17102
J. Phys. Chem. C 2008, 112, 17102–17108
Growth of Nanoscale Magnetic Films Using a Supercritical CO2/Ferric Acetylacetonate Batch Process Near Room Temperature Silvia De Dea,† David R. Miller,†,* and Robert E. Continetti‡,* Department of Mechanical and Aerospace Engineering and Department of Chemistry and Biochemistry, UniVersity of California, San Diego, 9500 Gilman DriVe, La Jolla, California 92093-0001 ReceiVed: June 16, 2008; ReVised Manuscript ReceiVed: August 15, 2008
Magnetic nanoscale films have been produced using a batch process based on the decompression of a supercritical CO2/Ferric acetylacetonate (Fe(acac)3) solution near room temperature. Fe(acac)3 was solubilized in supercritical CO2 at 140 bar, and the saturated supercritical solution was decompressed at rates of 1.5 and 45 bar/sec, forming cluster films on both silicon and gold substrates. The resulting nanoscale films have particles in the range of 16 to 350 nm and values of magnetic coercivity in the range 50-115 Oe. These batch films are compared with films grown previously by a continuous free-jet process, using the same supercritical solutions but with decompression times on the order of microseconds. The experiments suggest that the initial restructuring and decomposition of Fe(acac)3 into a magnetically ordered material takes place in the supercritical fluid phase, without influence of the film substrate, well below the known thermal decomposition temperature of 180 °C. This new batch process is straightforward, does not require hightemperature decomposition of a precursor, utilizes nontoxic and inexpensive starting materials, and the CO2 solvent is readily removed by decompression of the supercritical mixture. 1. Introduction Magnetic nanoparticles, as iron oxides, have attracted a lot of interest due to their applications in recording media,1 ferrofluids, catalysts,2-5 and biomedical applications such as targeted drug delivery.6-9 New synthetic routes for the preparation of magnetic nanoparticles are under constant investigation, and some of them include precipitation of metal salts from an aqueous solution, sol-gel synthesis, and aerosol spray pyrolysis.6,10 Each of the methods offers advantages and disadvantages in controlling composition, size, size distribution, and crystallinity of the particles formed. Metal organic complexes, such as ferric acetylacetonate, Fe(acac)3, have been very popular as volatile sources of metals in chemical vapor deposition (CVD) approaches to the preparation of ceramic superconductors11 and iron oxide magnetic thin films10 but have required hightemperature thermal decomposition of the metal organic precursor. In particular, Fe(acac)3 is known to decompose into magnetic iron oxides when heated well above 180 °C.12 We have recently reported13 that magnetically ordered nanoparticle films were unexpectedly formed near room temperature using the rapid expansion of supercritical solutions (RESS) process. For this RESS technique, a supercritical solution of Fe(acac)3 in CO2, typically at 140 bar and 70 °C, was decompressed to 1 bar or less by expanding the fluid through a micron-sized nozzle aperture. The resulting supersonic free-jet expansion was directed onto a cold silicon substrate where film growth occurred. The decompression occurs on the time scale of microseconds, the solubility decreases dramatically, and the solute obtains a very high degree of supersaturation, which leads * To whom correspondence should be addressed. Tel: +18585343131. Fax:+18585345355.E-mail:
[email protected](D.R.M.),Tel:+18585345559. Fax: +18585349856. E-mail:
[email protected] (R.E.C.). † Department of Mechanical and Aerospace Engineering, University of California, San Diego. ‡ Department of Chemistry and Biochemistry, University of California, San Diego.
to solute precipitation from solution. With the RESS process, we were able to grow magnetically ordered nanoparticle films on both hot and cold substrates, well below the known Fe(acac)3 decomposition temperature. In this article, we report that similar nanoscale magnetically ordered films can be grown on lowtemperature substrates, without a RESS process, by decompressing the supercritical Fe(acac)3/CO2 solution in a much simpler batch process. The batch experiments discussed here were in fact motivated by results presented in the previous RESS study.13 To understand if the supersonic RESS free-jet expansion, which can provide considerable translational energy to impinging Fe(acac)3 molecules and clusters, could have a unique effect on the decomposition of Fe(acac)3 and subsequent growth into a magnetic material, it was decided to utilize the batch process in a closed cell. The rate of decompression varies dramatically between the two processes, RESS (µs scale) and batch (sec scale). Particle formation and growth also occurs according to different mechanisms for the two processes. While we have found no substantial evidence for the formation of nanoscale solute clusters in the rapid RESS expansion,14 so that the dominant cluster growth occurred on the substrate, it is clear that large clusters do precipitate in the bulk fluid phase in the batch process prior to film formation. Therefore the batch process films can be expected to show different morphologies. A batch process has several advantages over a continuous RESS process. Often requiring a much smaller amount of solvent, a batch process is easier to operate, many samples can be grown at the same time, and a larger area can be coated. Therefore, a batch process might be more suitable to scale up to an industrial process for thin film growth. Perhaps the most famous batch process that utilizes supercritical fluids is the highpressure polyethylene polymerization, first developed in the late 1930s,15 where high-pressure ethylene is both the reactant and the solvent for the product. A batch process which is particularly important to mention, given the similarity with the batch process
10.1021/jp805314e CCC: $40.75 2008 American Chemical Society Published on Web 10/14/2008
Growth of Nanoscale Magnetic Films
J. Phys. Chem. C, Vol. 112, No. 44, 2008 17103
Figure 2. Particles precipitating from solution detected with the cloud point method.
Figure 1. Batch experimental apparatus.
presented in this study, is chemical fluid deposition (CFD).16,17 CFD has been developed as an alternative to CVD, overcoming precursor volatility constraints, mass transport limitations, and environmental issues. Other examples of batch processes with supercritical fluids include supercritical fluid extraction for coffee decaffeination, dry cleaning of laundry, and processing of pharmaceutical compounds.15,18,19 In the sections below, we briefly describe the batch experimental apparatus and present SEM structural and SQUID magnetic data for the resulting thin nanoparticle films. Surprisingly, we again find that magnetically ordered nanoscale cluster films can be fabricated at low temperatures. We compare the physical and magnetic properties of the films grown with the batch and the previous RESS processes. 2. Experimental Apparatus and Procedure Batch experiments were performed inside a specially designed 2.3 cm3 volume UV-vis stainless steel cell, equipped with three quartz windows and capable of pressures up to 180 bar. The cell has been described in detail elsewhere,14,20 and has been used in our laboratory for cloud point measurements of Fe(acac)3 solute solubility in supercritical CO2. For a typical batch film growth experiment, a known amount of solute and a substrate are placed in the solubility cell, pressurized, allowed to reach equilibrium and depressurized by opening a vent valve. The apparatus, shown schematically in Figure 1, includes a CO2 siphon-tube gas cylinder, an ISCO syringe pump, and the UV-vis solubility cell. An Ocean Optics HR2000 fiber optic UV-vis spectrometer is used to continuously monitor the in situ concentration of Fe(acac)3 during the pressurization and decompression cycle. In addition, a CCD camera can be mounted above the cell to observe particle formation, detecting the light scattered from a laser beam at 90° through a third quartz window.13 The experimental procedure is as follows: approximately 20 mg of Fe(acac)3 powder are loaded into the cell, sufficient to fully saturate the solution. A wire mesh is placed above the powder to isolate Fe(acac)3 from the substrate (a 7 × 4 mm silicon wafer), which is placed above the mesh. Usually, three separate substrates are loaded at one time to ensure reproducibility of the magnetic measurements. The cell is then heated to 313-318 K, purged with 2 bar of CO2, pressurized to 140 bar and isolated from the pump. At the same time the data acquisition program is started to record temperature, pressure, and concentration, as described elsewhere.14 The system is then left at 140 bar for approximately 3 h to nearly approach a saturated solution and subsequently depressurized by opening a vent valve. Typically, the cell pressure was reduced to about 50 bar, well below the cloud point, held at this pressure for a few minutes so that solute readily precipitates out of solution and onto the film substrate, and finally reduced to atmospheric pressure. Solute particles precipitating in the CO2 fluid phase
can be observed with the CCD camera, which was also done to obtain cloud point solubility data.14,20 A picture of particles precipitating from the bulk supercritical phase is shown in Figure 2; the dark bar is a magnetic stirrer in the cell bottom and the laser light is observed, due to scattering into the camera, as a narrow white beam prior to crossing the cloud point. As the pressure is reduced below the cloud point, particles precipitate from solution and scatter the laser light. Such cloud point measurements show that substantial particle growth occurs in the fluid phase above the substrates, although additional growth could occur on the substrate. The substrates are then removed and the deposited nanoparticle films analyzed with a FEI Quanta 600 Scanning Electron Microscope (SEM), and a Quantum Design MPMS-XL5AC superconducting quantum interference device (SQUID) based magnetometer. Experimental parameters studied include decompression rate and substrate material. Particles were collected on both silicon and gold (99.9% purity) substrates. The gold substrate was used to ascertain if there was any chemical role of solute-surface interaction, which would be more likely with the silicon surfaces (discussed further below). Because particles did not adhere well to gold flat substrates, the experimental procedure was modified by collecting the particles that precipitate out of solution in a gold cup, formed from the gold foil. For this purpose, an additional 9.5 cm3 volume was appended to the UV-vis cell.14 In these gold cup experiments, Fe(acac)3 starting material was loaded only in the UV-vis cell to avoid direct contact between the powder and the gold cup, whereas the gold cup was placed in the appended volume. Once the system was vented, the particles were collected from the gold cup with transparent adhesive tape, and their magnetic properties were analyzed with the SQUID based magnetometer. 3. Results and Discussion We find that magnetically ordered nanoparticles were deposited on silicon substrates and were also collected in the gold cup using the batch process based on the decompression of a Fe(acac)3 in CO2 supercritical mixture. The effect of two different rates of decompression in the batch process was studied for thin nanoparticle films deposited on silicon substrates. We refer to these as a slow decompression (dP/dt = -1.5 bar/s) and a fast decompression (dP/dt = -45 bar/s). It will be shown below that different morphologies can be obtained by such a change in the rate of decompression, so that this is a straightforward process control variable. 3.1. Morphology and Structure of Batch Nanoparticle Films. Figures 3 and 4 show SEM images at different magnifications of a film grown using a slower decompression rate, dP/dt = -1.5 bar/sec, onto substrates with Ts = 313 K. Parts (a)-(c) of Figure 3 show general features of the film. The film is characterized by long branched filaments of particles agglomerated or fused together, which lie over a thin coating. In addition to the long filaments, there are scattered particles of variable size deposited over the entire substrate. Details of the filaments and particles are shown in parts (b) and (c) of
17104 J. Phys. Chem. C, Vol. 112, No. 44, 2008
De Dea et al.
Figure 4. SEM images of particles deposited on a cold silicon substrate; Po ) 140 bar, To ) 313 K, slow decompression (dP/dt = -1.5 bar/sec).
Figure 3. SEM images of a film deposited on a cold silicon substrate; Po ) 140 bar, To ) 313 K, slow decompression (dP/dt = -1.5 bar/ sec).
Figure 3 where higher magnification SEMs are shown. In particular, parts (b) and (c) of Figure 3 show that the filaments are made of two different configurations of particles, which suggests two different types of growth of film materials. In one arrangement, particles on the order of 20-75 nm appear to selforganize in elongated chainlike structures (part (b) of Figure 3). In the other arrangement, particles are fused together to form filaments several micrometers long and approximately 70 nm wide (part (c) of Figure 3). Parts (a) and (b) of Figure 4 are SEM images of the particles distributed over the substrate and not organized in elongated structures. A trimodal size distribution of near spherical particles can be identified on the substrate. The larger particles, shown in parts (a) and (b) of Figure 4, have a minimum size of 100 nm, maximum size of 200 nm, mean size of 150 nm, and standard deviation of 33 nm. In particular, part (b) of Figure 4 shows that those spherical particles are in reality made by an assembly of even smaller particles. The second size distribution is due to particles that form the chainlike structure shown in part (b) of Figure 3 and the smaller ensemble of particles shown in part (a) of Figure 4, which are characterized by a minimum size of 32 nm, maximum size of 75 nm, mean size of 50 nm, and standard deviation of 16 nm. The third size distribution was difficult to measure because it was composed of very small particles distributed over the entire substrate with a minimum size of 20 nm, maximum size of 28 nm, mean size of 22 nm,
and standard deviation of 3.5 nm. It was not possible with the SEM available to us to resolve particles