled, assuming that the
melting latent heat of nanoscale Bi is close to the value for bulk Bi. The heat absorption ability
evaluated by the enthalpy of the melting process was 0.38 J·g-1, close to the expected value
based on the chemical composition at 0.47 J·g-1 and the melting latent heat fo bulk Bi. This
value is 100 times lower compared to Bi@Al2O3-Al prepared by powder metallurgy,23 but we
may improve it by increasing the amount of inclusion particles within the matrix. DSC over 15
consecutive heating-cooling cycles between 20 and 300 °C shows the cycling ability of the Bi
core after melting, as the DSC curves remained unchanged during cycling.

112
Figure 4.9. DSC analysis of a Bi@SiO2-Al nanocomposite (a) with zoom in the area 250-290 °C
during heating (b). for the 1st and 15th cycles. The DSC analysis of a reference Bi-Al sample is
also shown for the first cycle in (b) (orange).

We did not detect by DSC the phase change of small nanoparticles around 8 nm expected
from TEM observations, which occurred at around 240 °C in Ag matrix.21 This could be
explained by the decreased enthalpy with the particle size24 or the low volume fraction of
nanoparticles smaller than 10 nm. These hypothesis need to be studied by further experiments.

DSC further confirms the efficient role of the silica shell to limit the coalescence of Bi
nanoparticles, since the Bi-Al nanocomposite showed three endothermic peaks at 262, 269 and
271 °C overlapping together and shifted to higher temperature compared to the Bi@SiO2-Al
nanocomposite, which implies the progressive coalescence of nanoparticles into bulk particles
during cycling.

Compared to a Bi-Al composite prepared by powder metallurgy7 at the same heating

rate and with micron and nano-sized bismuth inclusions, the distribution of melting
temperatures for Bi@SiO2-Al prepared by our liquid metallurgy process is broader and shifted
to lower temperatures. This demonstrates a decreased Bi particle size achieved through our
protocol. The small initial nanoparticle size, the protective silica shell, the ultrasound treatment
induced disruption of Bi droplets and rapid thermal treatments by induction heating are the keys
to achieve small inclusion nanoparticles with a wide phase transition temperature range, which

113
is a considerable advantage for the final application as they can provide progressive heat
absorption in a wide temperature range.

4.4 Conclusions

In conclusion, we have successfully developed an ultrasound assisted casting synthesis of phase

change metal matrix nanocomposites. The phase change nanoparticles were embedded within
a silica shell, which acts as a barrier against extensive coalescence of the metal cores, which
are molten during the incorporation process.

We have made several attempts for in situ fabrication of Bi/Al2O3/Al by ultrasound-

assisted casting method (see Appendix S4.1 and S4.2). The formation of encapsulated
inclusions by aluminoreduction proved difficult to be applied to other types of inclusions such
as In/Al2O3/Al: the reaction conditions must be adjusted with the Gibbs free energy for the
formation of metal oxide. An additional challenge of prime importance relates to the decrease
of the melting point of the PCM nanoparticles. Reaching other phase transition temperatures
with metal alloy inclusions will require an indepth work in the future. We believe that such
light-weight aluminum-based functional materials with high heat dissipation ability, wide and
tunable phase transition temperature and low fabrication costs will be a relevant model from
the point of view of thermal control of vehicle battery.

114
References
(1) Wu, M. S.; Liu, K. H.; Wang, Y. Y.; Wan, C. C. Heat Dissipation Design for Lithium-
Ion Batteries. J. Power Sources 2002, 109 (1), 160–166. https://doi.org/10.1016/S0378-
7753(02)00048-4.
(2) Offer, G.; Patel, Y.; Hales, A.; Bravo Diaz, L.; Marzook, M. Cool Metric for Lithium-
Ion Batteries Could Spur Progress. Nature 2020, 582 (7813), 485–487.
https://doi.org/10.1038/d41586-020-01813-8.
(3) Jaguemont, J.; Omar, N.; Van den Bossche, P.; Mierlo, J. Phase-Change Materials
(PCM) for Automotive Applications: A Review. Applied Thermal Engineering. Elsevier
Ltd March 2018, pp 308–320. https://doi.org/10.1016/j.applthermaleng.2017.12.097.
(4) Qu, Z. G.; Li, W. Q.; Tao, W. Q. Numerical Model of the Passive Thermal Management
System for High-Power Lithium Ion Battery by Using Porous Metal Foam Saturated with
Phase Change Material. Int. J. Hydrogen Energy 2014, 39 (8), 3904–3913.
https://doi.org/10.1016/j.ijhydene.2013.12.136.
(5) Alipanah, M.; Li, X. Numerical Studies of Lithium-Ion Battery Thermal Management
Systems Using Phase Change Materials and Metal Foams. Int. J. Heat Mass Transf.
2016, 102, 1159–1168. https://doi.org/10.1016/j.ijheatmasstransfer.2016.07.010.
(6) Wu, W.; Yang, X.; Zhang, G.; Ke, X.; Wang, Z.; Situ, W.; Li, X.; Zhang, J. An
Experimental Study of Thermal Management System Using Copper Mesh-Enhanced
Composite Phase Change Materials for Power Battery Pack. Energy 2016, 113, 909–
916. https://doi.org/10.1016/j.energy.2016.07.119.
(7) Abraham, A.; Schoenitz, M.; Dreizin, E. L. Energy Storage Materials with Oxide-
Encapsulated Inclusions of Low Melting Metal. Acta Mater. 2016, 107, 254–260.
https://doi.org/10.1016/j.actamat.2016.01.074.
(8) Mohanty, P.; Mahapatra, R.; Padhi, P.; Ramana, C. H. V. V.; Mishra, D. K. Ultrasonic
Cavitation: An Approach to Synthesize Uniformly Dispersed Metal Matrix
Nanocomposites—A Review. Nano-Structures and Nano-Objects 2020, 23, 100475.
https://doi.org/10.1016/j.nanoso.2020.100475.
(9) Shchukina, E. M.; Graham, M.; Zheng, Z.; Shchukin, D. G. Chem Soc Rev
Nanoencapsulation of Phase Change Materials for Advanced Thermal Energy Storage
Systems. 2018, 4156–4175. https://doi.org/10.1039/c8cs00099a.
(10) Scheele, M.; Oeschler, N.; Meier, K.; Kornowski, A.; Klinke, C.; Weller, H. Synthesis
and Thermoelectric Characterization of Bi2Te 3 Nanoparticles. Adv. Funct. Mater. 2009,
19 (21), 3476–3483. https://doi.org/10.1002/adfm.200901261.
(11) STOBER, W. Sagittal Plane Spinal Mobility Is Associated with Dynamic Balance
Ability of Community-Dwelling Elderly People. J. Colloid Interface Sci. 1968, 26 (1),
62–69. https://doi.org/10.1016/0021-9797(68)90272-5.
(12) Wang, J.; Shah, Z. H.; Zhang, S.; Lu, R. Silica-Based Nanocomposites via Reverse
Microemulsions: Classifications, Preparations, and Applications. Nanoscale 2014, 6 (9),
4418–4437. https://doi.org/10.1039/c3nr06025j.
(13) Manasijević, D.; Minić, D.; Balanović, L.; Premović, M.; Gorgievski, M.; Živković, D.;
Milisavljević, D. Experimental Investigation and Thermodynamic Prediction of the Al–
Bi–In Phase Diagram. J. Alloys Compd. 2016, 687, 969–975.
https://doi.org/10.1016/j.jallcom.2016.06.262.
(14) Thompson, J. A.; Chapman, K. W.; Koros, W. J.; Jones, C. W.; Nair, S. Sonication-
Induced Ostwald Ripening of ZIF-8 Nanoparticles and Formation of ZIF-8/Polymer
115
 Composite Membranes. Microporous Mesoporous Mater. 2012, 158, 292–299.
https://doi.org/10.1016/j.micromeso.2012.03.052.
(15) Dirix, Y.; Bastiaansen, C.; Caseri, W.; Smith, P. Preparation, Structure and Properties of
Uniaxially Oriented Polyethylene-Silver Nanocomposites. J. Mater. Sci. 1999, 34 (16),
3859–3866. https://doi.org/10.1023/A:1004614604641.
(16) Barr, T. L. An XPS Study of Si as It Occurs in Adsorbents, Catalysts, and Thin Films.
Appl. Surf. Sci. 1983, 15 (1–4), 1–35. https://doi.org/10.1016/0378-5963(83)90003-X.
(17) Hughes A.E., Hedges M.M., S. B. A. Reactions at the Al/SiO2/SiC Layered Interface. J.
Mater. Sci. 1990, 25, 4856.
(18) Deqing, W.; Ziyuan, S. Aluminothermic Reduction of Silica for the Synthesis of
Alumina-Aluminum-Silicon Composite. J. Mater. Synth. Process. 2001, 9 (5), 241–246.
https://doi.org/10.1023/A:1015295200586.
(19) Salje, E. Heat Capacities and Entropies of Andalusite and Sillimanite: The Influence of
Fibrolitization on the Phase Diagram of the Al2SiO5 Polymorphs. Am. Mineral. 1986, 71
(11–12), 1366–1371.
(20) Suslick, S.; Fang, M. Acoustic Cavitation and Its Chemical Consequences. Phil. Trans.
R. Soc. Lond. 1999, No. 1927, 335–353.
(21) Liu, M.; Ma, Y.; Wu, H.; Wang, R. Y. Metal Matrix-Metal Nanoparticle Composites
with Tunable Melting Temperature and High Thermal Conductivity. Liu M, Ma Y, Wu
H, Wang RY. ACS Nano 2015, 9 (2), 1341–1351. https://doi.org/10.1021/nn505328j.
(22) Tran, N.; Zhao, W.; Carlson, F.; Davidson, J. H.; Stein, A. Metal Nanoparticle − Carbon
Matrix Composites with Tunable Melting Temperature as Phase-Change Materials for
Thermal Energy Storage. 2018. https://doi.org/10.1021/acsanm.8b00290.
(23) Khan, Z.; Khan, Z.; Ghafoor, A. A Review of Performance Enhancement of PCM Based
Latent Heat Storage System within the Context of Materials, Thermal Stability and
Compatibility. Energy Convers. Manag. 2016, 115, 132–158.
https://doi.org/10.1016/j.enconman.2016.02.045.
(24) Fu, Q.; Zhu, J.; Xue, Y.; Cui, Z. Size- and Shape-Dependent Melting Enthalpy and
Entropy of Nanoparticles. J. Mater. Sci. 2017, 52 (4), 1911–1918.
https://doi.org/10.1007/s10853-016-0480-9.
(25) N. Birks, G. H. Meier and F. S. Pettit. Introduction to the High Temperature Oxidation
of Metals. Cambridge Univ. Press 2006, 003, 17.
(26) Caroli, V. G. M. and S. Tables of Standard Electrode; 1978; Vol. 2.

116
General conclusions and perspectives

117
118
The electrical vehicle is expanding due to the gradual awareness and national policies of
reducing fuel consumption and emissions of pollutants. But until now, there are still challenges
that must be addressed to improve its performances and enable wide spread application. Long
endurance as one important concern that requires the decrease of the vehicle’s weight.
Electrification of the whole vehicle also encounters several challenges, particularly with regard
to heat regulation. Improving the thermal management of batteries is a particularly important
issue to ensure safety on board, to optimize cell aging and to preserve its performances. It is on
these topics that the work in this thesis focused: fabrication of a metal matrix nanocomposites
HfB2-Al with enhanced mechanical properties, and fabrication of a metal matrix nanocomposite
with phase-transition inclusions made of phase change Bi cores encapsulated in silica, as a proof
of concept for thermal regulation. The fabrication process applied in this work is liquid
metallurgy assisted by ultrasound treatment that could be scaled up and implemented in current
work flows for the production of aluminum alloys.

We have addressed the incorporation of nanoparticles synthesized in molten salts into

aluminum-based matrices by ultrasound-assisted liquid metallurgy. An experimental set-up was
designed and the process was carefully engineered to achieve total incorporation of
nanoparticles in the matrix with decreased porosity and homogeneous chemical composition
over the whole volume at the micron scale.

Several experiments were then performed by dispersing in molten aluminum 0.8 vol.%
of HfB2 nanoparticles synthesized in molten salts. Commercial SiC nanocomposites were also
assessed as references. According to electron microscopy, both types of nanoparticles are able
to maintain their size and morphology after the high temperature melting process, but
nanoparticles were segregated at the grain boundaries. HfB2 particles were only located at the
boundaries, but SiC nanoparticles were also observed inside the Al grains, indicating better
dispersion. Liquid nitrogen quenching was then applied to refine the microstructure. Parameters
regarding the improvement of wettability (composition of the matrix, type of the nanoparticles,
coating of the nanoparticles) or the quenching step (in particular other protocols developed to
manufacture amorphous alloys) might be explored to gain further control over the dispersibility.

The strengthening effect of the nanoparticles was studied with nanoindentation tests.
The mapping of the mechanical properties correlated with the optical image, which showed
119
heterogeneous hardness at the micro scale: a population with a relatively homogeneous
hardness corresponding to the matrix grains and a second population with significantly higher
hardness for the intergranular zones, in line with the high concentration of nanoparticles of hard
phases at grain boundaries. The same phenomenon is also observed in other nanocomposites
such as SiC-Al, HfB2-AlSi7Cu0.5Mg0.3. The average hardness of the sample with the inclusion
of HfB2 nanoparticles was improved by 14%. This value is lower than that of SiC
nanocomposite with the same volume ratio (21%), probably due to the difference in segregation
state. The elastic modulus measured by nanoindentation showed no difference between pure Al
and nanocomposites with 0.8 vol% loading of nanoparticles (SiC and HfB2). The tensile
strength might also be measured to reveal the differences of elastic modulus at the macroscale.

Three main pathways were explored to fabricate bismuth-aluminum composites by

liquid metallurgy. Firstly, Bismuth oxide was dispersed in molten Al and further reduced in situ
into metallic bismuth by aluminoreduction. This process was proven hard to be replicated for
other inclusions with are more difficult to be reduced, such as In or metallic alloys. Secondly,
bismuth nanoparticles were directly incorporated in the Al matrix. Nanoparticles underwent
large extent Ostwald ripening during the dispersion in molten Al, forming two groups of
particles with diameters below 10 nm and about several micrometers, respectively. Thirdly, we
have developed an encapsulation method by designing core-shell Bi@SiO2 nanoparticles, to
incorporate in molten Al. The nano-inclusions also underwent size distribution widening
towards smaller and larger sizes by Ostwald ripening, but to a smaller extent than the non-
encapsulated sample since the maximal particle size of the enlarged particles was about only
200 nm. This result indicates that although the silica shell underwent chemical reaction and
local mobilization, it acted efficiently against core coalescence. We have also demonstrated that
the silica shell underwent aluminothermic reduction, which led to silicon and an aluminosilicate
as intermediate. Thermal analysis indicated that around 90% of Bi element are still in nanoscale.
The average phase-transition temperature is lower than the Bi-Al2O3 nanocomposite prepared
by powder metallurgy described in the literature.2 Based on these results, the next tracks to
further explore is the major challenge of reaching phase change nanoparticles suitable for
thermal management of Li-ion battery, with an operating temperature at 60-80 °C. This could
be achieved by incorporating nanoparticles of low melting point alloys, such as Bi-In alloys.
By this method, we could expect to widen the phase transition temperature range to achieve

120
progressive heat absorption with the increase of the battery’s temperature. The next step would
then be to select low-cost elements and materials that would render the approach viable for an
industrial process.

These proofs of concept nonetheless appear to be very encouraging steps towards the
development of new metal matrix nanocomposites. Thanks to its adaptability, the described
process could be used for boride-based light-weight materials or for thermal management
materials for electronic/electrical device through the choice of appropriate inclusions and metal
matrix.

121
122
Appendix

S3.1 Design of the process for the preparation of nanocomposites

S3.1.1 Incorporation methods

The nanoparticles were first incorporated to the molten metal following the same procedure as
described in the literature,1 that is to say direct incorporation of the powders. Attempts have
been made by using BN nanosheets,31 SiC and HfB2 nanoparticles, and Bi2O3 microparticles
(several micrometres). Despite the removal of the native aluminum oxide layer at the surface
of the metal matrix grains prior to melting, we were not able to incorporate all the nanoparticles
inside the matrix by simple stirring methods (Appendix 3.1). For samples dispersed by 5
minutes ultrasound treatment, a loss of more than 90% of initial particles was detected by
WDXRF analysis (Appendix 3.2). Such a low inclusion content compared to the initial ratio
could be due to the precipitation of nanoparticles since brown powders were observed at the
bottom of the crucible. The difficulties to incorporate the nanoparticles into the melt lie in the
poor wettability and the high volume fraction between the nanoparticles powder and liquid
aluminum.

Appendix 3.1. (a) Al ingot after melting in an induction furnace, (b) Al ingot with BN
nanoparticles after melting in induction furnace. (c) Direct incorporation of SiC nanopowders
in pure molten aluminum with 5 min ultrasound treatment. Brown powders were found at the
bottom of the crucible. (d) Direct incorporation of HfB2 powders in pure molten aluminum with
5 min ultrasound treatment.

123
 Nanoparticles Matrix UST WDXRF approx. wt.%
Al Ti Bi Hf Si B
AlSi7Cu0.5Mg0.3 No 92 / / / 7 /
2 wt% BN AlSi7Cu0.5Mg0.3 No 92 / / / 7 Not
detected
Al Ti tip, 94 1.4 / / /

5 mins
2 wt% HfB2 Al Ti tip, 80 5.5 / 300 / Detected
ppm
5 mins
30 wt% Al Ti tip, 91 1.5 3.3 / / /
Bi2O3
5 mins
2 wt% SiC Al Ti tip, 78 0.9 / / 0.6 /

5 mins

Appendix 3.2. WDXRF analysis of samples with direct incorporation of nanoparticles in molten
Al

A master nanocomposite method was proposed by the group of Li via sintering of nanoparticles
and Al micropowder (50 wt % loading) together at 900 °C leading to a robust and compact
pellet.5 In our study, this method cannot be directly applied because of the cohesion of the
nanoparticles and of the potential reactivity between Al micron scale grains and the particles’
boron-rich shell before dispersion. An alternative two-step processing method was then
developed. Nanoparticles were dispersed in absolute alcohol in the presence of the aluminum
powder. This suspension was dried in an ultrasound bath to ensure its homogeneous deposition
over aluminum powder. The mixture was pressed into a pellet, then diluted in an aluminum
melt for ultrasound dispersion. This additional step of compaction of the nanopowder with
aluminum enhanced the overall incorporation. The macroscale homogeneity of the as-
synthesized nanocomposites was evaluated by WDXRF. For a HfB2-Al nanocomposite, the

124
chemical composition of 3 cut pieces was identical, and the inclusion ratio (Hf/Al) was close
to the expected value (Appendix 3.3a,b). The slight difference between expected and measured
values could be explained by the presence of Ti impurity.

Piece code Hf/Al at. ratio

Expected 0.025
Original 0.02
A1 0.02
B2 0.02
Bottom 0.02

Appendix 3.3. Two-step synthesis for a 0.8 vol.% HfB2-Al nanocomposites: (a) Photograph of
the sample, (b) WDXRF analysis on different cross sections. (c) Photograph of 0.8 vol.%
HfB2/Al nanocomposites.

The maximum load of inclusions prepared by this method was evaluated by increasing the
nanoparticle ratio in the pellet. The sedimentation of particles at the bottom part of the matrix
was observed in the case of HfB2-Al nanocomposites with nanoparticle content over 2 vol.%
(Appendix 3.3c). Furthermore, when the loading of inclusion was over 5 vol.%, either for SiC
or for HfB2, it was impossible to press nanoparticles/Al powder mixtures into a cohesive pellet
under our experimental conditions (8 tons for 2 minutes). A loose and fragile pellet was
obtained. Therefore, we chose to study nanocomposites with volume fraction < 2.4 vol.%. It
could also be noticed that at this concentration, it is impossible to detect nanoparticles by XRD.

S3.1.2 Improvement of the ultrasonic treatment setup

High intensity ultrasonic waves generate cavitation and acoustic streaming, which is most likely
to refine the microstructure, break aggregates and homogenize the distribution of nanoparticles.
However, as mentioned in chapter 1, this method requires a good control of processing
condition such as ultrasonic power, ultrasonic treatment time, erosion state of the tip and the

125
depth of dipping of the tip in liquid metal. Here we designed an experimental set-up adapted to
small volume of aluminum melt around 2 mL to optimize these conditions.

Concerning the choice of the tip, most of the commercial microtips are made of titanium alloy.
In our study, titanium tip brought around 1 wt. % of Ti after 10 minutes of ultrasound treatment
in pure aluminum. Besides, titanium is a strong boride former that could react with the boron
shell of HfB2 nanoparticles, leading to an increased ratio of Ti in WDXRF analysis. Komarov
et al. studied the erosion resistance of Nb tip and demonstrated a good stability over 10 hours
of ultrasound treatment.32 Thus, a niobium tip with 4 mm of diameter has been designed on
purpose by the supplier to replace the classic titanium alloy tip. EDS analysis of samples treated
with Nb tip showed no presence of Nb pollution (Appendix 3.4).

Appendix 3.4 EDS analysis of HfB2-Al nanocomposites treated by Nb ultrasonic probe for 5
minutes

The form of the crucible was specially designed as well to increase the dipping depth of the
ultrasonic tip (3 mm). Compared to the extremely large batches (ca. 1 L) described in previous
studies, this set-up is suitable for preliminary studies without needs to scale up the synthesis of
nanoparticles.

Previous works have proven that mechanical properties of nanocomposites obtained by

discontinuous ultrasonic treatments were better improved than with continuous ultrasonic
treatments for 20 min period.16 An interval of 5s on/ 5s off was then applied for all experiments.
This interval time could also prevent overheating of the ultrasound generator. Furthermore, it
was necessary to assess the influence of the ultrasound treatment power. A power varying from
10 to 40% of the nominal power was applied during the preparation of HfB2-Al samples. Strong
ultrasonic power over 40% resulted in the splash of liquid on the upper crucible wall leading to
126
a loss of around 42% of the sample. Intermediate ultrasonic power caused less liquid splash,
but pores with size varying from 0.05 to 1 mm were formed at the top part of the sample. The
formation of these bubbles could be explained by confinement of the cavitation bubbles during
solidification process. According to these observations, the best compromise was to use an
ultrasound power of 30% for ultrasonic dispersion, followed by 30 s degassing with a power of
10%. These conditions could act as a model to study the interactions between nanoparticles and
matrix environment.

Appendix 3.5. XRD patterns of a 0.8 vol.% HfB2-Al nanocomposite prepared with ultrasound
treatment through a Nb tip for 30 minutes, and a zoom between 21° and 46°. The peak at 29°
and 41° could be attributed to AlB12.

127
 Appendix 3.6 Al-B Phase diagram 19

Appendix 3.7. HRTEM image and corresponding Fast Fourier Transforms (FFTs) indexed
along the AlB12 structure for a 0.8 vol.% HfB2/Al nanocomposite prepared with ultrasound
treatment through a Nb tip for 30 minutes.

128
 Appendix 3.8. STEM-EDS mapping of a 0.8 vol% HfB2/Al nanocomposite prepared with
ultrasound treatment by a Ti tip for 5 minutes and cooled down in liquid nitrogen.

Appendix 3.9. SEM-EDS mapping and optical microcopy images for Al:AlSi7Mg0.3Cu0.5 1:1
treated at 600°C for 5 minutes

129
Appendix 3.10. (a-e,g,h) SEM-EDS mapping and (f,i) optical microscopy images for pristine
AlSi7Mg0.3Cu0.5

130
Appendix 3.11. Optical microscopy images of a 0.8 vol% HfB2-Al nanocomposite prepared with
ultrasound treatment through a Nb tip for 5 minutes and cooled down in liquid nitrogen. The
observations were made on a vertical cross section. (a) Center of the section, (b) bottom of the
section. Images of a 0.8 vol% SiC-Al nanocomposite prepared with ultrasound treatment
through Nb tip for 5 minutes and cooled down in the liquid nitrogen. (c) Near the surface, (d)
bottom near the crucible and (e) center part

131
Appendix 3.12. Optical microcopy images of (a) a Al/AlSi7Mg0.3Cu0.5 mixture prepared with
ultrasound treatment through a Nb tip for 5 minutes and cooled down in the water; (b) a
Al/AlSi7Mg0.3Cu0.5 nanocomposite prepared with ultrasound treatment through a Nb tip for 5
minutes and cooled down in liquid nitrogen, (c) a 0.8 vol% HfB2-AlSi7Mg0.3Cu0.5
nanocomposite prepared with ultrasound treatment through a Nb tip for 5 minutes and cooled
down in water; (d) a 0.8 vol% HfB2-AlSi7Mg0.3Cu0.5 nanocomposite prepared with ultrasound
treatment through a Nb tip for 5 minutes and cooled down in the liquid nitrogen.

132
Appendix 3.13. SEM observations and EDS mapping of 0.8 vol.% HfB2/Al nanocomposite
prepared with ultrasound treatment through a Nb tip for 30 minutes and cooled down in liquid
nitrogen

133
Figure S7. Optical images (left) and nanoindentation mapping of (middle) hardness and (right)
elastic modulus of Al (reference), 0.8 vol% SiC-Al, 2.4 vol% SiC-Al, Al-AlSi7Mg0.3Cu0.5, 0.8 vol%
HfB2-Al-AlSi7Mg0.3Cu0.5.

134
 Appendix 4.1. Image of Bi@SiO2-Al sample

Appendix 4.2. TEM of bare Bi nanoparticles heated to 320°C after DSC experiments

135
Appendix 4.3. TEM-HAADF image of Bi@SiO2-Al nanocomposite

136
Appendix 4.4. a-d. STEM-EDS mapping and e-h. EDS line profile analysis of a thin slice in a
Bi@SiO2-Al sample.

137
Appendix S4.1 Fabrication of metal matrix nanocomposites by aluminum reduction of
Bi2O3 via casting method

Fabricating composites by reducing high melting point oxides of a low melting point metal via
liquid metallurgy particles has been proposed as a way to fabricate metal matrix phase change
materials by solid metallurgy.7 We have attempted a similar process by liquid metallurgy
assisted by induction heating and ultrasonication. As in the work of Abraham et al.,7 the
dispersion of commercial 2.4 at. % Bi2O3 particles about 1 to 5 micrometres has been firstly
carried out with similar process as described in above in section 4.2 for Bi@SiO2/Al
nanocomposites.

XRD patterns (Appendix 4.6) displaying metallic Bi and Al phase confirmed the success
of aluminothermic reduction. An unidentified phase was also detected and could not be
attributed to alumina. This is assigned to contamination from the ultrasonic tip. STEM-EDS
mapping (Appendix 4.6) showed that the reduced Bi2O3 particles were located at the aluminum
grain boundaries. STEM-EDS maps of a single nanoparticle (Appendix 4.5 c-f) presented an
oxygen layer around 0.2 µm surrounding the Bi core, consistent with the formation of alumina
during the reduction process.

Appendix 4.6. XRD patterns of Bi2O3 /Al nanocomposite

138
 Appendix 4.7 SEM images (a, c) and SEM-EDS (b, d-f) maps of a Bi2O3-Al nanocomposite

DSC analysis was also performed for this sample in the same conditions as those described
previously in 4.3.3.(figure Appendix 4.8) The intense peak centred at 271 °C during the heating
ramp was in accordance with the melting process of bulk Bi.

139
 Appendix 4.8 DSC analysis of a Bi2O3-Al nanocomposite.

Appendix S4.2 Fabrication of metal matrix nanocomposites by aluminothermal reduction

of In2O3 via casting method

In the objective of varying the melting temperature of inclusion particles, we have attempted to
incorporate indium inclusions into an aluminium matrix, by dispersing commercial indium
oxide nanoparticles (Sigma-Aldrich, around 75 nm) in molten aluminum. Examination of
Ellingham diagrams25 (Gibbs free energy vs temperature) indicates that the Gibbs free energy
for the formation of In2O3 is less negative than that of Al/Al2O3, so In2O3 is expected to be
reduced by Al.

According to XRD patterns (Appendix 4.11), indium was mostly in its oxidized state and the
peak of metallic In presented a very low intensity. Indium oxide was not totally reduced by Al
for the reaction with 10 minutes dwell time at 740 °C. This lower reduction kinetic compared

140
with Bi2O3 could be explained by the lower difference of Gibbs free energy between In2O3 and
Al, as predicted in Ellingham diagram (Appendix 4.10).26 On the heating DSC curve (Appendix
4.12), reduced metallic In is accounted for the broad peak centred at 157 °C, with an onset
temperature at 146 °C. The melting temperature of bulk In is 156.5°C. The asymmetric peak
shape could be assigned to the nano-size In inclusion showing lower melting temperature
compared with bulk In.

These experiments proved the feasibility of in situ fabrication of Bi/Al2O3/Al by a casting

method. However, the formation of encapsulated inclusions by aluminoreduction is difficult to
be applied to other types of inclusions: the reaction condition needs to be adjusted with the
redox potential of the metal oxide and it is difficult to achieve other transition temperature.
However, our concept of direct incorporation of inclusion metal@SiO2 nanoparticles in molten
Al presents a promising pathway to widen the nature of inclusions.

Appendix 4.10. Ellingham diagram25

141
Appendix 4.11 XRD pattern of a In2O3-Al nanocomposite.

Appendix 4.12 DSC analysis of a In2O3-Al nanocomposite.

142
Résumé en Français

Introduction

Face aux défis énergétiques et environnementaux, le Groupe PSA investit depuis

plusieurs années dans la fabrication de véhicules plus respectueux de l'environnement. Pour
mener à bien cette stratégie, le groupe souhaite inventer et fabriquer de nouveaux matériaux
légers, peu coûteux, à haute ductilité et possédant également de bonnes propriétés thermiques.
Dans ce contexte, ce travail de thèse vise à étudier les nanocomposites à matrices métalliques
(metallic matrix nanocomposites MMnCs) avec des propriétés mécaniques et thermiques
améliorées. En particulier, les procédés de fabrication doivent être conçus pour répondre aux
besoins du processus actuel de coulée de l'alliage.

Parmi les techniques de fabrication des nanocomposites métalliques, la métallurgie

1,2
liquide assistée par ultrasons peut répondre à ces exigences. L'utilisation du traitement par
ultrasons est un bon moyen pour améliorer la dispersion des nanoparticules pendant le
processus de moulage de l'alliage d'aluminium. Ce procédé offre également des avantages tels
qu'un faible coût, un processus de moulage facile et une adaptabilité à la fabrication à grande
échelle. 1

L'examen de la littérature sur les MMnC actuels ouvre de nombreuses possibilités pour
le développement de nouveaux matériaux. En ce qui concerne les propriétés mécaniques des
MMnCs, les nanoparticules inorganiques commerciales se sont avérées utiles pour améliorer
efficacement la résistance et le module d’Young de la matrice métallique.3 Les nanoparticules
de renforcement doivent être conçues pour satisfaire les mécanismes de renforcement et les
conditions de la métallurgie à haute température. Les nanoparticules de borure synthétisées par
des sels fondus sont de bons candidats car elles sont de petite taille, légères, dures et stables à
haute température.4 À notre connaissance, ce type de nanomatériaux n’a jamais été incoroporé
au sein de nanocomposites. Une bonne dispersion des nanoparticules, essentielle pour les
propriétés mécaniques finales peut être obtenue par le contrôle des paramètres de composition
(mouillabilité de la matrice, nature des composés) ou par la modification des paramètres de
mise en œuvre (procédé de chauffage, puissance du traitement par ultrasons, vitesse de
chauffage/refroidissement). Cette partie du travail sera explorée en partie 1.
143
 Les matériaux à changement de phase ont une application limitée dans le contrôle
thermique de la température des batteries des véhicules en raison de leur faible conductivité
thermique et de leur faible résistance aux vibrations.5 L'incorporation de matériaux à
changement de phase dans une matrice métallique suscite un grand intérêt dans l'industrie
automobile en raison de la haute conductivité thermique, de leurs propriétés mécaniques élevées
et de la large gamme de températures de fonctionnement des matrices métalliques.6 Dans ce
cas, le matériau à changement de phase est en général un métal à bas point de fusion. Les
stratégies conventionnelles, notamment la métallurgie des poudres, nécessitent des coûts de
fabrication élevés pour ce type de composites.7 De plus, la température de travail est également
limitée : le changement de phase ne se produit qu'à la température de fusion du métal pur. Pour
surmonter ces problèmes, une méthode en phase liquide est proposée : les nanoparticules
métalliques (Bi) seront encapsulées dans une enveloppe de SiO2 pour empêcher l'immersion
dans la matrice. Ensuite, ces nanoparticules seront incorporées dans la matrice métallique
fondue par un procédé assisté par ultrasons. Cette partie du travail sera présentée dans la partie
2.

Partie 1. Nanocomposites à matrice métallique HfB2-Al et ses propriétés mécaniques

Un montage combinant un four à induction et un ultrasonicateur opérant à haute

température a été développé pour la préparation des nanocomposites par métallurgie liquide.
Plusieurs paramètres tels que le protocole d’incorporation des nanoparticules, la nature du
creuset, la forme du creuset, la puissance des ultrasons, l’intervalle du traitement ultrasons et la
rampe du chauffage ont été étudiés afin d’identifier les meilleures conditions pour la synthèse
des nanocomposites :

- L’incorporation des particules est réalisée en deux étapes afin d’assurer un transfer quantitatif
des particules dans la matrice métallique : 1) mélanger les nanoparticules et l’aluminium à l’état
de poudres puis presser ce pré-mélange en pastille, 2) diluer la pastille dans l’Al fondu.
- Un creuset en alumine a été spécialement conçu pour augmenter la hauteur du métal fondu et la
résistance au choc thermique.
- Une canne à ultrasons en niobium a été conçue par le fournisseur afin d’éviter la pollution des
matériaux finaux par la sonotrode usuelle (alliage de titane).
- Un intervalle on/off du traitement d’ultrason a été appliqué afin d’améliorer la dispersion et de
refroidir la sonotrode.
144
 - La puissance d’ultrasons est maintenue à 30% de la puissance maximale de l’appareil pendant
le traitement, puis le milieu est traité à 10% afin d’éviter la formation de bulles avant la
solidification. En effet, une haute puissance d’ultrasons (> 40% de la puissance maximale)
proboque des projections de métal fondu à l’extérieur de la zone de chauffage. Néanmoins, la
puissance d’ultrasons à 30% conduit aussi à de la porosité dans le matériau final, induite par la
cavitation provoquée par le traitement ultrasons.

Les nanoparticules de HfB2 ont été d’abord synthétisée en sels fondus.8 Les clichés de
microscopie électronique en transmission (TEM, Figure 1a) montrent que la morphologie est
de type cœur-coquille, avec un cœur de HfB2 de diamètre variant entre 4 et 14 nm, et une couche
de bore amorphe d'environ 3 nm qui recouvre les nanoparticules de HfB2. Les diagrammes de
rayon X confirment la phase de HfB2.

Plusieurs expériences ont ensuite été réalisées en dispersant 0,8 vol.% de nanoparticules
de HfB2 dans de l’Al en utilisant les meilleures conditions définies précédemment. Des
nanocomposites Al/SiC ont aussi été réalisés comme références.

Les images TEM (Figure 1d) pour l’échantillon 0,8 vol.% HfB2-Al (traitement aux
ultrasons par une pointe en niobium pendant 5 minutes, suivi d'une trempe dans l'azote liquide)
montrent que les nanoparticules forment des agrégats submicroniques entre les grains
d'aluminium. Les nanoparticules de HfB2 ont pu conserver leur taille après le processus de
fusion à haute température. Néanmoins, leur morphologie a légèrement changé, passant de
particules rondes à des particules facettés (figure 1c). La phase HfB2 a été confirmée par
HRTEM, indiquant également la stabilité thermique des nanoparticules HfB2.

Ces échantillons ont ensuite été analysés par microscopie électronique en balayage
(SEM) couplée à l'EDS. L'agglomération des particules de HfB2 a été observée le long des joints
de grain de l’aluminium plutôt qu'une distribution uniforme dans la matrice. Les mesures
d'analyse EDS donnent une composition hétérogène aux joints de grains : les nanoparticules de
HfB2 ont ségrégé vers ces zones. Les nanocomposites à base de nanoparticules commerciales
de carbure de silicium dispersées dans l’aluminium suivent la même tendance.

145
Figure 1. a) Clichés TEM des nanoparticules de HfB2 avant incorporation. b) Diagramme de DRX de la
poudre constituée de nanoparticules HfB2. c-d) Clichés TEM du nanocomposite 0,5% HfB2-Al. e)
Clichés de SEM du nanocomposite 0,5% HfB2-Al. e) Clichés SEM du nanocomposite 0,5% SiC-Al.

L'analyse de la taille moyenne des grains d’aluminium révèle un raffinement de la

microstructure, avec des grains de plus petite taille que les grains initiaux avant chauffage. Des
échantillons de HfB2/Al à 0,5 vol.% ont été refroidis à différentes vitesses de refroidissement.
La taille moyenne des grains est de 120 µm, 78 µm et 20 µm respectivement pour les taux de

146
refroidissement de 6, 18 et 100°C min-1. Le refroidissement par azote liquide permet un
refroidissement d’environ 100°C min-1 s'avère donc être la meilleure pour affiner la
microstructure des MMnCs.

L’effet de renforcement des nanoparticules ont été étudié avec des tests en
nanoindentation. La dureté moyenne de l’échantillon avec l’inclusion des nanoparticules HfB2
est de 0,53 GPa, plus élevé que la référence de l’aluminium pur préparée dans les mêmes
conditions à 0,42 GPa. Plus de 50% des mesures montrent une dureté plus élevée que
l’aluminium pure, correspondant la zone renforcée entre les joints de grains. La moyenne de
cette population est de 0,66 GPa, nettement supérieur à l’aluminium pur. Ces mesures
confirment l’effet renfort des nanoparticules HfB2.

La cartographie des propriétés mécaniques corrélée avec l’image optique permet de

mieux visualiser la différence entre l’intérieur des grains et les joint de grains (figure 2). On
observe une dureté hétérogène : une population avec une dureté relativement homogène
correspondant aux grains et une deuxième population sensiblement plus élevée pour les zones
intergranulaires. Ce phénomène est aussi observé dans d’autres nanocomposites tel que SiC-
Al, HfB2-AlSi7Cu0.5Mg0.3.

Figure 2. Cliché de microscopie optique (gauche) correlé à la cartographie des propriétés mécaniques
(dureté, module d’élasticité) du nanocomposite 0,5% HfB2-Al.

Pour conclure cette partie 1, nous avons testé l’incorporation des nanoparticules
synthétisé en sels fondus dans la matrice de l’Al par la métallurgie liquide assistée par ultrasons.

147
Les nanoparticules sont stables après le chauffage et agrégées au niveau du joint de grain. Ceci
permet de raffiner la microstructure et d’augmenter les propriétés mécaniques entre les joint-
grains. Cependant, l’effet de renforcement est moins marqué avec HfB2 comparé aux
nanocomposites à base de nanoparticules SiC commercial. Dans la littérature4, la dureté Vickers
du SiC en vrac est de 25,5 GPa et celle du HfB2 est de 30,5 GPa. Cette différence ne pourrait
pas être expliquée par la dureté entre les nanomatériaux. En outre, les grains de matrice pour
les nanocomposites contenant des nanoparticules de SiC sont plus durs que pour l’aluminium
pur, alors les mêmes grains dans les nanocomposites à base de HfB2 sont de même ordre de
grandeur que la référence d’aluminium, ce qui indique que les grains contiennent aussi des
particules de SiC, au contraire de HfB2 qui ségrège totalement. Cette mauvaise dispersion est
probablement due à l’état d’agrégation des nanoparticules de départ (après la synthèse en sels
fondus) et à la mouillabilité de HfB2 par l’aluminium, que l’on suppose faible. La trempe de
l’échantillon dans l’azote liquide permet d’affiner la microstructure de limiter la croissance des
grains d’aluminium. D’autres protocoles de trempe, développés notamment par P. Duwez pour
fabriquer des alliages amorphes,9 seraient envisageable pour obtenir un nanocomposite HfB2-
Al mieux dispersé.

Partie 2. Nanocomposites à matrice métallique Bi@SiO2-Al avec des propriétés de la

gestion thermique

La première étape de la préparation du composite consiste en la synthèse de

nanoparticules de bismuth avec une taille contrôlée (Figure 3a). Celle-ci a été réalisée selon une
procédure de chimie colloïdale précédemment rapportée par Scheele et al.10 Des nanoparticules
de bismuth d'environ 25 nm avec une distribution de taille relativement étroite ont été
synthétisées.

Dans un deuxième temps, les nanoparticules de bismuth ont été recouvertes d'une
coquille de silice de 7 nm d'épaisseur par la méthode de Stöber.11 Les nanoparticules de Bi sont
encapsulées dans les coquilles et forment des amas de plusieurs particules. Les agrégats ont une
taille allant jusqu'à 160 nm (Figure 3b).

Des nanocomposites d’aluminium dans lesquels sont incorporées des nanoparticules de

Bi@SiO2 cœur-coquille (Bi@SiO2-Al) et de Bi (Bi-Al) ont été obtenus par le procédé décrit

148
précédemment en partie 1. Les images SEM (Figure 3c) montrent que les zones riches en
bismuth fortement contrastées ont une taille allant jusqu'à environ 100 nm. Les observations
TEM (Figure 4d) met en évidence l'existence de nanoparticules de bismuth d'environ 10 nm
(Figure 4.3d) et 60 nm. L’élargissement de la distribution de taille des particules après
incorporation, vers les tailles plus petites et plus grosses peut être expliquée par le mécanisme
de maturation d'Ostwald au cours du processus d'incorporation, qui vient de la dissolution
partielle du bismuth dans l'aluminium fondu, comme attendu d’après le diagramme de phase
Bi-Al.12 Nous avons pu mettre en évidence la réduction in situ de la coque de silice par
aluminoréduction, qui conduit à du silicium et à un aluminosilicate. Pendant le chauffage assisté
par un traitement aux ultrasons, la coquille de silice est séparée du noyau de Bi, cela conduit à
la lixiviation des gouttelettes de Bi qui ont subi une maturation d'Ostwald et ont évolué en deux
populations de taille.

Le nanocomposite Bi-Al contient des particules de Bi de plusieurs micromètres. Ce

résultat indique que bien que la coquille de silice ait subi une réaction chimique et une
mobilisation locale, elle a agi efficacement contre la coalescence des cœurs par rapport à
l'échantillon sans coquille de silice.

149
Figure 3. a) Cliché TEM des nanoparticules de Bi initiales. b) Cliché TEM des nanoparticules de
Bi@SiO2. c) Cliché SEM du nanocomposite 0,25% Bi@SiO2-Al. d) Cliché TEM du nanocomposite
0,25% Bi@SiO2-Al.

Les courbes d’analyse colarimétrique différentielle (DSC) des composites (Figure 4)

présentent des pics endothermiques pendant le chauffage et un pic exothermique pendant le
refroidissement. Le nanocomposite Bi@SiO2-Al présente deux pics endothermiques avec une
température de démarrage de la fusion à 251 °C : un pic large à 263 °C correspondant à la fusion
des nanoparticules et un pic de faible intensité à 271 °C correspondant au bismuth massif. Le
déplacement de ces pics vers des températures plus élevées par rapport aux nanoparticules de
30 nm confirme que certaines des nanoparticules ont coalescé pendant le processus de fusion.
Le rapport d'enthalpie pour les pics à 263 °C et à 271 °C est de 9:1, ce qui indique qu'environ
90 % des éléments Bi sont encore à l'échelle nanométrique. La capacité d'absorption de chaleur
évaluée par l'enthalpie du processus de fusion est de 0,38 J g-1, proche de la valeur attendue

150
basée sur la composition chimique (0,47 J g-1). les courbes DSC sont restées inchangées après
15 cycles de chauffage-refroidissement entre 20 et 300 °C, ce qui montre la bonne stabilité au
cyclage du matériaux.

Le nanocomposite Bi-Al présente trois pics endothermiques à 262, 269 et 271 °C au

chauffage, et déplacés vers une température plus élevée par rapport au Bi@SiO2-Al, ce qui
implique la coalescence progressive des nanoparticules en particules massives pendant le
cyclage sans la coquille de SiO2.

Par rapport au composite Bi-Al préparé par métallurgie des poudres dans la littérature,6
la distribution des températures de fusion pour Bi@SiO2-Al préparé par notre procédé de
métallurgie liquide est plus large et décalée vers des températures plus basses. Cela démontre
une diminution de la taille des particules de Bi obtenue par notre protocole. La petite taille
initiale des nanoparticules, l'enveloppe protectrice en silice, et le chauffage rapide par le four
en induction sont les points clés pour obtenir des petites inclusions avec une large gamme de
température de transition, ce qui est un avantage considérable pour l'application finale.

Figure 4. Analyse DSC des nanocomposites Bi@SiO2-Al (gauche) avec aggrandissement d’une zone au
chauffage (droite) pour les 1er et 15e cycles. Anaylse DSC (droite, courbe orange) pour un composite
de reference Bi-Al.

Nous avons ainsi étudié plusieurs méthodes de fabrication in situ de Bi/Al2O3/Al par la
métallurgie liquide assistée par ultrasons. La formation d'inclusions encapsulées par
151
aluminoréduction est difficilement applicable à d'autres types d'inclusions comme In/Al2O3/Al
: les conditions de réaction doivent être ajustées en fonction du potentiel redox de l'oxyde
métallique. En outre, il est difficile d'atteindre une autre température de transition de phase avec
une inclusion d'alliage métallique par la voie de la co-réduction des oxydes.

Notre concept de nanocomposite métal/SiO2-Al, sans nécessité de passer par le

processus de réduction, ouvre la voie à l'obtention de nanocomposites à matrices métalliques
avec d'autres nanoparticules en inclusions, notamment des nanoparticules d'alliage métallique
qui ne sont pas miscibles avec l'aluminium liquide. Ces matériaux fonctionnels à base
d'aluminium présentent une capacité de dissipation de la chaleur relativement importante pour
cette classe de matériaux, une température de transition de phase réglable et un faible coût de
fabrication. Ils peuvent servir de modèle pertinent pour le contrôle thermique des appareils
électriques.

Conclusion

Plusieurs méthodes de dispersion de nanoparticules dans des matrices métalliques ont

été développées. En particulier, des expériences ont été réalisées en faisant varier les paramètres
tels que la puissance d’ultrason, le temps de traitement et la température de chauffage. Ainsi,
les conditions optimales du processus d’incorporation ont pu être identifiées. Ce travail permet
d’établir un protocole pour l’obtention de nanocomposites en utilisant des approches en voie
liquide. En particulier, un nouveau dispositif d’ultrasons a été mis en place pour l’amélioration
de la dispersion de nanoparticules dans les matrices métalliques tout en évitant les
contaminations.

Concernant la fabrication et la caractérisation des propriétés mécaniques sur les

nanocomposites contenant des inclusions de SiC et de HfB2, les tests de nanoindentation et de
microindentation ont permis de quantifier la dureté aux échelles microscopique et
macroscopique. Une hétérogénéité des propriétés mécaniques a été observée sur les
échantillons à matrices aluminium avec des nanoparticules de HfB2 et de SiC. Si nous avons pu
mettre en évidence une augmentation significative de la dureté par rapport à la matrice seule,
des efforts sont encore nécessaires pour améliorer la dispersion et atteindre des propriétés
mécaniques comparables avec la littérature en utilisant d’autres protocoles plus coûteux.1

152
 En outre, nous avons développé avec succès une synthèse de nanocomposites
métalliques à changement de phase par la métallurgie liquide assistée par ultrasons. Les
nanoparticules à changement de phase ont été incorporées dans une coquille de silice, qui agit
comme une barrière contre la coalescence des cœurs fondus pendant le processus
d'incorporation.

Des mesures de propriétés d’absorption de chaleur sont réalisées pour les composites
Bi@SiO2-Al et Bi-Al. Ces études révèlent la diminution du point de fusion avec la réduction
de la taille des inclusions dans l’échantillon Bi@SiO2/Al. Ce résultat a aussi confirmé que la
couche de silice permet de protéger partiellement les particules de bismuth de la coalescence
lors du traitement d’ultrason dans l’aluminium fondu.

Ces matériaux restent néanmoins des preuves de concept encourageant en vue de

l’élaboration de matériaux légers et fonctionnels pour le monde de l’automobile. Les
nanocomposites des borures décrits pourront être amélioré pour renforcer l’alliage de l’Al.
Grâce à l’adaptabilité du procédé de synthèse des Bi@SiO2-Al, des matériaux sur mesure
pourraient être crées pour la gestion thermique des pièces électriques ou électroniques.

153
Références

1. Casati, R. & Vedani, M. Metal Matrix Composites Reinforced by Nano-Particles—A

Review. 65–83 (2014) doi:10.3390/met4010065.
2. Prasad Reddy, A., Vamsi Krishna, P., Narasimha Rao, R. & Murthy, N. V. Silicon
Carbide Reinforced Aluminium Metal Matrix Nano Composites-A Review. Mater.
Today Proc. 4, 3959–3971 (2017).
3. Firestein, K. L. et al. High-strength aluminum-based composites reinforced with BN,
AlB2 and AlN particles fabricated via reactive spark plasma sintering of Al-BN powder
mixtures. Mater. Sci. Eng. A 681, 1–9 (2017).
4. Xie, C. et al. Stable and hard hafnium borides: A first-principles study. J. Appl. Phys.
125, (2019).
5. Li, W. Q., Qu, Z. G., He, Y. L. & Tao, Y. B. Experimental study of a passive thermal
management system for high-powered lithium ion batteries using porous metal foam
saturated with phase change materials. J. Power Sources 255, 9–15 (2014).
6. Abraham, A., Schoenitz, M. & Dreizin, E. L. Energy storage materials with oxide-
encapsulated inclusions of low melting metal. Acta Mater. 107, 254–260 (2016).
7. Mohanty, P., Mahapatra, R., Padhi, P., Ramana, C. H. V. V. & Mishra, D. K. Ultrasonic
cavitation: An approach to synthesize uniformly dispersed metal matrix
nanocomposites—A review. Nano-Structures and Nano-Objects 23, 100475 (2020).
8. Portehault, D. et al. A General Solution Route toward Metal Boride Nanocrystals.
Angew. Chemie 123, 3320–3323 (2011).
9. Klement W, Willens RH, D. P. No TNon-crystalline structure in solidified gold–silicon
alloys. itle. Nature 187(4740), 869–70 (1960).
10. Scheele, M. et al. Synthesis and thermoelectric characterization of Bi2Te 3
nanoparticles. Adv. Funct. Mater. 19, 3476–3483 (2009).
11. STOBER, W. Sagittal plane spinal mobility is associated with dynamic balance ability
of community-dwelling elderly people. J. Colloid Interface Sci. 26, 62–69 (1968).
12. Manasijević, D. et al. Experimental investigation and thermodynamic prediction of the
Al–Bi–In phase diagram. J. Alloys Compd. 687, 969–975 (2016).

154
Binghua MA – Thèse de Doctorat – 2021

Nanoparticules dans des matrices métalliques pour des applications en automobiles

Résumé
Ce travail porte sur l’élaboration de nanocomposites à matrices métalliques pour des
applications automobiles. Le procédé de fabrication repose sur la dispersion de nanoparticules
ayant des capacités de renforcement ou d’absorption de chaleur dans un métal par métallurgie
liquide. Après une étude bibliographique pour situer le sujet, le manuscrit porte d’abord sur
l’élaboration des nanoparticules comme briques élémentaires, en particulier des particules
cœur-coquille Bi@SiO2 à bas point de fusion. Ensuite, plusieurs méthodes sont étudiées pour
trouver les meilleures conditions d’incorporation des particules, qui reposent sur le couplage
entre le chauffage à induction et le traitement ultrasons dans la matrice fondue. Des
nanocomposites combinant des matrices à base d’aluminium et des nanoparticules de SiC ou
de HfB2 sont ainsi fabriqués pour renforcer les propriétés mécaniques. La microscopie
électronique révèle la ségrégation des particules aux joints de grains. Les mesures de
nanoindentation et de microindentation permettent de quantifier la dureté à l’échelle
microscopique, révélant ainsi une hétérogénéité des propriétés mécaniques entre les grains et
les joints de grains, plus durs en raison de leur forte concentration en particules. Enfin, des
nanocomposites métalliques à changement de phase sont élaborés en dispersant dans
l’aluminium les particules Bi@SiO2 préalablement synthétisées. L’étude par microscopie
électronique de l’interface particules/matrice met en évidence les mécanismes d’incorporation.
La coquille de silice protège partiellement les particules de bismuth de la coalescence lors du
traitement ultrasons, permettant un abaissement de leur point de fusion. Grâce à cette méthode,
nous pourrions élargir la plage de température de transition pour obtenir une absorption
progressive de la chaleur avec l'augmentation de la température des batteries.
Mots clés : [Nanocomposites, matériaux à changement de phase, propriétés thermiques]

Aluminum matrix nanocomposites for applications in automobile industry

Abstract
This PhD work is dedicated to the synthesis of metal matrix nanocomposites for automotive
applications, by dispersing functional nanoparticles in an aluminum matrix by liquid metallurgy
to improve mechanical and thermal regulation properties. After a bibliographic study to set the
context, we describe the preparation of the initial nanoparticles as building blocks, especially
core-shell Bi@SiO2 nanoparticles with low melting point. Then, several dispersion methods are
explored to find the best conditions for the particle’s incorporation, which relies on
ultrasonication in the molten matrix during induction heating. We then prepare nanocomposites
combining Al-based matrices with SiC or HfB2 nanoparticles, to improve mechanical
properties. Electron microscopy study reveals segregation of the nanoparticles at grain
boundaries. Nanoindentation and microindentation measurements enable the quantification of
the hardness at the microscopic scale, showing a heterogeneity of mechanical properties, with
higher hardness at the grain boundaries that are concentrated in nanoparticles. We then explore
the fabrication of phase change nanocomposites by incorporating in aluminium the above
described Bi@SiO2 particles. An electron microscopy study of the Al/nanoparticle interface
sheds light on reaction mechanisms during the incorporation. The silica shell partially protects
the nanoparticles against coalescence during the ultrasonic treatment in molten aluminum. We
then assessed the role of the inclusion size on the melting point of Bi, evidencing a decrease of
the melting point with the particle size. By this method, we could then widen the phase
transition temperature range to achieve progressive heat absorption with the increase of the
battery temperature.
Key words: [Nanocomposites, phase change materials, thermal properties]