An effective approach to reinforced closed-cell Al-alloy foams with multiwalled carbon nanotubes

Accepted Manuscript An effective approach to reinforced closed-cell Al-alloy foams with multiwalled carbon nanotubes Isabel Duarte, Eduardo Ventura, Susana Olhero, José MF. Ferreira PII:

S0008-6223(15)30182-2

DOI:

10.1016/j.carbon.2015.08.065

Reference:

CARBON 10230

To appear in:

Carbon

Received Date: 18 July 2015 Revised Date:

20 August 2015

Accepted Date: 21 August 2015

Please cite this article as: I. Duarte, E. Ventura, S. Olhero, J.M. Ferreira, An effective approach to reinforced closed-cell Al-alloy foams with multiwalled carbon nanotubes, Carbon (2015), doi: 10.1016/ j.carbon.2015.08.065. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

An effective approach to reinforced closed-cell Al-alloy foams with multiwalled carbon nanotubes Isabel Duartea,* Eduardo Venturaa, Susana Olherob, José MF Ferreirab a

Department of Mechanical Engineering, TEMA, University of Aveiro, Campus Universitário

RI PT

de Santiago, 3810–193 Aveiro, Portugal b

Department of Materials and Ceramics Engineering, CICECO, University of Aveiro, Campus

SC

Universitário de Santiago, 3810-193 Aveiro, Portugal

Abstract.

M AN U

Exploring the reinforcing role of carbon nanotubes to obtain materials (polymers, metals, ceramics) with enhanced properties has been often attempted, but the success is strongly limited by the dispersing degree of carbon nanotubes. Here we report on an innovative colloidal approach to disperse the carbon nanotubes in the powders mixture of the precursor materials in order to profit from their reinforcing potential and obtain a new class of closed-cell metal foams. The feasibility of the proposed approach was demonstrated for aluminium foams reinforced with

TE D

multi-walled carbon nanotubes. These nanocomposite metal foams synergistically combine the remarkable properties of both metal foams and carbon nanotubes. The results indicate that the tubular structure of carbon nanotubes is preserved throughout the entire the process. The carbon nanotubes are individually dispersed, stretched and randomly aligned in the aluminium-matrix of

EP

these closed-cell foams, thus potentiating their homogeneous 3D reinforcing role. Accordingly,

AC C

the Vickers micro-hardness of the closed-cell foams was greatly enhanced.

Keywords. Aluminium alloy foams; Carbon nanotubes; Nanocomposite metal foams; Colloidal processing; Powder Metallurgy.

*

Corresponding author. Tel.: +351 234 370830; Fax: +351 234 370953. E-mail: [email protected] (Dr.-Ing. I. Duarte).

1

ACCEPTED MANUSCRIPT 1. Introduction Lightweight, recyclable, non-flammable metal foams have excellent energy absorption capacity to impact and good sound damping, being used in commercial and military applications [1]. Despite the wide variety of foams and the very good control of the density [2] and cellular structure [3] allowed by some processes, their mechanical strength is not satisfactory for some

RI PT

advanced applications. To cope with this limitation, composite assemblies including Al-alloy foams incorporated into hollow structures [4-6] and sandwich panels [7] have been proposed. These combinations confer an overall enhanced mechanical strength to the assemblies, with the energy-absorbing function being reserved to the foam component. But increasing further the

SC

mechanical strength of these energy absorption materials still remains a major challenge and attempts using heat treatments have been also reported [7].

Carbon nanotubes (CNTs) exhibit a set of very attractive properties (excellent mechanical

M AN U

strength, high aspect ratios, and good thermal and electrical conductivities) and their reinforcing potential has been regarded with alacrity [8]. CNTs have been incorporated into different types of material matrices (metals, polymers and ceramics) as ideal ultra-lightweight reinforcements [9-12]. While the incorporation of CNTs into a polymer-matrix is relatively well established [910], applying the same principle to metallic matrices is still very challenging [11-12]. CNTs are

TE D

rather difficult to disperse in molten metals due to their intrinsic mutual incompatibility derived from significant differences in surface tensions and the consequent poor wetting. Also, CNTs tend to agglomerate into clusters in the metallic-matrix, as a result of their large aspect ratio and strong Van der Waals forces [11-12]. Moreover, CNTs react with molten metals leading to the

EP

formation of intermetallic compounds that deny the expected final properties. Most of the earlier research efforts attempted to overcome these process limitations (dispersing CNTs in the metalmatrix, and obtaining a good bonding between the matrix and the reinforcement for an efficient

AC C

load transfer). Several methods (powder metallurgy, molecular-level mixing, plasma spraying and casting) were tested but with limited success [11-19]. The results reported in the case of the powder metallurgy method were hardly encouraging due to the poor dispersion efficiency of the CNTs within the powders mixture [11-19]. Such unsuccessful attempts relied on unsuitable dispersion means (low energy ball milling under dry or wet environments processes) that also induced mechanical damage to CNTs. An innovative approach to uniformly disperse all the powder components involved in the preparation of Al-alloy foams reinforced with −COOH functionalised multiwall carbon nanotubes (MWCNTs−COOH) by combining the powder metallurgy method [20] with a colloidal processing step was just disclosed by our research group [21]. 2

ACCEPTED MANUSCRIPT Spraying the homogeneous suspension containing all the components into liquid nitrogen (freeze-granulation) enabled obtaining homogeneous granules, which could be then lyophilised and used for preparing the precursor materials by dry pressing [21]. Following the same approach, the present work provides an insightful study of some of the most relevant processing steps, including the characterization of the starting raw materials; the effects of the pre-oxidation

RI PT

heat treatment on the foaming agent; the specific roles of the dispersing agents and their synergetic dispersing actions, assessed through complementary techniques. Detailed SEM microstructural evidences are given about the achieved uniform dispersion of MWCNTs and TiH2 in the Al-alloy matrix, which was maintained along the entire process. The Vickers

SC

microhardness of Al-foams reinforced with 0.5 wt. % MWCNTs−COOH increased considerably up to 125% in comparison to the non-reinforced ones. This means that the common processing difficulties related to the poor dispersing ability of CNTs and the extremely harsh conditions

M AN U

prevailing upon their incorporation into molten metal melts can be successfully overcome. Therefore, 'carbon science' could greatly help in developing smarter materials. This new class of closed-cell metallic foams with high mechanical performance is likely to open new market opportunities for high-tech applications for which the commercially available metal foams have

2. Experimental 2.1. Materials

TE D

insufficient mechanical strength.

EP

Commercially available powders were used to prepare the closed-cell Al-alloy foams with different functions: Al-alloy powder as metallic matrix, and Titanium hydride (TiH2) as blowing

AC C

agent supplied by Alpoco (Nottingham, UK) and Rockwood Lithium, respectively. Table 1 summarises the main characteristics of these micro-sized powders. Multi-walled carbon nanotubes functionalized with carboxyl groups (MWCNTs−COOH) were directly purchased from Nano-Amor Nanostructured & Amorphous Materials, Inc. (USA). They were prepared by catalytic chemical vapour deposition and exhibit the following characteristics: content of COOH: 3.67−4.05 wt. %; purity >95%; inside diameter: 2−5 nm; outside diameter: <8 nm; length: 10−30 µm and surface area: 350−420 m2/g. A complex commercial surfactant, Nanosperse AQ (NanoLab Inc, Waltham, MA, USA, hereafter designated as NS) was selected for its efficacy in dispersing MWCNTs [22]. Poly(vinyl alcohol), PVA (Sigma, Dorset) was used in aqueous solution as binder and co-surfactant.

3

ACCEPTED MANUSCRIPT 2.2. Preparation of the nanocomposite Al-alloy foams The as-received TiH2 powder was pre-oxidised into a preheated furnace at 480 ºC for 3 h, aiming at shifting the decomposition temperature close to the melting temperature of the Al-based alloys [19]. The schematics shown in Fig. 1 compares the former traditional powder metallurgy method of producing non-reinforced closed-cell metal foams with the proposed novel approach to

RI PT

prepare Al-alloy foams reinforced with MWCNTs−COOH. The overall process was divided into three main preparation steps [21] for obtaining: (i) homogeneous spherical granules by freeze granulation (FG) from stable and high solids loading aqueous suspensions of (Al+TiH2) and (Al+TiH2+MWCNTs−COOH) (subsection 2.2.1); (ii) foamable precursor materials (subsection

SC

2.2.2); (iii) Al-alloy foams (subsection 2.2.3).

2.2.1. Preparation of stable and highly loaded solids aqueous suspensions

M AN U

Preliminary tests were performed for selecting the most appropriate contents of processing additives for preparing stable aqueous suspensions with high solids volume fractions (φ). An aqueous solution of 1.5 wt.% PVA was then used as background dispersing medium to investigated the effects of different φ (49−56 vol.%) on apparent viscosity versus elapsed time measured in a Brookfield viscometer (model DV II + Pro, Brookfield Engineering Laboratories,

TE D

Inc., Middleboro, USA). In another series of experiments, φ was fixed at 49 vol.% and the effects of different added amounts of NS on apparent viscosity were also studied. The starting Al-alloy powder was firstly pre-mixed with 0.6 wt.% pre-oxidised TiH2 in a tumbling mixer for 30 minutes and divided in two portions with different purposes: (i) to prepare

EP

an aqueous dispersion for FG (subsection 2.2.2); (ii) to be directly used for preparing foamable precursor specimens (subsection 2.2.2). For the composite system, two individual suspensions

AC C

were separately prepared in similar liquid media containing the proper amounts of processing additives: (i) The dispersion of MWCNTs−COOH was assisted by simultaneous ultrasonic and mechanical stirring; (ii) The starting powders (Al-alloy + 0.6 wt.% pre-oxidised TiH2) mixture was dispersed by mechanical stirring and ball milling. The first suspension was then drop-wisely added to the second one under mechanical stirring to obtain an intimate mixture of all components (Al-alloy, TiH2 and MWCNTs−COOH). The complete suspension was then ball milled in a low-speed rolling system for 72 h in a ceramic jar with alumina balls (weight ratio of solids/balls ~1/1.5) for ensuring a high-level homogeneity.

4

ACCEPTED MANUSCRIPT 2.2.2. Preparation of homogeneous spherical granules The well dispersed (Al-alloy+TiH2) and (Al-alloy+TiH2+MWCNTs−COOH) suspensions were passed through a sieve of 120 µm to remove impurities and then used to prepare spherical granules by freeze granulation (FG) (LS-2 model, PowderPro AB) [21-22]. Namely, they were sprayed through a nozzle of 1.2 mm into liquid nitrogen (−196 ºC) using a peristaltic pump at 30

RI PT

rpm under pressurized air of 0.8 bar (80 kPa). The quickly frozen spherical droplets were collected and lyophilised (Telstar lyophilizer) under low pressure (0.022 mbar − 2.2 Pa) and a temperature of ~−80 ºC) to avoid any segregation. The lyophilised spherical granules were sorted

2.2.3. Preparation of foamable precursor materials

SC

through sieving of 1 mm to get the required size distribution and to remove agglomerates.

M AN U

Foamable cylindrical specimens (10 g each, 30 mm diameter) were prepared from: (i) the composition without MWCNTs (99.4 Al-12Si + 0.6 TiH2, in wt.%), either as powder blend (hereafter designated as PB); (ii) spherical granules without MWCNTs (hereafter designated as FG00); (iii) spherical granules containing 0.5 wt.% of MWCNTs relative to the dry mass of the mixture [99.4 Al-12Si + 0.6 TiH2,], hereafter designated as FG05; (iv) the Al-12Si alone for comparison purposes. The specimens were pre-compacted at room temperature (RT) in die [19]

TE D

under different applied loads (1.4, 5, 10 and 20 ton) for a short time (1 s − 30 min). An in-house assembled hot-press press equipped with a movable furnace with a maximum load of 20 tons was used. Subsequently, the filled die was then pre-heated at 400 ºC and held at this temperature for various time periods (5−45 min) followed by a final pressing step under the same applied

EP

loads (1.4, 5, 10 and 20 ton) for 1 s − 50 min [19, 2]. Foaming of all types of precursor

AC C

specimens was carried out into a pre-heated furnace at 700 ºC [2, 20, 23].

2.3. Characterisation techniques Particle size, density and morphology of the Al-alloy and TiH2 powders were analysed by laser diffraction (Coulter LS 230, Miami, FL, Fraunhofer optical model), a multipycnometer (Quantachrome Instruments, based on Archimedes principle) and a scanning electron microscope (SEM, SU-70 model), respectively. Zeta potential (ξ-potential) of MWCNTs-COOH was measured (Malvern Zeta sizer, Nano ZS, Malvern, Worcestershire, UK) using 1 mM KCl solution as background electrolyte in the absence and presence of NS and PVA. pH adjustments were performed by adding 0.1 M solutions of either HCl or NaOH. The viscosity of suspensions was measured using a Brookfield viscometer (DV II + Pro model) within the rate range of 5

ACCEPTED MANUSCRIPT 0.005–400 Pa.s with Brookfield spindles LV2 and LV3. The granule size distribution was assessed by sieving using sieves with 1000, 500, 300, 180, 125 and 75 µm. The foamable precursor materials were characterised by measuring the apparent density and the Vickers microhardness. The distribution of TiH2 and the level dispersion of MWCNTs–COOH in aluminium-matrix of precursors and nanocomposite Al-alloy foams were assessed by SEM

RI PT

(Hitachi, S4100 and SU-70 models). The thermal behaviour of the granules and the individual powder components was evaluated by thermal analysis (Setaram Labsys™TG-DSC16 and Netzsch STA 409 EP) at a heating rate of 10 ºC), in conjunction with high resolution X-ray diffraction analysis (XRD, Rigaku Geigerflex D/Mac, C Series diffractometer). The ability of the

SC

precursor specimens to expand and form Al-alloy foams was investigated by hot-stage microscopy (Leitz, 2A-model) [24]. This technique allows recording the foaming process in situ and real time during heating the precursor. The expansion was determined by analysing the

M AN U

captured sequence of 2D-images using image-analysis procedures (ImageJ) and converting them into binary images (black and white). The projected area of the foam (A) at a given time normalised by that of the initial precursor (A0) is defined as “area expansion” (A/A0). The reinforcing ability of the MWCNTs−COOH was assessed by measuring the Vickers microhardness onto polished surfaces of foam specimens on their struts (cell-walls) under an

TE D

applied force (F) of 50 gf (0.49 N) with a tolerance of ± 15% for 10 s. A Shimadzu HMV-2000 equipment with square diamond indenter, two lenses of 10x and 50x together with the 10x of eyepiece, allowing for magnitudes of 100x and 500x was used.

EP

3. Results and Discussion

3.1. Characteristics of the micro-sized and nano-sized powder components

AC C

Fig. 2 shows the different morphologies and sizes of the starting raw materials (Al-12Si, TiH2 and MWCNTs−COOH) used in this research work. The particles of Al-12Si powder exhibit a quasi-spherical or/and an oblong shape (Fig. 2a), while the particles of TiH2 powder present an angular shape (Fig. 2b). The as received MWCNTs−COOH in nanotubes appear entangled (Fig. 2c− −d). The size distributions of the powders Al-12Si (average diameter, D50 ~27.6 µm), TiH2 (as-received, D50 ~5.6 µm and pre-oxidised, D50 ~6.1 µm) are shown in Fig. 3a. Fig. 3b, while Fig. 3c compares the XRD patterns and the SEM micrographs of the as-received and preoxidised TiH2. The particle size (Fig. 3a) and the surface roughness (Fig. 3c) of the TiH2 particles increased with this pre-oxidising treatment in air (480 ºC, 180 min) due to the formation of an oxide layer 6

ACCEPTED MANUSCRIPT [14-15, 20] as confirmed by the newly formed phases (rutile, TiO3, TiH1.5) identified in the respective XRD pattern of Fig. 3b, contrasting to the single phase nature of the as-received TiH2 powder. The TG/DTA thermograms of Al-12Si and TiH2 (as-received and pre-oxidised) are displayed in Fig. 4a and Fig. 4b, respectively. The endothermic peak centred at 586 ºC in Fig. 4a corresponds

RI PT

to the melting temperature of the Al-12Si alloy and is accompanied by a steep weight gain due to oxidation that has started at ~500 ºC. The “S” shape of TG curve in the ~500−700 ºC region can be attributed to the formation of an oxide surface layer around the Al-12Si particles, slowing the oxidation kinetics even under increasing harsh conditions. The decomposition temperature of

SC

pre-oxidised TiH2 shifted from 291 ºC to 550 ºC (Fig. 4b), close to the melting temperature of the Al-based alloys (Fig. 4a). This raise in the decomposition threshold of the blowing agent is essential for minimizing the temperature mismatch relative to the melting of the alloy [20]. There

M AN U

was a loss of ~25% hydrogen during this pre-oxidising treatment. This is in good agreement with date presented in Fig. 3 and in several literature reports [2, 20, 23]. 3.2. Dispersing ability of the MWCNTs − influence of processing additives The ξ-potential measurements give an idea about the magnitude of the electrostatic repulsion between dispersed nanoparticles in a liquid. In the absence of steric repulsions, the threshold of

TE D

colloidal stability is usually observed for ξ-potential values ≥25−30 mV [25]. The poor wetting of carbon by water is a major barrier to overcome upon dispersing MWCNTs in aqueous media. This justifies their surface modification with hydrophilic functional −COOH groups and a

EP

detailed study about the influence of the processing additives. The ξ-potential versus pH curves for the MWCNTs−COOH in either the absence or in the

AC C

presence of NS, PVA, and NS+PVA are shown in Fig. 5a. No isoelectric point could be detected for the MWCNTs within the analysed pH range. The most negative ξ-potentials are observed in the presence of NS, reaching values of ~−39 mV within the pH range of ~7−10. The values measured in the absence of processing additives are only slightly less negative, proving the effectiveness of the surface functionalization. The ξ-potential values measured in the presence of PVA alone were ~−3 mV along the entire pH range. The main reasons for this include: (i) the non-ionic nature of this additive; (ii) the large hydrophilic PVA molecules adsorbed onto the surface through hydrogen bridges [26-27] and protruding to the solution are likely to decrease the electrophoretic mobility of the MWCNTs. The simultaneous addition of both processing

7

ACCEPTED MANUSCRIPT additives enhanced the absolute values of ξ-potential due to the anionic nature of NS, but did not annulled the slowing effect of the dandling adsorbed PVA molecules. Fig. 5b illustrates the effects of the same processing additives on the sedimentation behaviour of MWCNTs−COOH performed in deionized water and in a 1.5 wt.% PVA aqueous solution. The labels “1” and “2” in the graduated cylinders stand for the absence and the presence of NS,

RI PT

respectively. The MWCNTs are not stable in deionised water in the absence of NS as judged from the systematically less dark supernatant, which becomes transparent after 92 h, accompanied by distinct sediment of MWCNTs−COOH. In PVA solution the stability against sedimentation was enhanced as judged from the turbid aspect of the supernatants after a longer

SC

time period of 120 h. From these results, it seems clear that either NS or PVA alone contribute to the stability of the MWCNTs−COOH, keeping them apart in the suspension, but a synergetic

M AN U

effect is obtained when both are present. Attempts for dispersing MWCNTs−COOH in ethanol [14] or in aqueous PVA solutions alone [15, 16] were apparently less successful. In both cases, clusters of the agglomerated carbon nanotubes were reported.

3.3. Rheological characterisation of concentrated the suspensions Based on the information gathered and reported in the previous section, namely the compatibility

TE D

of PVA with MWCNTs−COOH, the starting Al-12Si and TiH2 powders were dispersed in a 1.5 wt.% PVA aqueous solution. The ability of PVA molecules to mitigate the trend of coarse particles to sediment under gravity is a further favourable feature of this additive. The curves of apparent viscosity versus elapsed time for suspensions containing 49−56 vol.% solids (99.4 Al-

EP

12Si + 0.6 TiH2, in wt.%) in the absence of NS are shown in Fig. 6a. Significant overall increments in apparent viscosity can be observed for φ >52 vol.%, and the variations along the

AC C

measuring time are also more accentuated under these conditions. Because of that, φ = 49 vol.% was fixed to investigate the effects of adding different amounts of NS on the flow ability. The results presented in Fig. 6b reveal that there was a gradual decrease of the apparent viscosity with increasing added amounts of NS from 0−0.96 wt.% at given measuring times (ti = 0−360 s), while the measured values became more stable. These results confirm the effectiveness of this dispersant for dispersing the starting powders, anticipating a good compatibility among all the components in the composite system, in agreement with ξ-potential and sedimentation data displayed Fig. 5, and findings reported elsewhere [21, 22]. The apparent viscosity at t0 and at t360 dropped for about 3 and 2 times, respectively, when changing from 0−0.96 wt.% NS. Based on those results, the amount of 0.96 wt.% NS was selected to prepare the suspensions for freeze8

ACCEPTED MANUSCRIPT granulation. Such suspensions should be able to flow through the spraying nozzle without clogging it in order to produce spherical droplets. On the other hand, the viscosity should not be too low to avoid particle segregation and the formation of granules of small size and low density. Thus, the selected experimental conditions represent a suitable balance among all the

RI PT

requirements that are relevant for the suspensions for freeze-granulation.

3.4. Freeze-granulation and characterisation of the obtained granules

The size and spherical morphology of the granules prepared by FG did not change with the composition. No obstruction of the spray nozzle was observed during the preparation of these

SC

granules, confirming that the contents of the processing additives (PVA and NS) have been appropriately adjusted. Fig. 7 shows a typical cumulative size distribution curve of the FG05 granules determined by sieving. Most of the granules have sizes <1000 µm.

M AN U

The average of three measurements of apparent density for FG-00 and FG-05 granules gave the following values: 2.545±0.191 g/cm3 and 2.572±0.051 g/cm3, respectively. Therefore, no significant differences in density can be observed between the granules derived from the two compositions. This means that MWCNTs−COOH have been effectively dispersed and did not disturb the flow behaviour of the suspension.

TE D

The SEM images of FG05 granules with different sizes presented in Fig. 8 (left side) illustrate their typical spherical morphology. The SEM images under higher magnifications on the right side reveal interesting details on how the MWCNTs−COOH are uniformly and randomly distributed within the granules. The large and round Al-12Si particles are also clearly visible.

EP

The SEM/EDX images of FG00 granules under different magnifications shown in Fig. 9 also reveal that the angular intermediate sized particles of TiH2 appear well-distributed. All these

AC C

results confirm the suitability of the freeze-granulation-lyophilisation an effective technique to preserve the high level of the homogeneity achieved during preparation of the high solids loaded aqueous suspensions containing all the components in spite of their different sizes, shapes and chemical nature.

The TG/DTA curves for FG00 and FG05 granules registered in air atmosphere (Fig. 10a) do not reveal any significant differences. This can be understood considering the small amount of added MWCNTs−COOH. The first mass loss of about 1.3% identified within the ~285–550 ºC range is essentially due to partial dehydration of PVA (see Fig. 4b) accompanied by polyene formation [28-29]. The sharp endothermic peak centred at 579 ºC is due to the thermal decomposition of TiH2, with the concomitant hydrogen mass loos. But suddenly after 579 ºC, an exothermic effect

9

ACCEPTED MANUSCRIPT superimposed and was accompanied by a weight gain due to oxidation of Ti and the formation of titanium oxide phases as shown in Fig. 3. Such phases could not be identified in the XRD patterns displayed in Fig. 10b probably because of the trace amounts formed upon heating up to 1000 ºC are below the detection limit. The same can be stated concerning the eventual formation of undesirable intermetallic compounds, e.g. Al4C3 - JCPDS 01-071-3787 (RDB) and 00-050-

RI PT

0740 (RDB). The chemical reactions between CNTs and metal matrices are likely to readily occur at high temperatures, leading to the formation of the interfacial intermetallic products that degrade the structural nanoscale reinforcements. This situation is particular serious for singlewall nanotubes since their tubular integrity is lost. Such undesirable intermetallic products will

SC

hardly form within the practical temperature range used for foaming (~700 ºC). Contrarily, the incorporation of CNTs into molten metallic-matrices is rather difficult due to the harsh fabrication conditions employed, and because such conditions favour their degradation either by

M AN U

direct oxidation or by the formation of intermetallic products [11-12]. The broad exothermic band observed within the ~700-900 ºC range can be attributed to the oxidation of the Al-12Si particles. This is consistent with the formation of the alumina that was clearly identified in the XRD patterns of Fig. 10.

TE D

3.5. Preparation and characterisation of foamable precursor specimens Preliminary compacting experiments at RT under different applied loads and time periods revealed that apparent density did not change much with time, enabling the duration of this step to be as short as 1 s. The specimens were then heat treated up to 400 ºC for thermal stabilisation

EP

and further hot pressed under different time durations. Earlier findings revealed that excessive dwell times at this temperature should be avoided in order to prevent premature thermal

AC C

decomposition of TiH2 [2].

Fig. 11 shows the overall aspect of PB, FG00 and FG05 specimens (30 mm diameter) and of their SEM microstructures. The specimens shown in Fig. 11a-c were consolidated under a load of 1.4 ton and the ones shown in Fig. 11d-f were consolidated under a load of 20 ton. While PB specimens could be safely consolidated under a load of 1.4 ton (Fig. 11a [20], higher loads were required when starting from spherical granules, FG00 and FG05 (Fig. 11g, Table 2). The doted circles in the optical micrographs (Fig. 11e-f) engulf some TiH2 particles and aim at underlining their homogeneous distribution. Apparent density data reported in Table 2 reveal that under a load of 1.4 ton the highest value was obtained for Al-12Si. Slightly lower density was registered for PB, and noticeably lower values were measured for FG00 and FG05 specimens. These results mean that: (i) the addition of a small amount of rigid TiH2 particles somewhat decreased the 10

ACCEPTED MANUSCRIPT packing ability of the PB in comparison to ductile Al-12Si alloy; (ii) the spherical granules of the same composition are more difficult to accommodate and smash; (iii) the incorporation of MWCNTs (FG05) does not significantly affect the apparent density of the specimens. As expected, for both FG00 and FG05 granules, increasing the applied load enhanced the compaction degree, without any significant difference being observed between non-reinforced

RI PT

and reinforced precursor samples, as illustrated in Fig. 11g. Accordingly, this load was included in the optimum pressing cycle (pre-compacting at RT for 1 s, heating up to 400 ºC within ~14 min, holding at this temperature for thermal stabilisation during 40 min, and then applying the load for further 10 min).

SC

Fig. 12 compares the fracture surfaces of the FG00 and FG05 precursor materials. Both confirm the relatively high compaction degrees achieved. But most interestingly, individual MWCNTs protruding from the fracture surface of FG05 can be clearly seen, contrasting with their complete

M AN U

absence in the case of the FG00 precursor.

3.6. Foaming of the precursor specimens and characteristics of the foams Heating the precursor specimens to temperatures slightly above the melting point of the metallic alloy allows it to expand under the internal pressure exerted by the hydrogen (H2) released from

TE D

the thermal decomposition of the blowing agent (TiH2), forming closed-cell metal foams. The metal expands developing a highly porous closed-cell internal structure due to the simultaneous melting of the aluminium and thermal decomposition of the blowing agent in gas (H2) [20]. The foamed parts have a highly porous internal structure and are covered by a dense metal skin that

EP

improves their mechanical properties and provides good surface finish [30]. The hot-stage microscopy (HSM) revealed to be a suitable technique to describe the foaming behaviour of a single precursor [24]. Fig. 13a shows a sequence of representative images taken

AC C

upon heating a small cube shaped specimen of the FG05 precursor. It can be seen that most of the sharp edges have almost disappeared at 671 ºC and that expansion is progressing rapidly, even considering the hindering effects derived from the small size of the sample and its higher surface area/volume ratio. The formed oxide surface layer tends to oppose expansion, delaying the gas release from the foaming agent. Their maximum expansion ratio [(A/A0)max] was 1.97. These observations are consistent with the thermal events identified in TGA/DTA analyses (Fig. 4), namely, the thermal decomposition of the pre-oxidized TiH2 at ~550 ºC (Fig. 4a), and the melting of the AlSi12 alloy at ~586 ºC (Fig. 4b). The SEM micrographs shown in Fig. 13b reveal microstructural details of a spherical pore wall surface of FG05 foam. Individualized and stretched MWCNTs can be seen randomly distributed 11

ACCEPTED MANUSCRIPT in the aluminium-matrix. This non-agglomerated condition confirms the effectiveness of the dispersion achieved in the colloidal processing step. The stretched condition is likely to be boosted upon foaming, being favourable to reinforcement. The results of Vickers microhardness determined on the struts (cell-walls) of FG05 foams varied within the range of 55-125% in comparison to the non-reinforced ones, with a mean value of 93.43±19.30 HV. These values

RI PT

were significantly superior to those of non-reinforced FG00 foams (mean = 60 HV±5.18 HV). These results confirm the reinforcing role of MWCNTs in this new generation of metal foams, conferring them enhanced mechanical properties. The high standard deviation of reinforced foams can be attributed to: (i) the small added amount (0.5 wt.%) of MWCNTs; (b) the

SC

neighbourhood between of the indentation point and the reinforcing MWCNTs. This means that hardness can randomly change from one point to another depending on the closeness vicinity of a reinforcing MWCNT. The more homogeneous structure of non-reinforced foams explains their

M AN U

small standard deviation; the same being applied to the AlSi12 specimen (Table 2). The tubular structure of MWCNTs was preserved throughout the entire the process. Several strengthening mechanisms have been already proposed for MWCNTs in metal-matrix composites (MMCs) [31-35], including load transfer from matrix to MWCNTs [30]; grain refining [32] and texture strengthening [33] by pinning effect of MWCNTs; dispersion

TE D

strengthening of MWCNTs [34]; solution strengthening of carbon atoms [35]; strengthening of in-situ formed or participant carbide particles [35]; and thermal mismatch between MWCNTs and matrix [6]. But the composite strength might be a synergetic result of several strengthening mechanisms although the specific contribute of each one is not easy to discriminate from these

EP

previous reports [31-35]. In an attempt to shed further light on this issue, Chen et al. [36] examined the failure behaviours of MWCNTs (produced by chemical vapour deposition) in an Al metal matrix composite reinforced with 0.6 wt.% MWCNTs prepared by a powder metallurgy

AC C

route. They performed in-situ tensile tests were by operating the tensile stage inside a FE-SEM chamber. The tensile sample was prepared by extrusion and machining. This in-situ advanced tensile testing technique enabled them concluding that the mechanical behaviour of MWCNTs in composites is essentially regulated through a load transfer strengthening mechanism. When a force is applied to the composite, the MWCNTs initially act like a bridge to suppress crack growth. As further force is applied, the outer walls of the nanotubes in contact with the Al matrix start to break. The inner walls then fracture, either breaking vertically or unpeeling to expose the next inner walls, and so on.

12

ACCEPTED MANUSCRIPT Our results are consistent with the findings of Chen et al. [36], demonstrating the feasibility of manufacturing nanocomposite metal foams reinforced with MWCNTs by the powder metallurgy method by adding a colloidal processing step to efficiently disperse the MWCNTs.

4. Conclusions

RI PT

The feasibility of an innovative method for fabricating a new class of closed-cell reinforced Alfoams reinforced with MWCNTs was demonstrated. The novel approach is a modification of the traditional powder metallurgy (PM) method by adding an advanced colloidal processing step that includes Freeze-Granulation-Lyophilisation. The entire process allows: (i) achieving an effective

SC

dispersion of the MWCNTs−COOH in aqueous media and their homogeneous mixing with the other components of the system; (ii) preserving the homogeneity and structural integrity (tubular structure) of the carbon nanotubes through the process; (iii) a strong bond between the

M AN U

MWCNTs and the metal matrix, which provides an efficient load transfer. Accordingly, in comparison to the non-reinforced (FG00) Al-foams, the mean values of Vickers microhardness of reinforced (FG05) ones increased within the range from 55-125%, depending on the neighbourhood between of the indentation point and the reinforcing MWCNTs. Further research efforts will be required to systematically investigate the effect of other relevant experimental

TE D

variables not covered in this such as: (i) the reproducibility and overall quality control of the foaming process; (ii) determining the maximum allowable content MWCTs that can be incorporated without degrading the foaming process or the quality of the foams; (iii) applying the method to other Al-alloys and metallic systems; (iv) evaluating the properties and

EP

performance of the foams in high-tech applications and demonstrate their unique and

AC C

competitive advantages over their existing counterparts.

Acknowledgments

This work was supported by TEMA (CC/ID/CS/2.30.400.20.12, Centre for Mechanical Technology and Automation, www.ua.pt/tema). This work was also supported by the European Regional Development Fund (FEDER) through the COMPETE, by the Portuguese Government through the Portuguese Foundation for Science and Technology (FCT), in the scope of the projects UID/CTM/50011/2013 (Aveiro Institute of Materials, CICECO, www.ciceco.ua.pt)”.

13

ACCEPTED MANUSCRIPT References [1] Banhart J. Light-Metal Foams—History of Innovation and Technological Challenges Adv Eng Mater 2013; 15(3): 82–111. [2] Duarte I, Banhart J. A study of aluminium foam formation—kinetics and microstructure.

RI PT

Acta Mater 05/2000; 48(9): 2349-2362. [3] Banhart J. Manufacture, characterisation and application of cellular metals and metal foams. Prog Mater Sci 2001; 46(6): 559–632.

[4] Duarte I, Vesensak M, Krstulović-Opara L, Anžel I, Ferreira JMF. Manufacturing and

SC

bending behaviour of in situ foam-filled aluminium alloy tubes. Mater Des 02/2015; 66:532– 544.

M AN U

[5] Duarte I, Vesenjak M, Krstulović-Opara L, Ren Z. Static and dynamic axial crush performance of in-situ foam-filled tubes. Compos Struct 2015; 124: 128-139. [6] Baumeister J, Weise J. Metal Foams. In: Lehmhus D, Busse M, Herrmann AS, Kayvantash K. editors. Structural Materials and Processes in Transportation, Weinheim, Germany, WileyVCH Verlag GmbH & Co. KGaA; 2013, p.415-444.

TE D

[7] Banhart J, Seeliger, H-W. Aluminium foam sandwich panels: manufacture, metallurgy and applications. Adv Eng Mater 2008; 10(9): 793.

[8] Harris PJF. Carbon nanotube science: synthesis, properties and applications. Cambridge University Press. 2nd edition. Cambridge. 2011.

EP

[9] Tjong SC. Structural and mechanical properties of polymer nanocomposites. Materials Science and Engineering: R: Reports 2006. 53(3–4): 73–197.

AC C

[10] Nesher G, Serror M, Avnir D, Marom Gad. Silver coated vapor-grown-carbon nanofibers for effective reinforcement of polypropylene–polyaniline. Compos Sci Technol 2011; 71(2): 152–159.

[11] Tjong C. Recent progress in the development and properties of novel metal nanocomposites reinforced with carbon nanotubes and graphene nanosheets. Mater Sci Eng R 2013; 74: 281-350. [12] Thostenson ET, Ren Z, Chou T-W. Advances in the science and technology of carbon nanotubes and their composites: a review. Compos Sci Technol 2001; 61(13): 1899–1912. [13] Esawi A, Morsi K. Dispersion of carbon nanotubes (CNTs) in aluminum powder. COMPOS PART A- APPL Sci Manuf 2007; 38(2): 646–50. 14

ACCEPTED MANUSCRIPT [14] Deng C, Zhang X, Wang D, Lin Q. A Li.Preparation and characterization of carbon nanotubes/aluminum matrix composites. Mater Lett 2007; 61(8–9): 1725–1728. [15] Jiang L, Li Z, Fan G, Cao L, Zhang D. The use of flake powder metallurgy to produce carbon nanotube (CNT)/aluminum composites with a homogenous CNT distribution. Carbon 2012; 50(5): 1993–1998.

RI PT

[16] Jiang L, Fan G, Li Z, Kai X, Zhang D, Chen Z, Humphries S, Heness G, Yeung WY. An approach to the uniform dispersion of a high volume fraction of carbon nanotubes in aluminum powder. Carbon 2011; 49 (6): 1965–1971.

[17] Liu ZY, Xiao BL, Wang WG, Ma ZY. Singly. Dispersed carbon nanotube/aluminum

SC

composites fabricated by powder metallurgy combined with friction stir processing. Carbon 2012; 50 (5): 1843–1852.

M AN U

[18] Liu ZY , Xiao BL, Wang WG, Ma ZY. Analysis of carbon nanotube shortening and composite strengthening in carbon nanotube/aluminum composites fabricated by multi-pass friction stir processing. Carbon 2014; 69: 264–274

[19] Liu ZY , Xiao BL, Wang WG, Ma ZY. Developing high-performance aluminum matrix composites with directionally aligned carbon nanotubes by combining friction stir processing

TE D

and subsequent rolling. Carbon 2013; 62: 35–42.

[20] Duarte I, Oliveira M. Aluminium alloy foams: Production and properties. In: Powder Metallurgy. Kondoh K, editor. Rijeka. InTech; 2012. p.47-72.

EP

[21] Duarte I, Ventura E, Olhero S, Ferreira JMF. A novel approach to prepare aluminium-alloy foams reinforced by carbon-nanotubes. Materials Letters 2015; 160: 162-166.

AC C

[22] Singh MK, Shokuhfar T, Gracio JJ, De Sousa ACM, Ferreira JM, Garmestani H, et al. Hydroxyapatite modified with carbon-nanotube-reinforced poly(methyl methacrylate): A nanocomposite material for biomedical applications. Adv Funct Mater 2008; 18(5): 694-700. [23] Baumgartner F, Duarte I, Banhart J, Industrialization of powder compact foaming process. Adv Eng Mater 2000; 2(4): 168 - 174. [24] Duarte I, Ferreira JMF. 2D Quantitative analysis of metal foaming kinetics by hot-stage microscopy. Adv Eng Mater 01/2014; 16(1): 33-39. [25] Hunter RJ, Ottewill RH, Rowell RL. Zeta potential in colloid science. Principles and applications. 1981 Academic Press. Elsevier.

15

ACCEPTED MANUSCRIPT [26] Wiśniewska M. The temperature effect on electrokinetic properties of the silica-polyvinyl alcohol (PVA) system. Colloid Polym Sci 2011; 289(3): 341–344. [27] Francisco E, Cruz D, Zheng Y, Torres E, Li W, Song W. Zeta potential of modified multiwalled carbon nanotubes in presence of poly (vinyl alcohol) hydrogel. Int J Electrochem Sci 2012; 7: 3577–90.

RI PT

[28] Y Tsuchiya, K Sumi. Thermal decomposition products of poly (vinyl alcohol). J Polym Sci Part: A-1: Polym Chem 1969; 7(11): 3151-3158.

[29] Tubs RK, Ting KW. Thermal properties of polyvinyl alcohol. In: Finch CA (Ed.), Polyvinyl

SC

alcohol-properties and applications, London: John Wiley & Sons; 1973, pp. 167–182.

[30] Duarte I, Vesenjak M, Krstulović-Opara L. Variation of quasi-static and dynamic

182. [31] George R,

Kashyap

KT,

M AN U

compressive properties in a single aluminium foam block. Mater Sci Eng A 10/2014; 616:171-

Rahul

R,

Yamdagni

S.

Strengthening in

carbon

nanotube/aluminium (CNT/Al) composites. Scr Mater 2005; 53: 1159–1163. [32] Nam DH, Cha SI, Lim BK, Park HM, Han DS, Hong SH. Synergistic strengthening by load transfer mechanism and grain refinement of CNT/Al–Cu composites. Carbon 2012; 50: 2417–23.

TE D

[33] Wei H, Li Z, Xiong D-B, Tan Z, Fan G, Qin Z, Zhang D. Towards strong and stiff carbon nanotube-reinforced high-strength aluminum alloy composites through a microlaminated architecture design. Scr Mater 2014; 75: 30–3.

[34] Yoo SJ, Han SH, Kim WJ. Strength and strain hardening of aluminum matrix composites

711–4.

EP

with randomly dispersed nanometer-length fragmented carbon nanotubes. Scr Mater 2013; 68:

AC C

[35] Li S, Sun B, Imai H, Kondoh K. Powder metallurgy Ti–TiC metal matrix composites prepared by in situ reactive processing of Ti-VGCFs system. Carbon 2013; 61: 216–28. [36] Chen B, Li S, Imai H, Jia L, Umeda J, Takahashi M, Kondoh K. Load transfer strengthening in carbon nanotubes reinforced metal matrix composites via in-situ tensile tests. Composites Science and Technology 2015; 113: 1-8.

16

ACCEPTED MANUSCRIPT Figure Captions Fig. 1. A schematic comparison between the simple traditional powder metallurgy method of producing non-reinforced closed-cell metal foams (a), and the novel approach that includes a colloidal processing step in the former powder metallurgy method to produce reinforced closedcell metal foams (b).

alloy; (b) TiH2 powder; (c, d) as-received MWCNTS-COOH.

RI PT

Fig. 2. SEM morphological assessment of the particles of the starting raw materials: (a) Al-12

Fig. 3. Characteristics of the starting powders: (a) Particle size distributions of Al-12Si, and of TiH2 as-received and pre-oxidised; (b) XRD patterns, and (c) SEM micrographs of as-received

SC

and pre-oxidised TiH2.

Fig. 4. TG/DTA thermograms of the powders: (a) Al-12Si; (b) as-received and pre-oxidised

M AN U

TiH2.

Fig. 5. Dispersing ability of MWCNTs−COOH in aqueous media in the absence and in the presence of NS, PVA or NS+PVA: (a) Zeta potential versus pH; (b) Sedimentation tests − the labels “1” and “2” in the graduated cylinders stand for the absence and the presence of NS, respectively.

TE D

Fig. 6. Influence of the processing parameters on the apparent viscosity of (Al-12Si + 0.6 wt.% TiH2) suspensions dispersed in 1.5 wt.% PVA aqueous solution: (a) Effect of solids loading (absence of NS); (b) Effect of added amount of NS.

EP

Fig. 7. Cumulative size distribution curve of the spherical granules obtained by FG as determined by sieving.

AC C

Fig. 8. SEM micrographs of lyophilised FG05 granules of different sizes revealing their typical spherical morphology (left side). Detailed aspects of the surface of the granules showing the uniform distribution of MWCNTs−COOH (right side). Fig. 9. SEM micrographs of FG00 granules: (a) overall spherical morphology; (b-c) surface details under higher magnifications; (c) EDX elemental mapping revealing the TiH2 particles onto the oblong Al-alloy ones. Fig. 10. Influence of MWCNTs on thermal behaviour and crystalline phase assemblage after heat treating the FG00 and FG05 up to 1000 ºC in air atmosphere at a constant heating rate of 10ºC/min: (a) TG/DTA curves; (b) X-ray diffraction patterns.

17

ACCEPTED MANUSCRIPT Fig. 11. Foamable precursor specimens (30 mm in diameter) derived from PB (a, d); FG00 (b, e); FG05 (c, f). The materials were compacted under different applied loads of: 1.4 ton (a, b, c); and 20 ton (d, e, f). Variation of the density and the Vickers hardness of the precursor specimens as a function of the applied load (g). An indentation in the metal matrix of a foamable precursor material (h).

RI PT

Fig. 12. Fracture surfaces of precursor materials prepared from the granules: (a) FG00; (b) FG05. Fig. 13. (a) Sequence of representative HSM images of the FG05 precursor upon heating at 10ºC/min illustrating its foaming behaviour; (b) SEM micrographs of the internal surfaces of cellular pores of FG05 foam, showing the uniform distribution of MWCNTs in the metallic

AC C

EP

TE D

M AN U

SC

matrix.

18

ACCEPTED MANUSCRIPT Table Captions Table 1. Characteristics of the starting Al-alloy and TiH2 powders. Table 2. Densities and Vickers micro hardness (HV) of foamable precursor materials derived from FG00 and FG05 granules as function of the applied load. The presence of 0.5 wt.% of

AC C

EP

TE D

M AN U

SC

RI PT

MWCNTs did not affect these properties.

19

ACCEPTED MANUSCRIPT Table 1. Powder Al-12Si Powder

Al 86.65 Purity (%) 98.8

Si 12.46

Main alloying elements (wt.%) Fe Cu Mn Zn Ti 0.41 0.156 0.088 0.059 0.043 Impurities (wt. %) H (min.) N Fe Cl Ni Si 3.8 0.3 0.09 0.06 0.05 0.15

Mg 0.07

Ti (min.) 95

Cr 0.029

Zr 0.005

Mg 0.04

C 0.03

AC C

EP

TE D

M AN U

SC

RI PT

TiH2

Particle size (µm) D10 D50 D90 7.98 27.65 61.71 Particle size (µm) D10 D50 D90 1.1 6.2 13.5

1

ACCEPTED MANUSCRIPT Table 2.

Applied load (ton) 1.4 HV± standard deviation

2.58 ± 0.040

88.6 ± 13.1

2.54 ± 0.085

20 HV± standard deviation

density ± standard deviation (g/cm3)

HV± standard deviation

–

–

–

–

–

–

–

2.64 ± 0.004

67.6 ± 1.5

2.01 ± 0.044

42.32 ± 2.2

2.44 ± 0.074

63.7 ± 1.4

2.54 ± 0.017

67.8 ± 2.6

2.03 ± 0.045

42.34 ± 2.1

2.47 ± 0.060

65.6 ± 1.8

2.54 ± 0.030

69.2 ± 2.7

AlSi12 AlSi12 + TiH2

AC C

EP

TE D

M AN U

SC

Granules FG00 (without MWCNTs) Granules FG05 (with MWCNTs)

10 density ± standard deviation (g/cm3)

density ± standard deviation (g/cm3)

RI PT

Precursor foamable material

1

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT