- Email: [email protected]
Preparation of iron nanoparticles by chemical vapor condensation
October 2002
Materials Letters 56 (2002) 289 – 294 www.elsevier.com/locate/matlet
Preparation of iron nanoparticles by chemical vapor condensation C.J. Choi a,*, O. Tolochko b, B.K. Kim a a
Korea Institute of Machinery and Materials 66, Sangnam-Dong, Changwon, Kyungnam 641-010, South Korea b Material Science Faculty, State Technical University, 195251 Saint Petersburg, Russia Received 11 August 2001; accepted 15 August 2001
Abstract Iron nanoparticles were synthesized by the chemical vapor condensation (CVC) process using the pyrolysis of iron pentacarbonyl (Fe(CO)5). The influence of CVC parameter on the formation of nanoparticle and size distribution was studied. The synthesized nanoparticles consisted of core-shell type structure with nearly spherical shape and 6 – 25 nm in mean diameter. Obtained particles were studied by transmission electron microscopy and X-ray diffraction methods. The decrease of lattice constant of iron with increasing particle size was explained by the core – shell interaction. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Chemical vapor condensation; Iron; Nanoparticles; Size distribution; Lattice constant
1. Introduction In recent years, much attention has been paid to the synthesis and investigation of nanoparticles because of their wide range of potential applications [1]. Particularly, magnetic nanoparticles can have the special characteristic of exhibiting single-domain magnetism and can be used in magnetic tapes, ferrofluid, magnetic refrigerants, etc, because of their ultrafine size less than magnetic domain size [2]. A wide range of techniques to fabricate nanoparticles has been developed rapidly over the past decades [3– 5]. Among them, chemical synthesis of nanoparticles is a rapidly growing field because of its versatile
applicability to almost all materials and high rate of production capability with little agglomeration. Since the properties of these nanoparticles are basically determined by their mean size, size distribution, external shape, internal structure, and chemical composition, the characteristics of powders must be controlled during the production of the nanoparticles so that they are suitable to specific applications [3]. In this paper, we synthesized iron nanoparticles by chemical vapor condensation (CVC) by the pyrolysis of organometallic precursor of Fe(CO)5. The effect of processing parameters and annealing process on the microstructure and size of Fe nanoparticles were investigated.
2. Experimental details *
Corresponding author. Tel.: +82-551-280-35-32; fax: +82551-280-35-99. E-mail address: [email protected] (C.J. Choi).
The basic setup for CVC is similar to that described in literature elsewhere [6]. To produce iron na-
0167-577X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 7 7 X ( 0 2 ) 0 0 4 5 7 - 3
290
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
Fig. 1. TEM micrographs of iron nanoparticles. Insertion shows HRTEM image of oxide – metal phase interface.
noparticles, a carrier gas of high-purity argon or helium is fed through a heated bubbling unit containing the liquid iron pentacarbonyl (Fe(CO)5) precursor. Vaporization temperature of iron carbonyl was optimized at 150 jC. The flow of carrier gas entraining precursor vapor passed through the heated tubular
furnace to the work chamber. The precursor decomposed in that furnace and condensed in the clusters or particles. The experiment was conducted with the tubular furnace heated at temperature in the range of 400 – 1100 jC. All particles were well deposited on the surface of liquid nitrogen-cooled chiller in the work chamber, from which powders can be scrapped off and collected. To collect nanoparticles, the mixing gas of Ar contained a trace O2 flowed for 10 h with pressure of 1 kg/cm2 for a passivation process, which slowly oxidized nanoparticles. The morphologies and particle size distributions were determined by transmission electron microscopy using JEOL JEM-2000FXII equipment. The powder for TEM investigations was ultrasonically dispersed in ethanol and dropped on a carbon-coated copper grid. The average particle size of each sample was calculated as the center of gravity from particle size distribution. The phase analysis of samples was carried out on RIGAKU Geigerflex diffractometer with monochromatic CuKa radiation. The lattice parameter of the BCC phase was determined by extrapolation technique (see, for example, Klug and Alexander [7]). The silicon was
Fig. 2. X-ray diffraction patterns of as-produced iron nanoparticles with median diameter of 19 nm (1), 12 nm (2), and 8 nm (3) (a). Nanoparticles of 25 nm size: as-produced (1), after heating up to 220 jC (2), and 320 jC (3) in argon atmosphere (b).
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
291
used as an internal standard. The reproducibility was generally found to be better than F 0.0005 A.
3. Results and discussions TEM images of nanoparticles are shown in Fig. 1. They form intricate long stands to minimize the magnetic energy. All particles consist of dark core and light shell; the particle shape is nearly spherical. The core is metallic and the shell is composed of metal oxides [8]. The shell thickness is about 3– 4 nm irrespective of particle size. In order to investigate the effect of decomposition temperature on particle size, several particle samples were synthesized at various pyrolytic temperatures from 400 to 1100 jC, using argon or helium as a carrier gas; meanwhile, other preparation conditions were the same. Fig. 2 illustrates XRD patterns of the samples formed at different decomposition temperatures. It is shown that crystalline iron nanoparticles were synthesized by CVC. The peak intensity increases and its width decreases with increasing the decomposition temperature. No visible peaks of BCC phase were found for the samples produced at precursor decomposition temperature less than 750 jC in helium atmosphere. X-ray diffraction pattern does not show any additional peaks except peaks from BCC iron. However, slight diffusive peaks, which can belong to oxides, are distinguished. Fig. 3a presents the normalized size distributions for particles prepared at decomposition temperatures of 400, 750, and 1100 jC using argon as a carrier gas. The figures show that the mean particle size decreases with decreasing decomposition temperature of precursor. At the decomposition temperature of 400 jC, particles have a minimal size and actual symmetric size distribution. At lower temperatures, powders could not be synthesized. As the decomposition temperature increased, the distribution became wider and more asymmetric. The increase of median particle diameter with increasing the decomposition temperature in argon and helium atmospheres is shown in Fig. 3b. The increase of saturation vapor pressure with an increase of the decomposition temperature can enhance the growth of nucleus, which results in larger particle formation. Also, the high kinetic energy of gas mole-
Fig. 3. Particle size distributions (a) and median particle diameter (b) versus precursor decomposition temperature for iron nanoparticles, which were synthesized using argon or helium as carrier gas.
cules and as-formed iron particles in the gas phase can lead to the increase in number of collision between the particles and, consequently, to the preferable growth of larger ones. The size of particles prepared under helium atmosphere is significantly smaller, probably due to the higher mobility and heat conductivity of He, which lead to more rapid cooling of particles in the environment of He gas as compared with Ar. Fig. 4a,b shows normal probability and log probability plots of the data shown in Fig. 3a. The straight line in the corresponding plot indicated that the distribution is normal or lognormal. For the particles prepared at the decomposition temperature of 400 jC,
292
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
size can be explained by the interaction between the metallic core and oxide shell if the growth of oxide is assumed to be epitaxial (Fig. 1). Fung et al. [9] also showed that the epitaxial growth of oxide shell on the iron nanoparticles has a lattice misfit of about 3%. That can lead to compressive stresses induced in oxide shell and tensile stresses in metallic core, which causes increasing lattice constant in oxide-coated nanoparticles. Another possible reason for lattice parameter increase is the influence of dissolved interstitial atoms. The admixtures of interstitial atoms such as carbon or oxygen can be introduced in the lattice of iron particles during their formation by the vapor condensation and then fixed by subsequent rapid
Fig. 4. Normal-probability (a) and log-probability (b) plots of the data shown in Fig. 3a.
the particle size distribution is more close to normal than lognormal ones; the correlation coefficients with the straight line are equal to 0.93 and 0.99 for the lognormal and normal distribution, respectively. According to Ref. [3], it may mean that absorption growth mechanism predominates at lowest decomposition temperatures; particle growth is going on by separate atoms absorption. The increase of decomposition temperature leads to coalescence growth mechanism predomination; the distribution becomes lognormal, and the mean particle size increases. If particle size increases above 8 nm, the crystalline X-ray patterns of BCC solid solution can be detected and thus allow lattice parameter calculations. The measured values of lattice constant are larger than ˚ ) and increase that of pure iron (a(BCC – Fe) = 2.8664 A with decreasing particle size as shown in Fig. 5. Dependence of lattice constant on average particle
Fig. 5. The lattice constant of BCC iron as a function of mean particle diameter (a) and heating temperature (b).
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
293
quenching (estimated cooling rate in CVC process is about 106 K/s [3]). The second reason can cause a general increase of lattice constant, which is independent of particle size. Fig. 6 shows TGA curves that were obtained upon heating in air and argon atmospheres. Initial weight loss together with the endothermic effect is associated with moisture evaporation. Subsequent heating leads to weight increase due to the oxidation of iron cores. The oxidation is going on from the surface to center of each particle and leads to the increasing shell thickness (Fig. 7). The first and second stages of weight increase correspond to iron cores oxidation to Fe3O4 and Fe2O3, respectively. In the case of oxidation in air atmosphere, the experimentally observed Fig. 7. TEM micrographs of iron particles: as-prepared (a) and after low temperature heat treatment (b).
weight increase is in good correspondence with the calculated one. The calculations, which were made for full oxidation of round particle of 25-nm diameter with shell thickness of 3.5 nm, show an estimated value of mass increasing to be about 122%. To achieve a slow growth of oxide shells, the lowtemperature heat treatment was carried out in a flow atmosphere of argon containing about 10 4 vol.% of oxygen. In the argon atmosphere, the oxidation processes come very slowly and exactly two separated stages of oxidation were observed (Fig. 6). X-ray diffraction results and BCC iron lattice constant versus heating temperature are shown in Figs. 2b and 5b, respectively. We believe that the drastic decreasing of lattice parameter of BCC phase during low-temperature heating is due to the formation of defect structure in the oxide – metal phase interface by the oxide film growth and relaxation of internal stresses.
4. Conclusion
Fig. 6. Relative mass changes (M/M0) (a) and mass changes rate (dM/dT) versus temperature upon heating at the rate of 10 K/min in air (solid lines) and argon (dashed lines).
Iron nanoparticles were successfully synthesized by the CVC process using iron pentacarbonyl as a precursor under Ar or He atmosphere. The spherical nanoparticles of the mean diameter of 6 – 25 nm comprise the metal core and oxide shell. Average particle size increases and size distribution becomes
294
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
wider and more asymmetric with increasing decomposition temperature. Particles produced by CVC have a larger size if argon as carrier gas was used. The increase of lattice parameter of metallic core with the decreasing particle size can be explained by the interaction between metal cores and oxide shells.
References [1] L. Hu, M. Chen, Mater. Chem. Phys. 43 (1996) 212 – 219. [2] Hadjipanayis, G.A. Prinz, Science and Technology of Nanostructured Magnetic Materials, Plenum, New York, 1991.
[3] T. Sugimoto (Ed.), Fine Particles—Synthesis, Characterization and Mechanism of Growth, Marcel Dekker, New York, 1996. [4] W.R. Canon, S.C. Danforth, J.H. Flint, J.S. Haggerty, R.A. Mara, J. Am. Ceram. Soc. 65 (1992) 330 – 341. [5] S. Veintemillas-Verdaguer, M.P. Morales, C.J. Serna, Mater. Lett. 35 (1998) 227 – 231. [6] W. Chang, G. Skandan, S.C. Danforth, B. Kear, Nanostruct. Mater. 4 (1994) 507 – 520. [7] H.P. Klug, L.E. Alexander, The Precision Determination of Lattice Constant, X-Ray Diffraction Procedures, Wiley, 1974, pp. 594 – 597. [8] C.J. Choi, X.L. Dong, B.K. Kim, Scr. Mater. 44 (2001) 2225 – 2229. [9] K.K. Fung, B. Qin, X.X. Zhang, Mater. Sci. Eng. A 286 (2000) 135 – 138.
Materials Letters 56 (2002) 289 – 294 www.elsevier.com/locate/matlet
Preparation of iron nanoparticles by chemical vapor condensation C.J. Choi a,*, O. Tolochko b, B.K. Kim a a
Korea Institute of Machinery and Materials 66, Sangnam-Dong, Changwon, Kyungnam 641-010, South Korea b Material Science Faculty, State Technical University, 195251 Saint Petersburg, Russia Received 11 August 2001; accepted 15 August 2001
Abstract Iron nanoparticles were synthesized by the chemical vapor condensation (CVC) process using the pyrolysis of iron pentacarbonyl (Fe(CO)5). The influence of CVC parameter on the formation of nanoparticle and size distribution was studied. The synthesized nanoparticles consisted of core-shell type structure with nearly spherical shape and 6 – 25 nm in mean diameter. Obtained particles were studied by transmission electron microscopy and X-ray diffraction methods. The decrease of lattice constant of iron with increasing particle size was explained by the core – shell interaction. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Chemical vapor condensation; Iron; Nanoparticles; Size distribution; Lattice constant
1. Introduction In recent years, much attention has been paid to the synthesis and investigation of nanoparticles because of their wide range of potential applications [1]. Particularly, magnetic nanoparticles can have the special characteristic of exhibiting single-domain magnetism and can be used in magnetic tapes, ferrofluid, magnetic refrigerants, etc, because of their ultrafine size less than magnetic domain size [2]. A wide range of techniques to fabricate nanoparticles has been developed rapidly over the past decades [3– 5]. Among them, chemical synthesis of nanoparticles is a rapidly growing field because of its versatile
applicability to almost all materials and high rate of production capability with little agglomeration. Since the properties of these nanoparticles are basically determined by their mean size, size distribution, external shape, internal structure, and chemical composition, the characteristics of powders must be controlled during the production of the nanoparticles so that they are suitable to specific applications [3]. In this paper, we synthesized iron nanoparticles by chemical vapor condensation (CVC) by the pyrolysis of organometallic precursor of Fe(CO)5. The effect of processing parameters and annealing process on the microstructure and size of Fe nanoparticles were investigated.
2. Experimental details *
Corresponding author. Tel.: +82-551-280-35-32; fax: +82551-280-35-99. E-mail address: [email protected] (C.J. Choi).
The basic setup for CVC is similar to that described in literature elsewhere [6]. To produce iron na-
0167-577X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 7 7 X ( 0 2 ) 0 0 4 5 7 - 3
290
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
Fig. 1. TEM micrographs of iron nanoparticles. Insertion shows HRTEM image of oxide – metal phase interface.
noparticles, a carrier gas of high-purity argon or helium is fed through a heated bubbling unit containing the liquid iron pentacarbonyl (Fe(CO)5) precursor. Vaporization temperature of iron carbonyl was optimized at 150 jC. The flow of carrier gas entraining precursor vapor passed through the heated tubular
furnace to the work chamber. The precursor decomposed in that furnace and condensed in the clusters or particles. The experiment was conducted with the tubular furnace heated at temperature in the range of 400 – 1100 jC. All particles were well deposited on the surface of liquid nitrogen-cooled chiller in the work chamber, from which powders can be scrapped off and collected. To collect nanoparticles, the mixing gas of Ar contained a trace O2 flowed for 10 h with pressure of 1 kg/cm2 for a passivation process, which slowly oxidized nanoparticles. The morphologies and particle size distributions were determined by transmission electron microscopy using JEOL JEM-2000FXII equipment. The powder for TEM investigations was ultrasonically dispersed in ethanol and dropped on a carbon-coated copper grid. The average particle size of each sample was calculated as the center of gravity from particle size distribution. The phase analysis of samples was carried out on RIGAKU Geigerflex diffractometer with monochromatic CuKa radiation. The lattice parameter of the BCC phase was determined by extrapolation technique (see, for example, Klug and Alexander [7]). The silicon was
Fig. 2. X-ray diffraction patterns of as-produced iron nanoparticles with median diameter of 19 nm (1), 12 nm (2), and 8 nm (3) (a). Nanoparticles of 25 nm size: as-produced (1), after heating up to 220 jC (2), and 320 jC (3) in argon atmosphere (b).
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
291
used as an internal standard. The reproducibility was generally found to be better than F 0.0005 A.
3. Results and discussions TEM images of nanoparticles are shown in Fig. 1. They form intricate long stands to minimize the magnetic energy. All particles consist of dark core and light shell; the particle shape is nearly spherical. The core is metallic and the shell is composed of metal oxides [8]. The shell thickness is about 3– 4 nm irrespective of particle size. In order to investigate the effect of decomposition temperature on particle size, several particle samples were synthesized at various pyrolytic temperatures from 400 to 1100 jC, using argon or helium as a carrier gas; meanwhile, other preparation conditions were the same. Fig. 2 illustrates XRD patterns of the samples formed at different decomposition temperatures. It is shown that crystalline iron nanoparticles were synthesized by CVC. The peak intensity increases and its width decreases with increasing the decomposition temperature. No visible peaks of BCC phase were found for the samples produced at precursor decomposition temperature less than 750 jC in helium atmosphere. X-ray diffraction pattern does not show any additional peaks except peaks from BCC iron. However, slight diffusive peaks, which can belong to oxides, are distinguished. Fig. 3a presents the normalized size distributions for particles prepared at decomposition temperatures of 400, 750, and 1100 jC using argon as a carrier gas. The figures show that the mean particle size decreases with decreasing decomposition temperature of precursor. At the decomposition temperature of 400 jC, particles have a minimal size and actual symmetric size distribution. At lower temperatures, powders could not be synthesized. As the decomposition temperature increased, the distribution became wider and more asymmetric. The increase of median particle diameter with increasing the decomposition temperature in argon and helium atmospheres is shown in Fig. 3b. The increase of saturation vapor pressure with an increase of the decomposition temperature can enhance the growth of nucleus, which results in larger particle formation. Also, the high kinetic energy of gas mole-
Fig. 3. Particle size distributions (a) and median particle diameter (b) versus precursor decomposition temperature for iron nanoparticles, which were synthesized using argon or helium as carrier gas.
cules and as-formed iron particles in the gas phase can lead to the increase in number of collision between the particles and, consequently, to the preferable growth of larger ones. The size of particles prepared under helium atmosphere is significantly smaller, probably due to the higher mobility and heat conductivity of He, which lead to more rapid cooling of particles in the environment of He gas as compared with Ar. Fig. 4a,b shows normal probability and log probability plots of the data shown in Fig. 3a. The straight line in the corresponding plot indicated that the distribution is normal or lognormal. For the particles prepared at the decomposition temperature of 400 jC,
292
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
size can be explained by the interaction between the metallic core and oxide shell if the growth of oxide is assumed to be epitaxial (Fig. 1). Fung et al. [9] also showed that the epitaxial growth of oxide shell on the iron nanoparticles has a lattice misfit of about 3%. That can lead to compressive stresses induced in oxide shell and tensile stresses in metallic core, which causes increasing lattice constant in oxide-coated nanoparticles. Another possible reason for lattice parameter increase is the influence of dissolved interstitial atoms. The admixtures of interstitial atoms such as carbon or oxygen can be introduced in the lattice of iron particles during their formation by the vapor condensation and then fixed by subsequent rapid
Fig. 4. Normal-probability (a) and log-probability (b) plots of the data shown in Fig. 3a.
the particle size distribution is more close to normal than lognormal ones; the correlation coefficients with the straight line are equal to 0.93 and 0.99 for the lognormal and normal distribution, respectively. According to Ref. [3], it may mean that absorption growth mechanism predominates at lowest decomposition temperatures; particle growth is going on by separate atoms absorption. The increase of decomposition temperature leads to coalescence growth mechanism predomination; the distribution becomes lognormal, and the mean particle size increases. If particle size increases above 8 nm, the crystalline X-ray patterns of BCC solid solution can be detected and thus allow lattice parameter calculations. The measured values of lattice constant are larger than ˚ ) and increase that of pure iron (a(BCC – Fe) = 2.8664 A with decreasing particle size as shown in Fig. 5. Dependence of lattice constant on average particle
Fig. 5. The lattice constant of BCC iron as a function of mean particle diameter (a) and heating temperature (b).
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
293
quenching (estimated cooling rate in CVC process is about 106 K/s [3]). The second reason can cause a general increase of lattice constant, which is independent of particle size. Fig. 6 shows TGA curves that were obtained upon heating in air and argon atmospheres. Initial weight loss together with the endothermic effect is associated with moisture evaporation. Subsequent heating leads to weight increase due to the oxidation of iron cores. The oxidation is going on from the surface to center of each particle and leads to the increasing shell thickness (Fig. 7). The first and second stages of weight increase correspond to iron cores oxidation to Fe3O4 and Fe2O3, respectively. In the case of oxidation in air atmosphere, the experimentally observed Fig. 7. TEM micrographs of iron particles: as-prepared (a) and after low temperature heat treatment (b).
weight increase is in good correspondence with the calculated one. The calculations, which were made for full oxidation of round particle of 25-nm diameter with shell thickness of 3.5 nm, show an estimated value of mass increasing to be about 122%. To achieve a slow growth of oxide shells, the lowtemperature heat treatment was carried out in a flow atmosphere of argon containing about 10 4 vol.% of oxygen. In the argon atmosphere, the oxidation processes come very slowly and exactly two separated stages of oxidation were observed (Fig. 6). X-ray diffraction results and BCC iron lattice constant versus heating temperature are shown in Figs. 2b and 5b, respectively. We believe that the drastic decreasing of lattice parameter of BCC phase during low-temperature heating is due to the formation of defect structure in the oxide – metal phase interface by the oxide film growth and relaxation of internal stresses.
4. Conclusion
Fig. 6. Relative mass changes (M/M0) (a) and mass changes rate (dM/dT) versus temperature upon heating at the rate of 10 K/min in air (solid lines) and argon (dashed lines).
Iron nanoparticles were successfully synthesized by the CVC process using iron pentacarbonyl as a precursor under Ar or He atmosphere. The spherical nanoparticles of the mean diameter of 6 – 25 nm comprise the metal core and oxide shell. Average particle size increases and size distribution becomes
294
C.J. Choi et al. / Materials Letters 56 (2002) 289–294
wider and more asymmetric with increasing decomposition temperature. Particles produced by CVC have a larger size if argon as carrier gas was used. The increase of lattice parameter of metallic core with the decreasing particle size can be explained by the interaction between metal cores and oxide shells.
References [1] L. Hu, M. Chen, Mater. Chem. Phys. 43 (1996) 212 – 219. [2] Hadjipanayis, G.A. Prinz, Science and Technology of Nanostructured Magnetic Materials, Plenum, New York, 1991.
[3] T. Sugimoto (Ed.), Fine Particles—Synthesis, Characterization and Mechanism of Growth, Marcel Dekker, New York, 1996. [4] W.R. Canon, S.C. Danforth, J.H. Flint, J.S. Haggerty, R.A. Mara, J. Am. Ceram. Soc. 65 (1992) 330 – 341. [5] S. Veintemillas-Verdaguer, M.P. Morales, C.J. Serna, Mater. Lett. 35 (1998) 227 – 231. [6] W. Chang, G. Skandan, S.C. Danforth, B. Kear, Nanostruct. Mater. 4 (1994) 507 – 520. [7] H.P. Klug, L.E. Alexander, The Precision Determination of Lattice Constant, X-Ray Diffraction Procedures, Wiley, 1974, pp. 594 – 597. [8] C.J. Choi, X.L. Dong, B.K. Kim, Scr. Mater. 44 (2001) 2225 – 2229. [9] K.K. Fung, B. Qin, X.X. Zhang, Mater. Sci. Eng. A 286 (2000) 135 – 138.