Metallurgical and Materials Transactions B

, Volume 41, Issue 4, pp 841–856

Synthesis of FeCu Nanopowder by Levitational Gas Condensation Process

Authors

    • ARC-International
  • A. M. Sriramamurthy
    • ARC-International
  • M. Vijayakumar
    • Defence Metallurgical Research Laboratory
  • G. Sundararajan
    • ARC-International
  • Kamanio Chattopadhyay
    • Department of Materials EngineeringIndian Institute of Science
Article

DOI: 10.1007/s11663-010-9370-8

Cite this article as:
Sivaprahasam, D., Sriramamurthy, A.M., Vijayakumar, M. et al. Metall and Materi Trans B (2010) 41: 841. doi:10.1007/s11663-010-9370-8

Abstract

Condensation from the vapor state is an important technique for the preparation of nanopowders. Levitational gas condensation is one such technique that has a unique ability of attaining steady state. Here, we present the results of applying this technique to an iron-copper alloy (96Fe-4Cu). A qualitative model of the process is proposed to understand the process and the characteristics of resultant powder. A phase diagram of the alloy system in the liquid–vapor region was calculated to help understand the course of condensation, especially partitioning and coring during processing. The phase diagram could not explain coring in view of the simultaneous occurrence of solidification and the fast homogenization through diffusion in the nanoparticles; however, it could predict the very low levels of copper observed in the levitated drop. The enrichment of copper observed near the surface of the powder was considered to be a manifestation of the lower surface energy of copper compared with that of iron. Heat transfer calculations indicated that most condensed particles can undergo solidification even when they are still in the proximity of the levitated drop. It helped us to predict the temperature and the cooling rate of the powder particles as they move away from the levitated drop. The particles formed by the process seem to be single domain, single crystals that are magnetic in nature. They, thus, can agglomerate by forming a chain-like structure, which manifests as a three-dimensional network enclosing a large unoccupied space, as noticed in scanning electron microscopy and transmission electron microscopy studies. This also explains the observed low packing density of the nanopowders.

Introduction

During the last two decades, more attention has been paid to nanomaterials because of their unusual properties[15] when compared with conventional polycrystalline materials. In the area of powder metallurgy, nanostructured,[6,7] nanograined,[8] and nanosize particulate[911] materials are being paid great attention. Both chemical and physical techniques exist for the preparation of nanopowders. In the preparation by physical methods, vapor condensation is an important technique that can produce strain-free particles of very high purity. These powders are characterized by very clean reactive surfaces. Consequently, handling such powders requires great care, and in general, nanopowders are consolidated in situ into bulk samples.[12,13] In most of these techniques based on vapor condensation, a fixed quantity of materials is melted and evaporated in an inert gas atmosphere and condensed on walls cooled by circulating very cold fluids like liquid nitrogen. Such techniques are well suited for producing elemental powders. However, for alloy powders, the composition of the melt will be changing continuously during the process as a result of the preferential evaporation of relatively more volatile components. There is no continuous addition of alloy to the evaporating liquid pool that could result in a steady-state process. In levitated drop gas condensation (LGC), material of a chosen composition is fed continuously into the levitated liquid pool. This leads to a steady-state process, producing powder of the same composition as the alloy being fed. Furthermore, because the molten alloy in this process is not in contact with any other material, the powder produced is completely free from any contamination, and the composition is expected to be uniform throughout the process period, unlike other physical vapor processes. Unfortunately, little literature[1417] has been published on this technique, and whatever work was done covers only elemental and oxide nanopowders. This article covers the application of this technique to prepare Fe-Cu nanopowder and its characterization. A model has been proposed to aid in understanding the results qualitatively. We have chosen Cu as an alloying element because the literature suggests that it catalyzes the reduction of iron oxide that forms during passivation—a necessary step before the removal of the powder from the apparatus. We also have studied elemental iron to serve as reference.

Experimental

LGC

Figure 1 shows the schematic diagram of the LGC apparatus used in this investigation. The apparatus consists of three important units, viz., (1) a radio frequency (RF) generator with a high-frequency (440 kHz) countercurrent inductor for melting and levitating metals/alloys inside a 15-mm internal diameter quartz tube (QT), (2) an inert gas-purging unit to supply inert gas (Ar or He) to sweep the vapor over the levitating molten drop, and (3) a wire-feeding unit (WFU), which can feed continuously the materials in the form of a 0.5-to-1-mm diameter wire into a levitating drop throughout the entire duration of the operation to compensate for the loss of material by evaporation. A molten drop of around 7 mm generally is levitated after evacuating the full apparatus to less than 2-kPa pressure and then starting the flow of inert gas at the operating pressure of 13 to 40 kPa. The inert gas flow can be carried out in two different modes, open cycle in which the flowing inert gas is let off continuously and a closed cycle in which the inert gas is cooled and recirculated. The WFU has two independently operated drives that can feed wires at any required ratio as well as speed. Alloy or intermetallic nanopowders can be produced either by using two elemental wires or by using a single alloy wire. The nanopowder-argon gas aerosol comes out from the bottom of the QT and cools in a water-cooled chamber. Subsequently, the nanopowders formed are separated out by a filter unit equipped with a tapping system for periodic cleaning and collected in powder-collecting container. The as-synthesized nanopowders with ultra-clean surfaces are highly reactive and can be passivated at the end of the operation by evacuating the apparatus to less than 2 kPa and backfilling with any chosen gas. Alternatively, they also can be coated or chemically treated in situ just at the exit end of the QT and collected for fundamental studies by attaching a transmission electron microscopy (TEM) sampling module to the apparatus. More details about this experimental setup and the general physical characteristics of various nanopowders synthesized in this technique are given elsewhere.[14]
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig1_HTML.gif
Fig. 1

Schematic diagram of the levitational gas condensation apparatus: CW = coil with wire, WUD = wire unwinding device, UCU = unwinding device control unit, MD = melted drop, CCI = counter current inductor, SH = sampling head, SR = sampling rod, SB = separable boat, SA = separable ampoule, RT = reactor tube, F = filter, R = cooled receiver, RT1, RT2 = rotometers, and M1, M1 = manometers

FeCu Nanopowder Synthesis

Fe wire of 99.99 pct purity (Alfa Aesar, Ward Hill, MA) and Cu wire of 99.9999 purity (oxygen-free high-conductivity copper) in the weight ratio 96Fe:4Cu (Fe0.965Cu0.035) was bundled into a spherical ball of 1 gm and heated, melted, and levitated at a temperature of 2275 K (2002 °C). The vapor that came out of the molten drop was swept by a stream of high-purity argon gas (total impurity level less than 50 ppm) that entered the QT of the apparatus from the top and left from the bottom. The condensation process was carried out in an open-cycle mode, with an argon flowing rate of 0.01 m3 per minute. To compensate for the material leaving the molten drop, the drop was fed continuously with elemental wires of Fe and Cu in the same weight ratio (96Fe:4Cu) used initially. Also, to sustain the size of the molten drop, the rate of feeding was maintained equal to the rate of powder produced (g/hr). It is noted that just below the levitated drop we saw a blank zone of about 6 to 8 mm, after which, we saw a rain of agglomerates. It was also noted that the raining agglomerates were confined to about 7-mm diameter around the axis, as shown in Figure 2. The nanopowder was separated from the aerosol by a cloth filter and subsequently was passivated by passing air, oxygen, or nitrogen at 0.15 cm3/sec before exposing it to atmosphere. The composition of the drop after reaching steady state was determined by allowing it to fall into an aluminum cup by switching off the power to the RF generator and subjecting it to chemical analysis by inductively-coupled plasma-atomic emission spectroscopy (ICP-AES) for Cu. The temperature of the molten drop and the nanopowder during passivation was measured using a two-color pyrometer and an iron-constantine (J-type) thermocouple, respectively. Various important experimental parameters measured during the LGC process are given in Table I. Experiments also were done on pure iron by a similar procedure for the purpose of comparison.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig2_HTML.jpg
Fig. 2

Photograph showing the levitated FeCu drop and the raining agglomerates below the drop

Table I

Experimental Parameter Measured During the Synthesizing of FeCu Nanopowder in LGC

Parameters

 

Rate of wire feeding (g/hr)

1.1–1.2

Pressure (KPa)

30

Argon flow rate (m3/hr)

0.60

Inductor coil power (KW)

4.5

Temperature of the molten drop (K (°C))

2275 (2002)

Argon gas velocity in QT (m/s)

93.75

Characterization

The particle size of the FeCu nanopowder was characterized by (1) scanning electron microscopy (SEM) equipped with a high-resolution field emission gun (FEG-SEM-Hitachi S4300 SE/N, Hitachi High Technology America, Pleasanton, CA), (2) TEM and (3) the Brunauer, Emmett, and Teller (BET) method in a Micromeritics ASAP 2020 system (Micromeritics Instrument Corporation, Norcross, GA) using a 373 K (100 °C) degassing temperature. The particle size distribution was measured (on a sample of about 800 particles) on 100 KX SEM micrographs using an image analyzer (Image Pro Plus; Media Cybernetics, Silver Spring, MD) software. The crystallite sizes in the nanopowder, phases such as the passivating oxides over the nanopowder and the amount of various phases were characterized by X-ray diffraction (XRD) in a Bruker’s diffractometer (AXS Model Number D8 Advance System, Bruker AXS GmbH, Karlsruhe, Germany) using Cu Kα radiation (40 kV and 40 mA). The XRD scan was carried out over the 2θ from 20 deg to 120 deg at 0.005 deg/s−1 scan rate. The crystallite size of the nanopowder was determined by the Williamson–Hall method,[18] and the amount of oxide was quantified based on the peak height method. The structure, morphology, passivating oxide layer thickness, and chemical composition of individual nanopowder were investigated by TEM. The samples for TEM were prepared by mixing FeCu nanopowder with gatan two component epoxy glue and casting the mixture on to a thin sheet over aluminum foil followed by curing at 353 K (80 °C) on a hot plate. The glue sheet then was thinned down to 100 μm by mechanical polishing followed by cutting a disk of 3-mm diameter. The disk again was polished down to 40 μm mechanically, which was polished further into a dimple to about 10 μm. This was thinned further by ion beam milling with an argon ion source until perforation. TEM investigations were carried out at an excitation voltage of 200 kV using an FEI TECNAI G2 (FEI Worldwide, North America Nanoport, Hillsboro, OR) transmission electron microscope equipped with energy-dispersive X-ray spectrometry (EDS) for compositional analysis. Electron diffraction of a few individual FeCu nanoparticles was carried out to investigate the crystal structure. Composition analysis of several individual nanoparticles was carried out at random locations. The nanopowder also was examined in TEM by dispersing the as-prepared powder in ethanol and spreading one or two drops on a copper TEM sample grid to study the structure of agglomerates. X-ray photoelectron spectroscopy (XPS) (Omicron system, Omicron Nanotechnology GmbH, Taunusstein, Germany) equipped with an Ar+ sputtering gun was used to investigate the composition of the nanopowder surface and the depth profiling of the Cu concentration. For this purpose, the nanopowder was compacted into a thin disc (10-mm diameter and 0.5-mm thickness) with a smooth surface and then analyzed using monochromatized Al Kα source. The depth profiling was carried out by alternating ion (Ar+) sputtering and XPS analysis, and the Ar+ beam was rastered over an area of 3 × 3 mm2 and was run at a voltage of 3 KV. The surface composition was determined from the characteristic Fe and Cu peaks after different durations of sputtering.

Results

Physical Characteristics and Structure FeCu Nanopowders

The various physical characteristics of the FeCu and Fe nanopowders produced under identical experimental conditions are summarized in Table II. Figures 3(a) and (b) show the typical SEM image along with the particle size distribution of the FeCu nanopowder. The nanopowders are spherical but highly aggregated, with a log-normal particle size distribution typical of any vapor phase synthesized nanopowders.[19] The particle size measured from BET and TEM are comparable, whereas the one measured with the FEG-SEM image analyzer was higher. However, the crystallite sizes measured by X-ray diffraction are comparable with particle sizes measured by BET and TEM, indicating that the nanopowders are predominantly single crystals. Figure 3(c) shows very small agglomerated chains of particles that branch occasionally and develop into a three-dimensional network as in a reticulated sponge. One such spongy agglomerate is shown in Figure 3(d). The packing density of the as-synthesized nanopowders was around 0.1 to 0.35 pct of the theoretical density. This is consistent with three-dimensional networked structure encompassing a large unoccupied space made of a chain of nanoparticles.
Table II

Physical Characteristic of the FeCu and Fe Nanopowders

Physical Characteristic

FeCu Nanopowder

Fe Nanopowder

BET surface area, m2/g

22.98

19.91

Mean particle size, nm

 BET

33.2

38.3

 FEG-SEM

50

59

 TEM

31

38

Crystallite size, nm

34.6

35

Packing density, g/cm3

0.0262

0.0072

https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig3_HTML.gif
Fig. 3

(a) An SEM image of FeCu nanopowder, (b) the particle size distribution, and (c) and (d) three-dimensional reticulated sponge structures made of nanoparticle chains

Figures 4(a) and (b) show the bright field TEM micrograph of the FeCu nanopowder along with the diffraction pattern taken over the cluster of particles. As is shown in the micrograph, the nanoparticles are spherical, having core-shell type structure with a shell of about 3- to 4-nm passivating oxide layer on the surface of each particle. Most particles in the chain are sintered together to the extend-of-neck-to-diameter (X/D) ratio of 0.6. Indexing of the electron diffraction pattern showed rings made of spots corresponding to α-Fe {dhkl: 2.08 (110), 1.19 (211), and 1.45 (200)} along with a broad diffused rings (dhkl: 2.58, 1.57, and 2.99) that corresponds to the surface layer of oxides of Fe. The dark field TEM micrograph produced using the diffuse ring corresponding to dhkl–2.58 is shown in Figure 4(c). This micrograph shows that the diffraction ring used is from the passive layer on the surface of the particle, and it is made of ultrafine oxide grains. Figure 5 shows the X-ray diffraction pattern of FeCu nanopowder. Along with body centred cubic (BCC) Fe, a broad single peak at 2θ = 35.52 deg was present that could be from the surface passive oxide layer. Nano Fe powders synthesized by physical vapor condensation and subsequently passivated by slow exposure to air normally have 2 to 4 nm thick surface oxides layer, consisting mainly of γ-Fe3O4 or Fe2O3.[20] Both oxides have a spinel structure with very close d-spacing. In addition, the very low grain size of the oxides leads to peak broadening. Together, these factors precluded the quantification of the proportion of the two oxides. Further studies on the surface chemistry of the nanopowder by XPS indicated the presence of copper oxides in addition to iron oxides on the surface. Figures 6(a) and (b) show the XPS spectra of the powders for Cu 2p and Fe 2p levels, with and without sputter cleaning to remove the surface oxide layers. Table III gives the binding energy peak positions for Fe (2p3/2 and 2p1/2) and Cu (2p3/2 and 2p1/2) of the FeCu nanopowder along with those of the standard samples, viz., Fe, Fe3O4, Fe2O3, Cu2O, and CuO, which were determined using the same instrument under the same conditions. It is evident from the data that Fe 2p3/2 and Cu 2p3/2 binding energy peaks of FeCu nanopowder are close to those of Fe2O3 (710.9 ev) and CuO (934.8 ev), respectively, suggesting that the surface is predominantly covered by a solid solution of CuO and Fe2O3.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig4_HTML.jpg
Fig. 4

(a) TEM micrographs of FeCu nanopowders, (b) a corresponding SAED pattern (* corresponds to oxide and ** corresponds to Fe), and (c) the dark field image of the same area shown in (a)

https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig5_HTML.gif
Fig. 5

X-ray diffraction patterns of FeCu nanopowder

https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig6_HTML.gif
Fig. 6

XPS spectra showing the biding energies corresponding to (a) Fe 2p and (b) Cu 2p of FeCu nanopowder

Table III

Fe 2p and Cu 2p Binding Energy Peaks of Various Standard Samples Along with FeCu Nanopowder

Materials

Fe

Cu

2p1/2

2p3/2

2p1/2

2p3/2

96Fe4Cu nanopowder

724.6

710.8

953.2

933.2

α-Fe foil*

719.7

706.7

Fe3O4**

724.1

710.3

Fe2O3**

724.9

710.9

Cu2O**

952.2

932.3

CuO**

954.6

934.8

*99.99 pct pure foil after 4 kV, 120 min sputter cleaning using Ar+ gun

**Nanopowders from Alfa Aesar

Calculation of the composition of various constituents present, based on the XRD peak heights after background subtraction, showed that the overall oxide content on FeCu nanopowders is 7.1 pct. The lattice parameter “a” measured from Fe d[220] and d[310] planes of FeCu nanopowders was 2.868A deg and 2.869A deg, respectively. These values are higher than the ones for Fe nanopowders (d[220] = 2.864A deg and d[310] = 2.866A deg) synthesized from same raw material under identical conditions. Because the atomic radius of Cu (1.278A deg) is higher than that of Fe (1.241A deg), it is expected that the lattice parameter of Fe will increase with Cu dissolution. This is evident in a study by Kneller[21] on an Fe-Cu solid solution prepared by the simultaneous vapor deposition of Fe and Cu on a cold substrate; this study reports a significant increase in the aFe with a Cu concentration particularly below 15 wt pct. In this present study, the aFeCu of the FeCu nanopowder is comparable with the theoretically calculated aFeCu based on the atomic radii of Fe and Cu.

Chemical Properties

The TEM micrograph and the corresponding EDS pattern taken on a single FeCu nanopowder using a nanoelectron beam are shown in Figure 7. The composition of several such individual nanoparticles revealed that the concentration of copper varies from particle to particle, and in some particles, the copper content was as low as 1.6 wt pct. However, the composition of the individual agglomerates quantified using SEM-EDS showed that the copper concentration in different areas of the agglomerates is around 4.0 ± 0.2 wt pct, which is comparable with the overall composition of the FeCu nanopowders measured by ICP-AES (4.05 wt pct). These results indicate that although the average composition of the powder produced is around 4.0 wt pct, there is a difference in the Cu concentration from particle to particle.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig7_HTML.jpg
Fig. 7

The TEM micrograph of FeCu nanopowders and the corresponding EDS pattern taken on the single FeCu nanopowder using a nanoelectron beam

Figure 8 shows a variation of Cu concentration in the FeCu nanopowder as a function of Ar+ sputtering time from the surface to inside, quantified based on the XPS peak area. The concentration of Cu at the surface is as high as 21.06 wt pct, and with surface etching, the concentration gradually decreases. After 180 minutes of etching, the composition reaches 4.4 wt pct, a value very close to the average composition. The depth to which etching was done is uncertain. We believe it could be a few nanometers based on the previous published report[22] on atomized steel powder. Because of the presence of a significant amount of porosity in the compacted sample used for XPS, the contribution from unetched surfaces of the nanoparticles to the overall Cu concentration always will be present. Hence, even in the 180-minute etched sample that showed 4.4-wt pct Cu on the surface, the actual Cu percentage in the core of the nanoparticles could be less.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig8_HTML.gif
Fig. 8

Variation of the Cu concentration from the surface to inside the FeCu nanopowders, quantified by XPS depth profiling

Discussion

A sketch of the important portion of the apparatus incorporating a conceptual model of the process is shown in Figure 9. During the experiments, the levitated drop was held at 2275 K (2002 °C). As stated earlier, this is the only method in which the process is capable of attaining a steady-state condition. Table IV gives the composition of the molten drop levitated for different durations. It seems that the process attains steady state in less than 10 minutes from the time of melting of the 96Fe:4Cu. The steady-state composition of the levitated drop is 0.2 wt pct, which is significantly less than the initial composition or the composition of the nanopowder produced (4.05 wt pct Cu). At steady state, the material evaporated producing the nanopowder is compensated fully by continuously feeding the wires of Fe and Cu at the appropriate rates so that the composition of the levitated drop remains same. The vapor produced by the levitated drop is swept by the incoming argon gas. The argon gas, when it first mixes with the vapor, causes significant cooling, which results in burst homogeneous nucleation. This gas continues to flow over the drop, collecting the vapor being produced by the levitated drop that contributes to the growth of the nanoparticles. With a low Reynolds’s number (1600), the argon flows in a streamlined state. This is clear from the fact that the alloy-bearing gas is confined to a diameter of about 7 mm around the axis in the exit side of the tube, as evidenced by the visible rain of agglomerates of the powder confined to this region. Furthermore, as shown in Figure 9, a stagnant gas could be present above the levitated drop where the incoming argon impinges before it sweeps over the levitated drop and surrounds the eddy just below the levitated drop. Because the gas in the eddy circulates very close to the drop, they are loaded with vapor, possibly in condensed state, in addition to vapor. Because the gas in the eddies flow in closed loops, its contribution to the production of nanopowder is not clear. Even if they have some contribution, it will be negligible in comparison with the contribution of the vapor swept by the argon, flowing from the entry side to the exit side close to the levitated drop. The thickness of the vapor-bearing argon over the surface of the levitated drop was estimated to be approximately 0.9 mm based on the conservation of vapor-bearing volume of the argon but ignoring velocity distribution. The volume of the vapor-bearing argon was estimated from the diameter of the rain of particles in the exit side of QT and velocity of argon.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig9_HTML.gif
Fig. 9

Schematic showing the conceptual model of LGC process

Table IV

Composition of the Liquid FeCu Drop Levitated for Different Times

Time (s)

Cu (wt pct)

0*

4 ± 0.05

10

2.76 ± 0.05

30

2.52 ± 0.05

600**

0.2 ± 0.002

*Initial composition of the alloy before melting

**Steady-state composition of the levitating drop

The state of the nucleated nanoparticles within the vapor-bearing zone mainly depends on their temperature. An effort has been made to estimate the temperature of the condensed droplets/particles based on heat transfer considerations. The details of the calculation are described in Appendix A. The nanodroplets or particles of 18-nm radii (mode of particle size distribution) was taken for this calculation. For the purpose of comparison, a calculation also was done for smaller and larger particles. It turned out that the lowest estimated temperature of the nanodroplet or particle separated by 10 micrometers from the levitated drop under equilibrium conditions (i.e., dT/dt = 0) is around 1470 K (1197 °C), and at 0.9 mm—the end vapor-bearing zone—it is 1293 K (1020 °C). This suggests that most of the condensed particles in the vapor-bearing zone of 0.9 mm around the levitated drop are in solid state. The time required for the gas or particles to travel around the levitated drop and leave the latter is around 3 ms. During this period, the particles can grow further by the intake of vapor until they exit from the proximity of the levitated drop. The particles also can grow by coalescence with other particles that impinge on them at this temperature. Because the droplets and particles are extremely small, they will be in a state of Brownian motion and will collide and coalesce with each other while being carried by the flowing argon toward the exit side. The rate of cooling estimated through heat transfer calculations (Appendix A) is very high, of the order of 105 K/sec, when droplets are very close to levitated drop, and it falls as the distance from it increases.

To aid the understanding of the process of condensation, composition of phases, as well as possible coring, a liquid–vapor phase diagram was calculated based on the method adopted by Vijayakumar et al.[23] But the liquid–vapor phase diagram depends on the total vapor pressure. This was estimated to be 3040 Pa (0.03 atm.) from the rate of production of nanopowder, as shown in Appendix B. The calculation of the phase diagram is described in Appendix C. From this calculated phase diagram shown in Figure 10, the composition of the liquid that is in equilibrium with the vapor containing 4-wt pct Cu is 0.18 wt pct. This is consistent with 0.2-wt pct Cu found by the chemical analysis of the levitated drop. An issue that should be addressed is the high level of Cu observed near the surface of the particles. The phase diagram in the L + V (liquid–vapor) region suggests that there should be strong coring with the high level of Cu at the surface of the particle, which is expected to homogenize in a matter of few microseconds, considering its extremely small size. This should happen even in solidified particles while they are passing over the levitated drop because they take as much as 3 milliseconds to cross the levitated drop. The coring that might occur during solidification is likely to be insignificant, considering the narrow solidification zone and fairly high cooling rate of the order of 105 K/s (Figure A5). However, a fairly high level of Cu was observed at the surface. It seems that there should be some other force that sustains the high level of Cu near the surface. A study by Mukherjee et al.[24] on the surface segregation in a transition metal bimetallic alloy cluster suggests that Cu as a solute has a tendency to be on the surface in the Fe-Cu alloy, and this behavior also was confirmed experimentally. Furthermore, it should be noted that the surface energy of liquid Cu at its melting point of 1357 K (1084 °C) is 1.36 J/m2, whereas that of Fe at its melting point of 1809 K (1536 °C) is 1.87 J/m2.[25] It is expected that the surface energy of Cu can reduce even further, as the temperature of FeCu liquid droplets is expected to be higher than the melting point of Cu. This low surface energy of Cu in comparison with that of Fe could work against the homogenizing tendency, resulting in a balance that maintains a relatively high level of copper at surface, which is consistent with our observations.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig10_HTML.gif
Fig. 10

The liquid and vapor phase boundary of the Fe-Cu system calculated based on THERMOCALC data

Particle size distribution shows that it follows log-normal distribution. However, unlike other vapor-condensation processes, the levitated drop technique shows a much broader distribution, and particularly, particles with larger diameters are significant. This could be a result of a much longer time available for the growth of the droplet nucleated at the entry side of the levitated drop, as stated earlier. Mixing helium with argon or using helium gas alone narrows down the particle size distribution, as reported by Jigatch et al. in Al powder; however, this is suitable only for low-melting-temperature metals and alloys. In the present study on the FeCu system, mixing of even 5 vol pct of He with Ar reduced the levitated drop temperature drastically and resulted in an insignificant nanopowder yield.

As stated earlier, we observe only agglomerates at the end of the apparent blank zone below the levitated drop. It seems reasonable to believe, based on the estimated temperature of the nanoparticles, that the initial stage of agglomerate formation occurs in this apparently blank zone immediately below the levitated drop surface. The hot nanoparticles that have just formed have very clean surfaces, and when two such particles impinge on each other, they can become bonded. The bond, which is a point, initially can develop rapidly into a large interface, producing a thick neck. Because the nanoparticles have a very large surface-to-volume ratio, surface diffusion can be expected to play a predominant role in the development of the interfaces. TEM studies showed that each of the particles is a single crystal. This is to be expected, as the atoms that deposit on the particle prefer to register with an existing lattice rather than initiating a new one that requires it to go over the nucleation barrier. Some literature[26,27] indicates that the agglomerates have a chain-like structure. After several attempts in SEM and TEM, we could isolate a few very small agglomerates made up of a chain of individual nanoparticles. Their structure seemed to be a part of a three-dimensional network in the early stage of evolution. The chains can develop into a three-dimensional network by either branching or by impinging the chains with one another. Branching develops nodes with three branches, whereas the impingement of chains on each other produces nodes with four branches. Both kinds of nodes could be seen in the TEM micrographs of the agglomerates shown in Figure 11. This three-dimensional structure seems to enclose a large empty space. Thus, we have a low packing density. This poor density gets even worse when these agglomerates are packed together to produce compacts because of a high inter-agglomerate friction caused by mechanical interlocking. Such loose packing of particles with large and clean surfaces provides excellent access to gases. This explains the rapid oxidation of the powder mass when it suddenly is exposed to air or oxygen at atmospheric pressure. Therefore, the powder is passivated by letting in passivating gases, usually oxygen or air at a very low and controlled rate so that the surface is covered with oxides without raising the temperature significantly. The temperature at the start of the passivation in the present experiment was 308 K (35 °C), which rises to 311 K (38 °C) during the initial 20 to 30 minutes and then falls back to 308 K (35 °C). Attempts were made to passivate with nitrogen that could cover the surface of the powder by adsorption so that the latter can be desorbed when required. Though a satisfactory level of passivation was achieved, it was observed that the powder contains oxides in small quantity. It is believed that a small amount of air entered the system through a leakage in the apparatus and caused a lower rate of oxidation. Attempts to passivate with nitrogen without causing oxidation currently are in progress.
https://static-content.springer.com/image/art%3A10.1007%2Fs11663-010-9370-8/MediaObjects/11663_2010_9370_Fig11_HTML.jpg
Fig.11

TEM micrograph of FeCu nanopowder showing chain-like features with branching and nodes

An interesting issue of fundamental importance is the formation of nanoparticle aggregates in a chain-like configuration. Such behavior was observed in all types of material, including metals and oxides.[2830] Tang et al.[30] in their article on the one-dimensional assembly of nanoparticles indicated several driving forces that can lead to template-free self-assembly forming chain-like agglomeration. Out of these, magnetic dipole moments and crystallographically oriented agglomeration are relevant to LGC. As stated earlier, the individual particles produced are single crystals. Because their size is smaller than the magnetic domain size of bulk iron, they are single domain, single crystal magnetic particles. Therefore, they can assemble along their magnetic axis. Because the magnetic vectors of the individual particles are oriented along certain specific crystallographic direction, the chains formed in the magnetic materials also may be referred to as crystallographically oriented agglomerates. The merit of magnetic dipole driven agglomeration is that each of the particles is associated with a magnetic field, which can exert a physical force on other similar particles that are present in, or enter, its magnetic field. When the particles come together, they align their magnetic vector before impingement. An attempt to prove the alignment of magnetic axis in the chain using TEM were unsuccessful because of beam drift. Because of the existence of physical forces in the agglomerate formation of magnetic particles, the growth rate of agglomerates is expected to be much faster than, crystallographically oriented growth alone. Because only the magnetic vector between adjacent particles is aligned, the grain boundary or interface has 1 degree of freedom (i.e., rotation about the magnetic vector), which remains unfixed. Therefore, they are twist boundaries, which are likely to have lower energy than general grain boundaries. This situation should result in a large dihedral angle between adjacent particles. The development of a thick neck between adjacent particles, as shown in Figure 7, supports this view.

We believe that the agglomeration will occur in the blank zone. Heat transfer calculation (Appendix A) performed for a particle with a 36-nm diameter indicates that the particles cool to Curie temperature just 2 mm below the levitated drop in the blank zone. Because the particles forming the chain are generally around 30- to 40-nm diameter, they must cool to Curie temperature much sooner before reaching the end of the blank zone.

Conclusions

A nanopowder of 96Fe:4Cu was made by the levitational gas condensation method. The resultant powder was studied for physical, microstructural, and chemical characteristics. An attempt was made to understand the characteristics through a liquid–vapor phase diagram calculated using the available thermodynamic data, process modeling through heat transfer analysis, and magnetic nature of the powder.

The composition of the levitated drop giving out a vapor of 4-pct Cu in steady state matched well with the prediction of the calculated phase diagram. Though a high level of Cu at the surface of the powder is to be expected from the phase diagram, it takes a negligibly short time for complete homogenization because of the small size and prevalent high temperature. It was argued that the low surface energy of Cu is the major driving force that could stabilize Cu at the surface of the powder.

Modeling of the process involving heat transfer analysis, which indicated that most droplets can undergo solidification while they are still in the proximity of the levitated drop. During the period of residency close to the levitated drop, they can grow by intake of vapor. Once the particles start moving away from the levitated drop, they cool rapidly. Although agglomeration can initiate above the curie temperature, it proceeds rapidly below the curie temperature under the influence of the magnetic field of the particles. The chain-like structure formed, thus, occasionally can develop branching, leading to the formation of a reticulated three-dimensional sponge like structure that is responsible for the observed very low packing density of the resultant powder.

Acknowledgments

The authors acknowledge Dr. Sandeep Bysakh for his help with TEM investigation and Dr. Sankara Subramanian, DMRL, Hyderabad, for his help with developing a code for the numerical solution of the heat transfer problem.

Copyright information

© THE MINERALS, METALS & MATERIALS SOCIETY and ASM INTERNATIONAL 2010