Aluminum composite reinforced with multiwalled carbon nanotubes ...

9 downloads 37899 Views 4MB Size Report
b AMPAC and Nanoscience Technology Center, University of Central Florida, Orlando, FL 33816, ... Surface & Coatings Technology 203 (2009) 1544–1554.
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy Surface & Coatings Technology 203 (2009) 1544–1554

Contents lists available at ScienceDirect

Surface & Coatings Technology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s u r f c o a t

Aluminum composite reinforced with multiwalled carbon nanotubes from plasma spraying of spray dried powders Srinivasa R. Bakshi a, Virendra Singh b, Sudipta Seal b, Arvind Agarwal a,⁎ a b

Plasma Forming Laboratory, Nanomechanics and Nanotribology Laboratory, Department of Mechanical and Materials Engineering, Florida International University, Miami, FL 33174, USA AMPAC and Nanoscience Technology Center, University of Central Florida, Orlando, FL 33816, USA

a r t i c l e

i n f o

Article history: Received 19 September 2008 Accepted in revised form 4 December 2008 Available online 13 December 2008 Keywords: Carbon nanotubes Aluminum composites Dispersion Nanoindentation Spray drying Plasma spraying

a b s t r a c t Homogenous dispersion of carbon nanotubes (CNTs) in micron sized aluminum silicon alloy powders was achieved by spray drying. Excellent flowability of the powders allowed fabrication of thick composite coatings and hollow cylinders (5 mm thick) containing 5 wt.% and 10 wt.% CNT by plasma spraying. Two phase microstructure with matrix having good distribution of CNT and CNT rich clusters was observed. Microstructural evolution has been explained using single splat and the infiltration of CNT clusters by liquid metal. Partial CNT surface damage was observed in case of the 10 wt.% CNT coating due to CNT mesh formation and smaller size of spray dried agglomerate. Increase in the elastic modulus and improvement in the yield strength and elastic recovery properties due to CNT addition was observed by nanoindentation. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Metal matrix composites have been proposed for many automobile [1–4] and space applications [4,5] due to their superior mechanical properties like higher specific strength and stiffness, desirable coefficient of thermal expansion and good damping properties [6]. Multiwalled carbon nanotubes (CNTs) possess unique properties like high stiffness (~ 970 GPa), strength (~ 63 GPa) [7] and thermal conductivity up to 3000 W m− 1 K− 1 [8]. Excellent mechanical and thermal properties of CNTs along with their low density make them ideal as reinforcements for composites. Polymer [9–11] and ceramic [12–15] matrix composites have been prepared with CNTs as reinforcement in order to improve the mechanical and/or thermal properties. Although majority of the research work on development of CNT composites have been carried out using polymers as matrices due to the relative ease of fabrication. Metal matrix composites with CNT reinforcement have generated significant interest in last five years. Composites have been prepared with metal/alloy matrices of Al [16–19], Cu [20–23], Ni [24–26], Mg [27–29] and metallic glasses [30,31]. Several fabrication routes namely powder metallurgy, ball milling, extrusion, hot pressing, equalchannel angular pressing, spark plasma sintering, electro-deposition, electro-less deposition and thermal spraying have been used to synthesize metal matrix composites with CNTs. Chen et al. [22] have observed a decrease in the wear loss by 80% and coefficient of friction ⁎ Corresponding author. E-mail address: agarwala@fiu.edu (A. Agarwal). 0257-8972/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2008.12.004

by 40% by addition of CNTs in Ni-P-CNT coatings prepared by electroless deposition technique. Cha and co-workers [21] have observed a 200% increase in the yield strength in 10 vol.% CNT reinforced Cu composites prepared by a molecular level mixing method, highlighting the significance of dispersion in effective strengthening. Noguchi et al. [16] have reported a sevenfold increase in compressive yield strength in 1.6 vol.% CNT reinforced Al composites prepared by a nano-scale dispersion method. Zhou and co-workers [17] have shown a 30% decrease in the wear rate and friction coefficient of Al-CNT alloys by addition of 20 vol.% of CNTs. Tang et al [32] have shown a 65% decrease in coefficient of thermal expansion of aluminum composite reinforced with 15 vol.% single walled CNTs. These studies indicate that there is lot of scope for strengthening and enhancement of properties by addition of CNTs. In the previous work from our research group [13,14,33], it has been shown that plasma spraying can be successfully used to fabricate bulk CNT reinforced metal and ceramic composites. Blending was used by Laha et al. [33] to mix carbon nanotubes and aluminum-23 wt.% silicon alloy powder. However, this lead to poor dispersion and clogging of the powder feeding tubes, due to which the thickness of the cylindrical shell fabricated by plasma spray forming was limited to 0.635 mm [33]. The goal of the present work is to improve the dispersion of the CNTs in the aluminum composite by employing spray dried powders to enable fabrication of thick coatings and bulk freestanding structure. The microstructure evolution of the coatings from spray dried powders is reported with the help of the single splat structure. The change in mechanical properties with CNT addition is also investigated.

Author's personal copy S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

1545

of the particles at 100 mm distance from the plasma gun, which is where they strike on the substrate to form the coating.

2. Experimental details 2.1. Powder feedstock

2.3. Microstructure and phase characterization Aluminum–silicon (Al–Si) eutectic alloy powder of composition Al-11.6% Si-0.14% Fe by weight obtained by inert gas atomization process and having a mean particle size 2.4 ± 1.2 µm (D90 = 3.8 µm) was obtained from Valimet Inc. (Stockton, CA, USA). Multiwalled carbon nanotubes (CNT), obtained from Inframat Advanced Materials (Willington, CT, USA), had a purity of more than 95% and diameter of 40–70 nm and length 1–3 µm. Al–Si powder and CNTs were mixed to make aqueous slurry with poly-vinyl alcohol (PVA) as binder. Two compositions of 5 wt.% CNT and 10 wt.% CNT were prepared. The slurry was then subjected to spray drying. During spray drying [34], the slurry is atomized into droplets in a chamber, with hot gas flowing in counterclockwise direction, where the droplets dry to form large agglomerates. The 5 wt.% and 10 wt.% composite agglomerates obtained were named SD Al-5CNT and SD Al-10CNT respectively. Gas atomized Al–Si powder of the same composition as above but particle size 14 ± 9 µm (D90 = 26 µm) was used for fabrication of pure Al–Si coating without any nanotubes and will be referred to as Al–Si hereafter. This coating was used as a reference to estimate strengthening effect due to CNT addition. 2.2. Plasma spraying The powders described above were plasma sprayed onto mild steel substrate to form coatings with a thickness varying from 0.5–5.5 mm. SG-100 gun (Praxair Inc., Danbury, Connecticut, USA) was used for spraying. It has a thoriated tungsten cathode (Praxair # 02083-730) and a concentric copper anode (Praxair # 01083A-720). The anode plate has an orifice of 3.125 mm and the powders are injected through a feed tube of 1.5 mm internal diameter. The plasma power was ~ 22 kW and the primary, secondary and career gases used along with their flow rates were Ar (42.5 slpm), He (30.5 slpm) and Ar (11.9 slpm) respectively and the gun traverse velocity was 25 mm s− 1. Powder was fed internally into the plasma plume. The mild steel substrate was grit blasted and kept at a distance of 100 mm from the gun for spraying thick coatings. The powder feed rates for the Al–Si, SD Al-5CNT and SD Al-10CNT powders were calculated to be 12, 7 and 6 g/min respectively for the chosen carrier gas flow. The microstructural evolution of coatings, mainly the clustering tendency of the CNTs, was studied using single splat of the powder particles. Single splat was obtained by making a single sweep of the plasma gun over the glass substrate with same processing conditions. In-flight particle velocity and temperature was measured using the Accuraspray sensor (Tecnar Automation LTEE, QC, Canada). It contains a dual fiber device which collects signals from particles at two different locations and then uses cross correlation to measure the time delay between the two signals. From the known distance between the fibers the velocity is calculated. The minimum velocity that can be measured in this manner is 5 m/s with 0.5% precision. Temperature is measured using two color pyrometry which is based on measuring the intensity of the signal at two different wavelengths. Subsequently, the temperature of the radiating body is obtained using Planck's Law and Wein's approximation.   e λ ;T  5 I λ ln eð 1 Þ λλ21 Ið 2 Þ ðλ2 ;T Þ ðλ Þ 1   1 = 1 1 T C2 λ1 − λ2

ð1Þ

where λ1 and λ2 are the two wavelengths, C2 is a constant (=1.4388 cm K) and ε(λ,T) is the spectral emissivity. The minimum temperature that can be measured is 900 °C and the precision is 0.1%. The detector was placed so as to measure the velocity and temperature

Morphology of the spray dried composite powder was examined using secondary electron imaging mode using a JEOL JSM 630F scanning electron microscope (SEM). Samples were prepared by sprinkling the powder on an adhesive tab. To examine coating cross sections, samples were cut using a low speed cutting saw and were mounted in the resin. They were then ground and polished to a 0.1 µm finish using diamond suspension. Polished cross sections were etched in Keller's reagent (5 ml HNO3, 3 ml HCl, 2 ml HF and 190 ml H2O). Density was measured by the water immersion method. The coating was cut from the substrate using low speed cutting saw resulting in free standing composite. Coating fracture surfaces were prepared by breaking a part of the free standing coating under tension and observing inside SEM in such a manner that the fracture surface was at an angle to the normal. The length and diameter of CNTs were measure from representative SEM micrographs using Image J. Length and diameter were reported as the average of 20–50 measurements. XRD was carried out using CuKα (λ = 1.542 Å) radiation in a Siemens D-500 X-ray Diffractometer operating at 40 kV and 20 mA at a scan rate of 1.2 deg/min. A Philips/FEI Tecnai F30 field emission gun transmission electron microscope (TEM) operating at an accelerating voltage of 300 kV was used to study the high resolution microstructure of the matrix and the nanotubes. The samples for TEM were in the form of a 3 mm disc, which were punched from the coating less than 100 µm thick prepared by grinding on a 600 grit abrasive paper, followed by dimpling at the center using a dimple grinder (Model 656 Mk3, Gatan, Inc., CA, USA). The final thinning was carried out by twinjet polishing (Model 110, E.A. Fischione Instruments, Inc., PA, USA) using a 30 vol.% mixture of 6N HNO3 in ethanol as the electrolyte until a hole was formed. Micro-Raman spectroscopy was carried out in the backscattering mode using an Argon ion laser of wavelength 514.5 nm and 18 mW to compare CNT structure in powder and coating. For testing powders, they were sprinkled on an adhesive tape and subjected to the laser beam. Mounted samples used for optical metallography were also used for micro-Raman spectroscopy. The laser beam was focused on the matrix and the CNT clusters to record the data from different regions. 2.4. Mechanical property characterization Microhardness of the thick coatings deposited on mild steel substrate was measured using a Vickers microhardness Tester (Shanghai Taiming Optical Instrument Co. Ltd., model HXD-1000 TMC, Shanghai, China) A load of 200 g was applied for a dwell time of 15 s for the purpose. The average value of at least 6 indents is reported here. Nanomechanical properties were measured by carrying out nanoindentation using a Hysitron Triboindenter (Hysitron Inc., Minneapolis, MN, USA). A Berkovich type diamond indenter having tip radius of 100 nm was used. Indentations were carried out at loads of 2000 µN, 3000 µN and 4000 µN. The load was applied linearly up to the maximum load in 5 s followed by a halt of 2 s at the maximum load followed by unloading in 5 s. Nine indents were made for each load on the matrix part of the nanocomposite coating, which makes it a total of 27 values of hardness and elastic modulus per sample. A total of 27 indents at the three loads were sufficient to get the average value of the properties. The reduced elastic modulus and hardness of the coatings were computed using Oliver and Pharr method [35]. Scanning Probe Microscopy (SMP) image of the indent was generated using the same indenter tip. SPM images were processed using the image processing software SPIP™ (Image Metrology A/S, Horsholm, Denmark).

Author's personal copy 1546

S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

3. Results and discussions 3.1. Carbon nanotube dispersion in powder feedstock Powder feedstock morphology and its characteristics have very significant effect in thermal spraying. The distribution of CNTs in the powder feedstock will influence the coating characteristics which will be reflected in the microstructure. Fig. 1a and b shows the SEM micrographs of the spray dried spherical agglomerates of SD Al-5CNT and SD Al-10CNT powders. The particle size distribution of both powders is shown in the inset and found to be 57 ± 21 µm (D90 = 80 µm) and 39 ± 15 µm (D90 = 54 µm) respectively. Spray drying

results in formation of large spherical agglomerates of sizes up to 50 times of the constituent particles. Spherical shape of the particle causes low inter-particle friction and hence leads to excellent flowability. Fig. 1c and e shows the high magnification SEM image of the outer surface of SD Al-5CNT and SD Al-10CNT agglomerate. Fig. 1d and f shows the inside of broken agglomerate of SD Al-5CNT and SD Al-10CNT powders, respectively. It is concluded from Fig. 1c and d that in case of SD Al-5CNT, CNTs are distributed uniformly on the surface as well as inside of the agglomerate. From Fig. 1e, it can be seen that there is dense network of CNTs forming a mesh on the outer surface of the SD Al-10CNT agglomerate that holds Al–Si particles on the surface. There is no mesh formation on the inside of the agglomerate (Fig. 1f).

Fig. 1. SEM micrographs showing SEM images of a) SD Al-5CNT powder agglomerates, b) SD Al-10CNT powder agglomerates, c) outer surface of a single SD Al-5CNT powder, d) inside view of fractured SD Al-5CNT powder, e) outer surface of a single SD Al-10CNT powder, and f) inside view of fractured SD Al-10CNT powder. Inset of a and b shows the powder size distribution.

Author's personal copy S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

During spray drying, where the CNT-metal slurry is atomized into droplets, the CNTs have a tendency to segregate on the surface of the droplets, owing to their non wetting property. Shrinkage caused by drying of the droplets brings CNTs closer. Also during drying, the movement of vapor from inside the droplet to outside would lead to transport of low density CNTs to the surface. This leads to mesh formation in the SD Al-10CNT powder. Mesh formation is not seen in SD Al-5CNT due to the lower concentration in CNTs. In powder metallurgy route, dispersion of the carbon nanotubes in the starting powder is one of the most critical steps that affect the CNT distribution in the final product. Physical blending [33], nanoscale dispersion [16], molecular level mixing [21] and mechanical alloying [36] have been used for dispersing CNTs in the metallic powder. Blending often uses a turbula mixer or a rotating mill which is not effective in dispersing CNTs in the matrix powder [36]. It has been reported that ball milling of the blended aluminum-CNT powders [36] for 48 h resulted in excellent dispersion. However, the particle size reached in excess of 1 mm after the ball milling process, which cannot be used for thermal spraying as well as most powder metallurgy based fabrication methods. Also, CNTs break and undergo damage due to impact from the milling media. Nanoscale dispersion method [16] is effective in dispersing CNTs in larger particles (~28 µm), which are an order of magnitude larger than those used in present study. Since the starting powder in our study is finer (1–3 μm), the CNT dispersion can be considered at a better resolution in the present case of spray drying. Molecular level mixing involves dispersion of CNTs in the salt solution followed by evaporation and calcination to form fine oxide dispersed with CNTs. The oxide is then reduced to metal with hydrogen. Though molecular level mixing provides promising results in terms of CNT dispersion [21], it is a multi-step process which is time consuming and expensive. In addition, there is a probability of retention of oxide due to incomplete reduction leading to impurities. Spray drying not only provides improved CNT dispersion but also improves the flowability of the powder due to the spherical nature of the agglomerates which is very advantageous in thermal spraying. Thus we can conclude that spray drying offers a simple and economical process for dispersing CNTs uniformly at the scale of a micrometer which can be utilized for large scale manufacturing of CNT reinforced nanocomposites by thermal spraying as well as other methods.

1547

Fig. 3. SEM micrograph of plasma sprayed a) Al-5CNT and b) Al-10CNT coatings. Inset shows Raman spectra from the coating matrix and CNT clusters within the matrix. Raman spectrum of the corresponding powder feedstock is also included.

3.2. Microstructural characterization of the coatings Fig. 2 shows a composite cylinder made by plasma spraying of SD Al-10CNT powder. The spraying was done on a 1 in. internal diameter ordinary carbon steel pipe whose surface was grit blasted. Attempts to remove the cylindrical coating have not been made. The dimensions of spray deposited cylinder are: 27 mm inner diameter, 5 mm wall thickness and 100 mm length. This cylinder was spray formed in

Fig. 2. Plasma spray formed Al-10CNT cylinder with a wall thickness of 5 mm and 100 mm length.

20 min which proves that composite with high (10 wt.%) CNT content can be rapidly fabricated by plasma spraying. Similar cylindrical structures were spray deposited using SD Al-5 CNT and Al–Si powder. It is emphasized that thick coatings could be deposited due to spray dried spherical agglomerates which assist plasma spraying process without clogging the powder feeding system. Thick coating was not deposited in our previous work [33] where blended CNT and aluminum powder feedstock caused clogging and inconsistent powder flow due to high surface forces between CNTs. 3.2.1. Coating microstructure and CNT dispersion Fig. 3 shows the SEM micrograph of the cross section of Al-5CNT and Al-10CNT coatings. Both coatings are uniform, dense and adherent to the substrate. Clusters of CNTs are also seen in both the coatings. The degree of CNT clustering is higher in the Al-10CNT coating as compared to Al-5CNT coating. The higher degree of CNT clustering in Al-10CNT coatings is attributed to CNT mesh formation in the corresponding SD Al-10CNT powder. As the spray dried powder gets molten, the poly vinyl alcohol binder evaporates. The surface tension of molten Al–Si alloy generates capillary forces which causes agglomeration of CNTs and the collapse of mesh structure leading to the formation of CNT cluster. This phenomenon is depicted schematically in Fig. 4. The degree of CNT clustering is low in Al-5CNT coating due to lower concentration of CNTs in SD Al-5CNT powder. The porosity present in the coatings is located in the vicinity of CNT

Author's personal copy 1548

S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

Fig. 4. Schematic of molten droplet obtained from agglomerated particle showing clustering due to surface tension forces.

clusters. The densities of the Al–Si, Al-5CNT and Al-10CNT coating as measured by water immersion technique was found to be 2.44, 2.36 and 2.35 g cm− 3 which corresponds to a density of 90%, 88% and 90% of the theoretical density, respectively. The inset in Fig. 3 shows the micro-Raman spectra taken from the matrix, clustered CNT region of the cross section of the coating and corresponding spray dried powder agglomerate. The micro-Raman spectra of the CNT cluster in Al-5CNT coating shows a strong Si peak at a Raman shift of 520 cm− 1 while the CNT cluster in Al-10CNT has a weak Si peak. This indicates that the CNT cluster in Al-5NT was metal infiltrated while the cluster in Al-10CNT was not. The presence of D and G peaks in the powder, coating matrix and CNT clusters indicates that the CNT structure remains intact after spray drying and plasma spraying. Uniform dispersion of CNTs in the matrix is necessary in order to achieve uniform properties and strengthening of the splats. Fig. 5 shows the SEM image of the fracture surfaces of Al-5CNT and Al-10CNT coatings. Fig. 5a shows an almost uniform distribution of CNTs within the splats in Al-5CNT coating. Fig. 5b shows CNTs distributed between splats and a CNT rich cluster in Al-10CNT coating. It is observed that clustering tendency is more in Al-10CNT due to the mesh formation in SD Al-10CNT powder.

powder. This is in accordance with the starting powder sizes. The effect of the CNT content is two-fold on splat formation. Firstly, it increases the thermal conductivity of melt which leads to higher heat transfer and a faster solidification rate. This could be another reason why the extent of spreading and the splat sizes are smaller for Al-10CNT due to high CNT content. Secondly, CNTs increase the heat content of the droplet. This is

3.2.2. Single splat microstructure and CNT-metal wetting To understand the mechanism of evolution of the microstructure, the single splats obtained from individual particles were analyzed in SEM. Figs. 6 and 7 shows the SEM images of individual splats of SD Al-5CNT and SD Al-10CNT powder respectively. Two kinds of splat morphologies namely fingered and disc shaped splats are observed for the SD Al-5CNT powder as shown in Fig. 6a and e respectively. Fingered splats are smaller than the disc splats indicating that they are formed from smaller powder particles. The shape of the splats is governed by the viscosity and thermal conductivity of the droplet which are dependent on the CNT content. Viscosity of the droplet from composite particles can be given as [37] μe =

μ 1−ϕð1 + RC =RP Þ

ð2Þ

Where µe is the effective dynamic viscosity, µ is the dynamic viscosity of the liquid phase, RC is the radius of CNT clusters and the RP is the droplet size and ϕ is the CNT fraction. Considering RC/RPbb1, Eq. (2) predicts 6.5% and 13.8% increase in viscosity of the molten Al–Si alloy due to addition of 5 wt.% and 10 wt.% of CNTs respectively. Most of the splats in Al-10CNT are disc shaped which is due to the increase in the viscosity of the melt caused by increased CNT content. The splat sizes ranged from 70–190 µm for Al-5CNT and 50–160 µm for Al-10CNT

Fig. 5. SEM image of a) fracture surface of Al-5CNT showing uniformly distributed CNTs in the splats, b) fracture surface of Al-10CNT coating showing CNTs between splats and a CNT rich cluster.

Author's personal copy S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

1549

Fig. 6. SEM images showing a) Fingered splat from SD Al-5CNT powder, b) splat finger showing CNT cluster infiltrated with metal marked by rectangles, c) and d) high magnification images of the CNT infiltrated cluster, e) Disc splat of Al-5CNT powder showing CNT rich cluster, and f) CNT rich cluster with poor metal infiltration.

deduced from the fact that the specific heat capacity of C (Graphite) is larger than that for Al-12%Si alloy at all temperatures. In particular, at the measured particle temperature of 2300K, specific heat capacity is equal to 2.145 J g− 1 K− 1 for graphite and 1.152 J g− 1 K− 1 for Al-12% Si alloy (taken from thermo-chemical software and database FactSage) respectively. The higher heat content would prolong the time required for heat loss to occur. Hence, the melt will be molten for a longer duration which would enable spreading and finger formation. It is concluded from the magnified images (Fig. 6c and d) that the CNT clusters in the fingers

have been infiltrated by the molten alloy. Distribution of CNTs in each of the fingers is very uniform. Fig. 6e shows a disc splat of SD Al-5CNT and a CNT rich cluster (Fig. 6f) which has been partially infiltrated with molten Al–Si alloy. The disc splat is larger (150 µm) compared with fingered splat (75 µm) for reasons explained earlier. Fig. 7 shows a disc splat from SD Al-10CNT particle showing CNTclusters which are partially infiltrated with metal. The size of the CNT clusters observed within a splat is between 10 and 30 µm which is same as the size of CNTclusters observed in the coating cross sections of Fig. 3). A few clusters of size up to 50 µm

Author's personal copy 1550

S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

Fig. 8. XRD plot for the three coatings showing Al4C3 formation in CNT containing coatings.

viscosity and surface tension would be very low which promotes infiltration. Hence, infiltration of molten metal between CNTs is expected to be dominated during in-flight as compared to impact and

Fig. 7. a) Disc splat from SD Al-10CNT powder showing CNT rich cluster, and b) high magnification image of the CNT rich cluster with poor metal infiltration.

are observed in the Al-10CNT coating cross section, which form due to contiguity of the CNT clusters from individual splats. Landry et al. [38] have shown that the wetting of Al–Si alloys on graphite improves with time i.e., the contact angle diminishes from 160 to 40 degrees with time due to formation of interface carbide layer over a period of 104 s at a temperature of 1190 K. This indicates that reaction products forming at the interface promote wetting and hence infiltration of molten alloy. Table 1 shows the powder properties along with the in-flight temperature and velocity. Considering the stand-off distance as 100 mm, the flight time has been calculated and shown in Table 1. The particle flight time is less than 1 ms. The measured temperature of the particles during flight is of the order of 2300 K. It is anticipated that infiltration of CNT clusters would be limited due to the small interaction time between the CNT cluster and the molten metal. However, under such high superheated conditions, the

Table 1 Properties and in-flight temperature and velocity of the particles Powder

Specific surface area m2/g

Temperature K

Velocity m/s

Particle size µm

Approximate flight time ms

Al–Si SD Al-5CNT SD Al-10CNT

1.44 4.50 5.90

2309 ± 34 2287 ± 3 2308 ± 6

109 ± 3 168 ± 1 180 ± 2

14 ± 9 57 ± 21 39 ± 15

0.9 0.6 0.6

Fig. 9. SEM images showing Al4C3 formation in Al-10CNT coating on a) cross section, and b) fracture surface.

Author's personal copy S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

solidification. Since in Al-10CNT a larger amount of CNT clusters are observed (due to higher CNT content), it is expected that there would be more clusters which have not been infiltrated with metal (Fig. 7b). 3.2.3. CNT stability and interfacial phenomena Structural stability of CNTs is essential for achieving strengthening. Fig. 8 shows the XRD spectra for the Al-Si, Al-5CNT and Al-10CNT coatings. Al4C3 is formed in both CNT reinforced coatings as per the following reaction [39]. 4Alðliq:Þ + 3C = Al4 C3 ; ΔGf = −258:32 + 0:0968T

ð3Þ

The free energy of formation of Al4C3 is −35.7 kJ mol− 1 at 2300 K that proves thermodynamic feasibility at the temperature experienced by inflight particles. The amount of Al4C3 formed was calculated by the ratio of the area under Al4C3 peaks to the area under all the peaks. Al4C3 was found to be 7.8% and 15.8% for Al-5CNT and SD Al-10CNT coatings, respectively. Yang and Scott [40] have shown formation of Al4C3 in a carbon fiber reinforced Al-7 wt.%Si alloy produced by liquid metal infiltration of fiber perform at 700 °C. De Sanctis et al. [41] have shown formation of lath like Al4C3 in carbon fiber reinforced Al-13 wt.%Si alloy

1551

produced by liquid metal infiltration at 690 °C. Landry et al. [38] found during sessile drop experiment that that Al4C3 is formed by the reaction between Al-13%Si alloy and graphite, while SiC is formed during the reaction between Al-20%Si alloy and graphite indicating that Si content has strong effect on the carbide formation. From our research group, Laha et al. [42] have reported earlier that β-SiC is formed during plasma spraying of Al-23 wt.%Si alloy powders blended with 10 wt.% CNT which results in improved wetting. Ci et al. [43] have shown the formation of needle like Al4C3 during annealing of CNTs over which Al had been deposited by magnetic sputtering. Plasma spraying is a rapid solidification process and the reaction time is very small (~ms). From the SEM of fracture surface or the coating cross section, no typical features corresponding to Al4C3 formation were found for Al-5CNT coating. But for Al-10CNT coating, Al4C3 could be exclusively seen. Fig. 9a and b shows the SEM images of Al-10CNT coating's cross section and fracture surface, respectively. Al4C3 needles are formed near CNT clusters where the activity of carbon is high. It is observed in Fig. 9a that the surface of the cluster seems to be made of carbonaceous mass resulting from damaged CNT due to reaction. In case of SD Al-10CNT, the large CNT content leads to mesh formation and there is large surface area available for reaction. The CNT rich clusters which are poorly infiltrated with metal undergo

Fig. 10. TEM images showing a) fine grained microstructure of Al-10CNT coating, inset shows the diffraction pattern showing Al4C3, Al and Si b) metal coated and reacted CNT in Al10CNT coating, inset shows the diffraction pattern indicating Al4C3 formation, c) undamaged CNT in Al-5CNT coating, and d) undamaged CNT in Al-10CNT coating with some reaction products on the surface.

Author's personal copy 1552

S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

Fig. 11. Schematic showing heat transfer mechanisms in the clusters.

Fig. 12. a) Images of the indent on the coatings obtained with 2000 µN load, b) representative load displacement curve from nanoindentation, c) variation of elastic modulus and hardness of the coatings with CNT vol.%, d) calculated yield strength ratio of the coating and the ratio of elastic work, and e) line profiles along a median of the indentation obtained with 2000 µN load showing pile-up surface.

Author's personal copy S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

1553

Table 2 Nano-mechanical properties of the coatings Coating

CNT vol.%

Microhardness GPa

Nanohardness GPa

Elastic modulus GPa

We/Wt

σy GPa

σ0.29 GPa

Al–Si Al-5CNT Al-10CNT

0 6.2 12.4

0.870 ± 0.03 1.350 ± 0.05 2.100 ± 0.04

1.61 ± 0.20 2.33 ± 0.27 2.89 ± 0.27

90 ± 9.5 107 ± 6 125 ± 7

0.17 ± 0.02 0.23 ± 0.03 0.33 ± 0.02

32.6 ± 2.0 38.3 ± 1.9 41.5 ± 1.8

38.3 ± 2.7 44.1 ± 1.3 59.1 ± 3.0

more damage due to direct interaction with the plasma plume. This will lead to generation of more defects on the CNTs which will lead to increase in reaction sites for carbide formation. These are the reasons for increased amount of Al4C3 formed in Al-10CNT coating. Fig. 10a shows TEM image of fine grained structure in Al-10CNT coating. Inset shows the SAD pattern confirming the presence of Al4C3. Fig. 10b shows a metal coated CNT in Al-10CNT coating which has undergone some reaction with the matrix. The inset shows the diffraction pattern which indicates Al4C3 formation. Fig. 10c and d shows HR-TEM images of undamaged CNTs in Al-5CNT coating and Al-10CNT coating, respectively. CNT surface is smooth in case of Al5CNT coating (Fig. 10c) whereas discontinuous reaction product is formed on the CNT surface in Al-10CNT coating (Fig. 10d). Surface walls of CNT shows slightly damaged structure in case of Al-10CNT as compared to Al5CNT coating. Table 1 indicates that the SD Al-10CNT powder experienced higher velocity and slightly higher temperatures due to the smaller size of the powders and CNT mesh formation which increases the thermal conductivity and emissivity of the powder. The high thermal conductivity of CNTs [8] helps in conducting heat to the interior of the cluster thereby assisting the damage. It was shown earlier that metal infiltration into the CNTcluster was more prominent in Al-5CNTcoating. The molten layer acts as a protective layer shielding the CNT clusters and helps prevent damage to CNTs in Al-5CNT coating. Fig. 11 shows a schematic of the heat transfer in metal coated and uncoated clusters. The heat transfer in coated clusters would be only by conduction and convection and the metal layer will shield the CNTs from the radiation of the plasma plume. The presence of larger number of uncoated CNT clusters in Al-10CNT would result in more damage to CNTs than in case of the Al-5CNTcoating. So it can be concluded that wetting of CNTs by molten alloy is critical in establishing dispersion as well as stability of CNTs.

39% by addition of 5 wt.% and 10 wt.% of CNTs, respectively. Further analysis of the load displacement curves was carried out using the method used by Chen et al [45]. The strength of the coatings has been calculated using the following relations: We hr = 1− Wt hmax

ð4Þ

     σy Pmax E = M σ 1 + ln + M 1 0:29 2 σy σ 0:29 h2max

ð5Þ

 2 σ 0:29 −σ y hr hr = 1−0:142 −0:957 0:29E hmax hmax

ð6Þ

where We and Wt refer to the elastic work and total work done during indentation, hmax and hr are the maximum depth and residual depth, Pmax is the maximum load, σy is the yield strength, σ0.29 is the stress at strain of 0.29, E is the elastic modulus and M1 = 6.618 and M2 = −0.875 are constants for Berkovich type indenter. The We/Wt ratio provides a

3.3. Mechanical property characterization Fig. 12a shows the scanning probe microscopy (SPM) images of the indents from Al-Si, Al-5CNT and Al-10CNT coatings, respectively. The decreasing size of the indents suggests strengthening effect due to addition of CNTs. Fig. 12b shows the representative load displacement curves for the three coatings at a load of 2000 µN. The decrease in maximum displacement (hmax) and the residual depth (hr) indicates strengthening due to addition of CNTs. Fig. 12c plots the hardness and reduced elastic modulus of the coatings as a function of CNT content. There is a consistent increase in the elastic modulus with CNT addition. The scatter in the data is due to the localized nature of the nanoindentation test in which the values are influenced by the local CNT/Si content. Table 2 shows the elastic modulus and hardness values (nano and micro) for the coatings. It is seen that the standard deviation of the hardness values obtained by micro-indentation were less than 4% of the average value indicating that the values are consistent. Also the standard deviation of the elastic modulus values of the CNT composites obtained by nanoindentation were less than 6% of the average value indicating consistency in the measured values. The nanohardness values were found to increase by 44% and 80% by addition of 5 wt.% and 10 wt.% CNT, respectively. The values of hardness obtained by nanoindentation are larger compared to that obtained by microindentation. This is attributed to the localized nature of the nanoindentation test. Values of mechanical properties obtained from microindentation involve several splats and are influenced by several features like inter-splat porosity and splat sliding [44]. Elastic modulus shows an improvement of 19% and

Fig. 13. SEM images showing a) CNT pullout in Al-5CNT coating, b) crack bridging in Al10CNT coating.

Author's personal copy 1554

S.R. Bakshi et al. / Surface & Coatings Technology 203 (2009) 1544–1554

measure of the elastic energy recoverable from the material. It can be seen from Table 2 that the values of σy calculated from nanoindentation are quite large. Strengthening effect is reflected by the increase in the values of σy and σ0.29 on addition of CNTs. The ratio of the strength of the composite coatings to that of Al–Si coating and the We/Wt ratio has been plotted in Fig. 12d. There is an increase of 17.5% and 27% in σy due to addition of 5 wt.% and 10 wt.% CNTs. The increase in We/Wt ratio with increase in CNT content indicates improvement of elastic properties of the composites. Fig. 12e shows the depth profiles taken along the median of the indent. Pile-up of material occurs on the surface of the indenter and is observed on the edges of the triangle shaped indent. The pile-up height for Al–Si, Al-5CNT and Al-10CNT coating are approximately 19 nm, 15 nm and 8 nm, respectively. The reduced depth of the indent and pileup also indicates strengthening effect of CNT addition on the composite. Fig. 13 shows the SEM images of the fracture surfaces indicating the strengthening mechanisms that could exist in CNT reinforced aluminum composites. Fig. 13a shows CNT pull-out phenomena in Al-5CNT coating. CNTs are pulled out from the matrix until they fail under tension. The aligned nature of the CNTs in Fig. 13a indicates that this mechanism could result in good strengthening effect. Fig. 13b shows a crack in Al-10CNT coating due to solidification shrinkage. It is observed that the CNTs in the matrix form bridges between the crack which could lead to toughening. These mechanisms would results in increase of strength and stiffness as measured by tensile test. The test would require fabrication of thick tensile samples, which will be our future work. 4. Conclusions Spray drying is an effective way of dispersing CNTs at the scale of a micrometer. Thick (up to 5 mm) aluminum alloy coatings containing 5 wt.% and 10 wt.% CNT were successfully fabricated by plasma spraying. Coating microstructure is two phase and is made up of matrix with distributed CNTs and CNT rich clusters. Single splat of spray dried particles shows CNT agglomeration and cluster formation due to poor wetting with the alloy. Extent of damage of CNTs is found to be more in the 10 wt.% coating compared to 5 wt.% CNT coating. Nanoindentation showed an increase in the elastic modulus of 19% and 39% by addition of 5 wt.% and 10 wt.% CNT to aluminum alloy. Pile up heights reduced and ratio of elastic work to total work increased by addition of CNT indicating improvement in the elastic recovery properties. Fiber pullout and crack bridging are the strengthening mechanisms for improvement in mechanical properties. Acknowledgement Authors acknowledge financial support from the National Science Foundation CAREER Award (NSF-DMI-0547178) and DURIP Grant from Office of Naval Research (N00014-06-0675). Srinivasa Rao Bakshi

would like to acknowledge Dissertation Year Fellowship from Florida International University for financial support. Authors would like to thank CeSMEC at FIU for micro-Raman spectroscopy, Dr. Tapas Laha for plasma spraying and Tanisha Richard for helping with the schematics. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45]

A. Sato, R. Mehrabian, Metall. Trans., B 7 (1976) 443. P. Rohatgi, JOM 43 (4) (1991) 10. A. Kelly, J. Mater. Sci. 41 (2006) 905. M.A. Taha, Mater. Des. 22 (2001) 431; S. Rawal, JOM 53 (4) (2001) 14. J.S. Shelly, R. LeClaire, J. Nichols, JOM 53 (4) (2001) 18. R. Schaller, J. Alloys Compd. 355 (2003) 131. M.F. Yu, O. Laurie, M.J. Dyer, K. Moloni, T.F. Kelly, R.S. Ruoff, Science 287 (2000) 637–640. P. Kim, L. Shi, A. Majumdar, P.L. McEuen, Phys. Rev. Lett. 87 (21) (2001) 215502. K.T. Lau, D. Hui, Composites B33 (2002) 263. R. Andrews, M.C. Weisenberger, Curr. Opin. Solid State Mater. Sci. 8 (2004) 31. J.N. Coleman, U. Khan, W.J. Blau, Y.K. Gun'ko, Carbon 44 (2006) 1624. G.D. Zhan, J.D. Kuntz, J. Wan, A.K. Mukherjee, Nat. Matters 2 (2003) 38. K. Balani, S.R. Bakshi, Y. Chen, T. Laha, A. Agarwal, J. Nanosci. Nanotechnol. 7 (2007) 1. K. Balani, T. Zhang, A. Karakoti, W.Z. Li, S. Seal, A. Agarwal, Acta Mater. 56 (2008) 571. W.A. Curtin, B.A. Sheldon, Mater. Today (Nov 2004) 44. T. Noguchi, A. Magaria, S. Fukazawa, S. Shimizu, J. Beppu, M. Seki, Mater. Trans. 45 (2) (2004) 602. S.M. Zhou, X.B. Zhang, Z.P. Ding, C.Y. Min, G.L. Xu, W.M. Zhu, Composites: Part A 38 (2007) 301. C.F. Deng, D.Z. Wang, X.X. Zhang, A.B. Li, Mater. Sci. Eng., A 444 (2007) 138. J.H. Bak, D.K. Young, S.S. Hong, B.Y. Lee, S.R. Lee, J.H. Jang, M. Kim, K. Char, S. Hong, Y.D. Park, Nat. Maters 7 (2008) 459. K.T. Kim, S.I. Cha, S.H. Hong, S.H. Hong, Mater. Sci. Eng., A 430 (2006) 27. S.I. Cha, K.T. Kim, S.N. Arshad, C.B. Mo, S.H. Hong, Adv. Mater. 17 (2005) 1377. W.X. Chen, J.P. Tu, L.Y. Wang, H.Y. Gan, Z.D. Xu, X.B. Zhang, Carbon 41 (2003) 215. P. Quang, Y.G. Jeong, S.H. Hong, H.S. Kim, Key Eng. Mater. 326–328 (2006) 325. P.-Q. Dai, W.C. Xu, Q.-Y. Huang, Mater. Sci. Eng., A 483–484 (2008) 172. G.-R. Shen, Y.-T. Cheng, L.-N. Tsai, IEEE Trans. Nanotechnol. 4 (5) (2005) 539. C.-S. Chen, X.H. Chen, Z. Yang, W.-H. Li, L.-S. Xu, B. Yi, Diamond Relat. Mater. 15 (2006) 151. E. Carreno-Morelli, J. Yang, E. Couteau, K. Hernadi, J.W. Seo, C. Bonjour, L. Forro, R. Schaller, Phys. Status Solidi 201 (2004) 53. C.S. Goh, J. Wei, L.C. Lee, M. Gupta, Nanotechnology 17 (2006) 7. C.S. Goh, J. Wei, L.C. Lee, M. Gupta, Comput. Sci. Tech. 68 (2008) 1432. Z. Bian, R.J. Wang, W.J. Wang, T. Zhang, A. Inoue, Adv. Funct. Mater. 14 (2004) 55. Y.B. Li, Q. Ya, B.Q. Wei, J. Liang, D.H. Wu, J. Mater. Sci. Lett. 17 (1998) 606. Y. Tang, H. Cong, R. Zhong, H.M. Cheng, Carbon 42 (2004) 3260. T. Laha, A. Agarwal, T. McKechnie, S. Seal, Mater. Sci. Eng., A 381 (2004) 249. S.L. Lukaseiwicz, J. Am. Ceram. Soc. 72 (4) (1989) 617. W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564. A. Esawi, K. Morsi, Composites: Part B 38 (2007) 646. V.V. Sobolev, J.M. Guillemany, Mater. Lett. 42 (2000) 46. K. Landry, S. Kalogeropoulou, N. Eustathopoulos, Mater. Sci. Eng., A 254 (1998) 99. T.Y. Kosolapova (Ed.), Handbook of high temperature compounds: properties, production, applications, Hemisphere, New York, 1990, p. 228. M. Yang, V.D. Scott, Carbon 29 (7) (1991) 877. M. De Sanctis, S. Pelletier, Y. Bienvenu, M. Guigon, Carbon 32 (1994) 925. T. Laha, S. Kuchibhatla, S. Seal, W. Li, A. Agarwal, Acta Mater. 55 (2007) 1059. L. Ci, Z. Ryu, N.Y. Jin-Phillipp, M. Ruhle, Acta Mater. 54 (2006) 5367. F. Tang, J.M. Schoenung, Scr. Mater. 54 (2006) 1587. Y. Chen, T. Laha, K. Balani, A. Agarwal, Scr Mater. 58 (2008) 1121.