Conformal nanocoating of zirconia nanoparticles by ... - CiteSeerX
INSTITUTE OF PHYSICS PUBLISHING
NANOTECHNOLOGY
Nanotechnology 16 (2005) S375–S381
doi:10.1088/0957-4484/16/7/010
Conformal nanocoating of zirconia nanoparticles by atomic layer deposition in a fluidized bed reactor Luis F Hakim1 , Steven M George2 and Alan W Weimer1,3 1
Department of Chemical and Biological Engineering, University of Colorado, Boulder, CO 80309, USA 2 Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA E-mail: [email protected]
Received 11 January 2005, in final form 24 March 2005 Published 15 April 2005 Online at stacks.iop.org/Nano/16/S375 Abstract Primary zirconia nanoparticles were conformally coated with alumina ultrathin films using atomic layer deposition (ALD) in a fluidized bed reactor. Alternating doses of trimethylaluminium and water vapour were performed to deposit Al2 O3 nanolayers on the surface of 26 nm zirconia nanoparticles. Transmission Fourier transform infrared spectroscopy was performed ex situ. Bulk Al2 O3 vibrational modes were observed for coated particles after 50 and 70 cycles. Coated nanoparticles were also examined with transmission electron microscopy, high-resolution field emission scanning electron microscopy and energy dispersive spectroscopy. Analysis revealed highly conformal and uniform alumina nanofilms throughout the surface of zirconia nanoparticles. The particle size distribution and surface area of the nanoparticles are not affected by the coating process. Primary nanoparticles are coated individually despite their high aggregation tendency during fluidization. The dynamic aggregation behaviour of zirconia nanoparticles in the fluidized bed plays a key role in the individual coating of nanoparticles. (Some figures in this article are in colour only in the electronic version)
1. Introduction The interest in the use of nanoparticles, those with dimensions less than 100 nm, has continuously increased over recent years. The processing of nanoparticles offers the potential to design novel materials at the nanoscale. If the surface properties of nanoparticles could be modified while keeping their bulk properties, the performance of nanopowders for many applications could be greatly enhanced. The deposition of an extremely thin film on a substrate can alter the surface characteristics without degrading the bulk properties of the underlying material. Coating techniques such as chemical vapour deposition (CVD) have been utilized to coat particles. During a CVD reaction the chemical reactants are allowed to coincide 3 Author to whom any correspondence should be addressed.
in the gas phase making this technology dependent upon reaction time, exposure and temperature [1–5]. The nature of CVD can lead to films that have a thickness of 1 µm or more [6, 7], are non-conformal and highly granular [8] or are deposited around aggregates of particles rather than on primary particles [9–11]. Various CVD technologies such as metal– organic chemical vapour deposition (MOCVD) and plasmaenhanced chemical vapour deposition (PECVD) have also been applied to particles [12–15]. However, granular coatings are commonly observed [12] and uniform performance has not been achieved due to particle agglomeration during processing [13]. Atomic layer deposition (ALD) provides a unique method for depositing ultrathin films on surfaces [16–18]. ALD uses sequential surface reactions to coat substrates with high conformality and precise thickness control at the atomic scale.
0957-4484/05/070375+07$30.00 © 2005 IOP Publishing Ltd Printed in the UK
S375
L F Hakim et al
1
To pump
2
3
1 8
To N2 supply
4
6 5
7
7
Figure 1. Schematic diagram of ALD-FBR apparatus: (1) pressure transducers, (2) sintered-metal filter, (3) reaction column, (4) vibro-motors, (5) spring supports, (6) pneumatic valve, (7) reactant containers, (8) mass flow controller.
Coating schemes have been developed for Al2 O3 ALD using a sequence of surface reactions [19–26]. Using doses of trimethylaluminium (TMA) and water in an ABAB. . . sequence alumina films can be deposited. The growth of ultrathin and conformal Al2 O3 nanolayers using sequential surface reactions is based on the binary reaction: 2Al(CH3 )3 + 3H2 O → Al2 O3 + 6CH4
(1)
Al2 O3 ALD is achieved by splitting this binary reaction into two separate half-reactions [19–22]:
2. Experimental details
(A) AlOH∗ + Al(CH3 )3 → AlOAl(CH3 )∗2 + CH4
(2)
(B) AlCH3∗ + H2 O → AlOH∗ + CH4
(3)
where ∗ indicates a surface species. During each dose, the reactants completely saturate the active sites on the substrate, making this a self-limiting and self-controlling process. The thickness of the film deposited after sequential exposures of Al(CH3 )3 and H2 O is nominally 1.1 Å per coating cycle at 450 K [19]. The growth of ultrathin films on nanoparticles has been successfully demonstrated at small scale [27–30] for several ALD systems. In these studies, in situ FTIR spectroscopy was used to monitor the surface chemistry during the sequential exposures. Metal and ceramic micron-sized particles have been coated with alumina using ALD in a fluidized bed S376
reactor [31–33]. Bulk processing of ultrafine particles allowed testing the improved properties of coated particles. Enhanced oxidation resistance and increased surface reactivity are some examples of these superior characteristics. As the interest in the use of nanoparticles continuously grows the need for bulk processing also rises. The present work demonstrates that primary zirconia nanoparticles can be conformally and uniformly coated at large scale with Al2 O3 using ALD in a fluidized bed reactor.
A schematic diagram of the atomic layer deposition fluidized bed reactor (ALD-FBR) is shown in figure 1. The ALD reactor is a 4.6 cm inside diameter (ID) stainless steel column with a 20 µm pore size porous metal disc as the gas distributor. A 316 L porous metal filter element (1.9 cm ID × 15.24 cm long; 0.5 µm pore size) was used at the inside top of the reactor column to prevent particles from leaving the system. Due to their small size, zirconia nanoparticles fluidize as aggregates and remain in the bed during the coating process. The fluidization system was capable of operating at reduced pressures with mechanical vibration. Reduced pressure was achieved in the fluidization column by controlling the opening of a Nupro® LD series diaphragm valve placed at the inlet of an Alcatel 2063 vacuum pump. Vibration was
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
Figure 2. (a) and (b). Transmission electron micrographs of uncoated zirconia nanoparticles.
applied to the system using two synchronized vibro-motors from Martin Engineering having variable amplitude. Vibration frequency was controlled using an ACS 140 speed controller from ABB Drive and Power Products. The vibro-motors were attached to a custom-built vibration platform. Four spring supports on this platform gave homogeneous vertical vibration to the entire system. The fluidization column was attached to this platform via a mounting assembly. Details are shown in figure 1. The fluidized system was controlled and monitored using LabView® . A script recorded the pressure drop across the bed of powder and allowed control of the fluidizing gas flowrate and all the system valves. High-purity nitrogen was used as the fluidizing gas. Its flowrate was controlled by an MKS® 1179 series mass flow controller. MKS® 902 series piezo transducers were located below the distributor plate and at the outlet of the fluidization column to measure the pressure drop across the bed of powder. During operation the system was vibrated to improve the fluidization of nanoparticles. A vibration frequency of 20 Hz and amplitude of 3.3 mm were used. A vacuum trap with an activated alumina insert from Laco Technologies was located at the inlet of the pump to prevent damage to the vacuum system. Particles were obtained from Nanoproducts Corp. They were 26 ± 3 nm in primary particle size and had a BET surface area of 40.9 ± 1.9 m2 g−1 . The bulk density of the powder was 320 kg m−3 and the density of primary nanoparticles was 5890 kg m−3 . Zirconia nanoparticles were not perfectly spherical and showed high agglomeration tendency as observed in figures 2(a) and (b). Some nanoparticles may show sintered regions to some extent. The resulting structures are also nanosized. This condition is not the result of the fluidized bed ALD processing. In order to perform the alumina-ALD reactions, TMA obtained from Sigma Aldrich and deionized water were used as reactants. A series of pneumatically activated SwagelokTM diaphragm valves controlled the automatic and sequential dosing of reactants during the coating cycles. The reactants were fed via their vapour pressures and the system was kept at low pressure (∼1 Torr) at all times. The entire reactor was encased by a clamshell-type furnace from Marshall Furnace Co. The reaction temperature was 450 K. The feeding lines were also heated to approximately 350 K using heavy-duty heating tape. This helped to elevate the temperature of gases before reaction as well as reducing the deposition of gaseous species on the internal walls of the feeding lines.
Prior to loading, powders were sieved using a 40 mesh screen. This screening was intended to eliminate large stationary aggregates (>420 µm) formed during shipping and storage. Particles were then dried at 400 K and 0.5 Torr for 2 h inside the reactor and under a continuous flow of nitrogen. The pressure inside the reactor was monitored at all times. As observed in figure 3, the half-reactions during a coating cycle can be easily identified by the characteristic dosing pressures of TMA and water. After each reactant dose, the system was flushed with inert gas to eliminate unreacted species as well as any methane formed during reaction. The reactants were fed for enough time so saturation of all active sites occurred for every dose. This time was calculated based upon the surface area of the nanoparticles and the sample mass utilized for each batch. Typical batch volumes for ALD processing of nanoparticles were of the order of 120 cm3 . The coating cycle was automatically repeated until the desired film thickness was achieved.
3. Results and discussion 3.1. Fluidization behaviour of zirconia nanoparticles Since the ALD coating was performed in a fluidized bed reactor, understanding the fluidization of zirconia nanoparticles is a key step in improving the coating process. In general, due to high cohesive forces, nanoparticles fluidize as aggregates [34–38]. This means that properties of aggregates, rather than those of primary particles, will determine the fluidization properties of nanoparticles. Aggregate properties such as size, density and sphericity will determine fluidization properties such as bed expansion and minimum fluidization velocity. A fluidized bed offers the advantage of high heat and mass transfer coefficients due to constant solid recirculation and excellent fluid–solid contacting. In an ALD system, this characteristic would help achieve fast saturation of active sites on the particle surface. Therefore, fluidization needs to be assured at all times during processing. Figure 4 shows a plot of pressure drop across the bed of particles versus superficial gas velocity. This plot is used to determine the minimum fluidization velocity of particles. Incipient fluidization is achieved when the pressure drop remains constant. For the case of zirconia nanoparticles, this value is about 0.035 cm s−1 . This shows that zirconia nanoparticles fluidize despite their small size and high cohesive forces and that this system is suitable for ALD processing. 3.2. Test for conformality, uniformity and composition of films A Philips CM 10 transmission electron microscope (TEM) was used to analyse the conformality of alumina films deposited on zirconia nanoparticles. TEM analysis was performed at 100 kV. Figures 5(a) and (b) show two micrographs of coated nanoparticles after 50 cycles. From visual inspection, the growth rate is calculated to be in the range of 1–1.4 Å/cycle. It is important to mention that the growth rate of films may vary with the size and geometry of the substrate. For particles with a high ratio of curvature, more active sites on the surface are exposed to the gas phase S377
L F Hakim et al
1.4 H2O Dose
1.2 TMA Dose
Pressure (Torr)
1 0.8 0.6
N2 Purge
N2 Purge
0.4 0.2 0 0
50
100
150
200
250
300
350
Time (sec) Figure 3. Pressure profile inside the ALD fluidized-bed reactor during a coating cycle.
0.4
Pressure Drop (kPa)
0.35 0.3 0.25 0.2 0.15 0.1 0.05 0
0.01
0.02
0.03
0.04
0.05
Superficial Gas Velocity (cm/s) Figure 4. Pressure drop across the fluidized bed versus superficial gas velocity for zirconia nanoparticles.
reactants. This is explained because less steric effect occurs during the adsorption step of the reaction, thus higher growth rates are expected. The TEM micrographs show extremely conformal and uniform films around nanoparticles. In figure 5(a), particles appear aggregated due to high cohesive forces commonly observed in nanoparticles. An important observation is the fact that films deposited on different nanoparticles are clearly differentiated. This can be observed on the sharp edge formed between coated nanoparticles. Figure 5(b) shows individually coated primary nanoparticles with a very uniform film throughout their surface. In order to further test the homogeneity of the deposited films, a high-resolution microscopy analysis was performed. A Hitachi S-5200 field emission scanning electron microscope (FESEM) was used. This equipment was coupled with a Princeton Gamma Tech Spirit energy dispersive spectrometer (EDS) with a Li doped Si detector. The EDS analysis was performed at 15 kV. Analyses were performed by Geoff S378
Courtin and Dr William Kroenke at the Center for MicroEngineered Materials at the University of New Mexico. Figures 6(a) and (b) show two sets of FESEM images on the left and their corresponding aluminium-EDS signal on the right. As observed in the images, the aluminium-EDS signal matches the shape of the coated particles and is distributed fairly uniformly throughout the substrate. The non-granular nature of ALD films allows the uniform deposition of alumina on nanoparticles. A Nicolet 750 Magna-IR Fourier transform infrared (FTIR) spectrometer was used to determine the composition of the deposited films. The vibrational mode of bulk alumina is located at about 930 cm−1 . As observed in figure 7, no features appear at this frequency for the uncoated zirconia sample. Al2 O3 bulk vibrational modes appear for coated particles after 50 and 70 cycles. The frequency of the vibrational modes for the coated particles matches the one of A-16 SG® alumina powder obtained from Almatis. This is a direct confirmation of the composition of the alumina films. The alumina vibrational
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
Figure 5. (a) and (b). Transmission electron micrographs of alumina-coated zirconia nanoparticles after 50 coating cycles.
a
spectrum appear. This behaviour is expected for conformally and uniformly coated particles [29, 30].
b
3.3. Effect of coating on particle size distribution and surface area
Figure 6. (a), (b) FESEM images and aluminium-EDS signal for alumina-coated zirconia nanoparticles.
modes for coated particles completely match the one for the reference alumina powder and no features from the zirconia
3.0
Bulk Alumina Feature 2.5
2.0
1.5
Infrared Absorbance
After 70 cycles After 50 cycles Alumina powder Uncoated Zirconia
As has been studied by several researchers [34–39], nanoparticles fluidize as dynamic aggregates of several hundreds of microns in size. If aggregates, rather than individual nanoparticles, were coated, this large size would be shown in a measurement of the aggregate size distribution. A 3225 Aerosizer® from TSI was used to determine the size distribution of dynamic aggregates of zirconia nanoparticles before and after coating (figure 8). This system only detects particles in the micron-sized range. Therefore, individual nanoparticles do not appear in this distribution. As observed in the plot, the size of dynamic aggregates remains fairly unchanged after the coating process, meaning that no aggregates are being coated. These are dynamic aggregates of coated particles and not coated aggregates. If aggregates of nanoparticles were coated and glued together, the size distribution after coating would drastically shift to the
1.0
1200
1100
1000
900
800
700
600
0.5 500
-1
Frequency (cm ) Figure 7. FTIR spectra of uncoated, reference alumina powder and alumina-coated zirconia nanoparticles after 50 and 70 cycles.
S379
L F Hakim et al
1.0
Uncoated Coated (50 cycles)
Normalized Frequency
0.8
0.6
0.4
0.2
0.0 0
1
2
3
4
5
6
7
8
9
10
Size (microns) Figure 8. Particle size distribution of uncoated and alumina-coated zirconia nanoparticles.
50.0 44.4
Surface Area (m2/g)
40.0 34.3
30.0 23.1 20.8
20.0 11.5
11.2
10.0
0
2
4
6
8
10
Film Thickness (nm) Figure 9. Prediction of change in surface area with increasing film thickness for different primary particle sizes. 26 nm ( ), 50 nm (), 100 nm ().
right. Measured nanoparticles remain aggregated only due to cohesive London–van der Waals forces [40, 41]. An additional method to verify that nanoparticles are coated individually is the determination of the surface area. If nanoparticles were sintered together during the coating process, the surface area would drastically decrease. Nevertheless, there is an expected change in the surface area of nanoparticles as the thickness of the film increases. This change occurs due to the increase in the particle diameter as well as the change in the effective density. Figure 9 shows the prediction in the change of surface area for three different sizes of zirconia nanoparticles as the thickness of the alumina film increases. The surface area of zirconia nanoparticles was measured before and after 70 coating cycles using a Quantachrome Autosorb® -1. The experimental surface areas for uncoated and coated samples are 38.1 ± 4.0 and 33.2 ± 7.9 respectively. The prediction mentioned above shows an expected change of S380
14.8% in the surface area after 70 cycles. The experimental change is 12.9%, lower than the predicted value, showing that nanoparticles are not being glued together during the coating process. It is important to mention that the prediction shown in figure 9 assumes spherical particles. However, as observed in TEM micrographs shown in figure 2, zirconia nanoparticles are not perfectly spherical due to sintering that occurs to some extent during the manufacturing process. Deviations from the predicted change in surface area can be attributed to this precoating condition. 3.4. Conformal coating of primary nanoparticles Probably the biggest challenge in trying to individually coat primary nanoparticles is to overcome their natural tendency for agglomeration. Due to cohesive London–van der Waals forces [41] nanoparticles aggregate into large structures that
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
can achieve several hundreds of microns in size. These aggregates are held together by a combination of interparticle forces [34, 37]. Processing ALD coating of nanoparticles in a fluidized bed offers an advantage over other reactor configurations. When interparticle forces are minimized a behaviour called dynamic aggregation is observed during fluidization of nanoparticles [34]. Dynamic aggregates partially break apart and reform due to constant solid recirculation and gas flow through the bed of particles. Even though this breakage is partial during a single collision event, due to recirculation and frequent impacts, the entire surface of all the particles will be eventually exposed. If the same gas atmosphere is kept during this time (the time it takes to expose the entire surface of all particles) the surface will be saturated and individual nanoparticles can be coated. Two competing mechanisms exist during the adsorption of gas precursors into the surface of nanoparticles. On one hand, dynamic aggregation allows for exposing the surface of the particle if processed for long enough times. On the other hand, diffusion of reactants inside aggregates can reach the surface of individual particles despite the fact that they are aggregated. However, the conduits formed between nanoparticles are extremely narrow and the diffusion mechanism would be very slow [42–44]. This means that surface adsorption of gas precursors will predominantly take place during aggregate breakage. This mechanism also assures that nanoparticles are not glued together due to coating as the adsorption and reaction of gas precursors occur only when nanoparticles are not aggregated. This implies that in the absence of dynamic aggregation coating of primary nanoparticles may not be feasible.
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]
4. Conclusions
[30]
Primary zirconia nanoparticles were conformally coated with alumina ultrathin films using atomic layer deposition (ALD) in a fluidized bed reactor. The composition of alumina nanolayers was confirmed by FTIR spectroscopy. Highly conformal and uniform films are observed via TEM and high-resolution FESEM-EDS analysis. The particle size distribution and surface area of nanoparticles are not affected by the coating process. Primary nanoparticles are coated individually due to their dynamic aggregation behaviour.
[31]
[34] [35]
Acknowledgments
[36]
This work was funded by the National Science Foundation (grant NER-0210670), the Department of Education GAANN Program in Functional Materials and the University of Colorado Engineering Excellence Fund. The authors thank Tom Giddings for his support in the TEM analysis. The authors would also like to thank Geoff Courtin and Dr William Kroenke at the Center for MicroEngineered Materials at the University of New Mexico for performing the FESEM-EDS studies on coated nanoparticles.
[32] [33]
[37] [38] [39] [40] [41] [42] [43] [44]
Buijnsters J G et al 2002 Diamond Relat. Mater. 11 1760–8 Hong B et al 1997 Diamond Relat. Mater. 6 55–80 Kellerman B K et al 1997 Surf. Sci. 375 331–9 Laikhtman A et al 1999 Diamond Relat. Mater. 8 725–31 Takatsuka T et al 2000 Appl. Surf. Sci. 162/163 156–60 Itoh H, Watanabe N and Naka S 1988 J. Mater. Sci. 23 43–7 Itoh H, Hattori K and Naka S 1989 J. Mater. Sci. 24 3643–6 Toda Y and Kato A 1989 Ceram. Int. 15 161–6 Sanjurjo A, Lau K and Wood B 1992 Surf. Coat. Technol. 54 219–23 Wood B J et al 1991 Surf. Coat. Technol. 49 228–32 Lau K H, Sanjurjo A and Wood B J 1992 Surf. Coat. Technol. 54 234–40 Karches M, Morstein M and Rudolf von Rohr P 2003 Surf. Coat. Technol. 169/170 544–8 Kim J, Han G Y and Chung C-H 2002 Thin Solid Films 409 58–65 Sanchez I et al 2001 Powder Technol. 120 134–40 Shin H S and Goodwin D G 1994 Mater. Lett. 19 119–22 George S M, Ott A W and Klaus J W 1996 J. Phys. Chem. 100 13121–31 Goodman C and Pessa M 1986 J. Appl. Phys. 60 R65–R81 Suntola T 1992 Thin Solid Films 216 84–9 Ott A W et al 1997 Thin Solid Films 292 135–44 Ott A W et al 1996 Appl. Surf. Sci. 107 128–36 Dillon A C et al 1995 Surf. Sci. 322 230–42 Higashi G S and Fleming C G 1989 Appl. Phys. Lett. 55 1963–5 Fan J F, Sugioka K and Toyoda K 1991 Japan. J. Appl. Phys. 2 30 1139–41 Kumagai H and Toyoda K 1994 Appl. Surf. Sci. 82/83 481–6 Ritala M et al 1996 Thin Solid Films 286 54–8 Soto C and Tysoe W T 1991 J. Vac. Sci. Technol. A 9 2686–95 Ferguson J D et al 2004 Appl. Surf. Sci. 226 393–404 Ferguson J D, Weimer A W and George S M 2002 Thin Solid Films 413 16–25 Ferguson J D, Weimer A W and George S M 2000 Thin Solid Films 371 95–104 Ferguson J D, Weimer A W and George S M 2000 Appl. Surf. Sci. 162/163 280–92 Wank J R, George S M and Weimer A W 2004 J. Am. Ceram. Soc. 87 762–5 Wank J R, George S M and Weimer A W 2004 Powder Technol. 142 59–69 Wank J R et al 2004 Coating fine iron particles with an oxidation-resistance alumina nanolayer using AlD in a fluidized bed reactor Fluidization XI, Present and Future of Fluidization Engineering (Brooklyn, NY: ECI International) Hakim L F et al 2005 Powder Technol. submitted Nam C H, Pfeffer R and Dave R N 2004 Fluidization XI 11th Int. Conf. on Fluidization (Naples: Engineering Conferences International) Wank J R, Weimer A W and George S M 2001 Powder Technol. 121 195–204 Yao W et al 2002 Powder Technol. 124 152–9 Zhu C et al 2004 Powder Technol. 141 119–23 Matsuda S, Hatano H and Tsutsumi A 2001 Chem. Eng. J. 82 183–8 Hamaker H C 1937 Physica 4 1058–72 Visser J 1989 Powder Technol. 58 1–10 Norberg P, Patersson L-G and Lundstrom I 1994 Vacuum 45 139–44 Philip J R 1995 Chem. Eng. Sci. 50 793–802 Shiundu P M, Williams P S and Giddings J C 2003 J. Colloid Interface Sci. 266 366–76
S381
NANOTECHNOLOGY
Nanotechnology 16 (2005) S375–S381
doi:10.1088/0957-4484/16/7/010
Conformal nanocoating of zirconia nanoparticles by atomic layer deposition in a fluidized bed reactor Luis F Hakim1 , Steven M George2 and Alan W Weimer1,3 1
Department of Chemical and Biological Engineering, University of Colorado, Boulder, CO 80309, USA 2 Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA E-mail: [email protected]
Received 11 January 2005, in final form 24 March 2005 Published 15 April 2005 Online at stacks.iop.org/Nano/16/S375 Abstract Primary zirconia nanoparticles were conformally coated with alumina ultrathin films using atomic layer deposition (ALD) in a fluidized bed reactor. Alternating doses of trimethylaluminium and water vapour were performed to deposit Al2 O3 nanolayers on the surface of 26 nm zirconia nanoparticles. Transmission Fourier transform infrared spectroscopy was performed ex situ. Bulk Al2 O3 vibrational modes were observed for coated particles after 50 and 70 cycles. Coated nanoparticles were also examined with transmission electron microscopy, high-resolution field emission scanning electron microscopy and energy dispersive spectroscopy. Analysis revealed highly conformal and uniform alumina nanofilms throughout the surface of zirconia nanoparticles. The particle size distribution and surface area of the nanoparticles are not affected by the coating process. Primary nanoparticles are coated individually despite their high aggregation tendency during fluidization. The dynamic aggregation behaviour of zirconia nanoparticles in the fluidized bed plays a key role in the individual coating of nanoparticles. (Some figures in this article are in colour only in the electronic version)
1. Introduction The interest in the use of nanoparticles, those with dimensions less than 100 nm, has continuously increased over recent years. The processing of nanoparticles offers the potential to design novel materials at the nanoscale. If the surface properties of nanoparticles could be modified while keeping their bulk properties, the performance of nanopowders for many applications could be greatly enhanced. The deposition of an extremely thin film on a substrate can alter the surface characteristics without degrading the bulk properties of the underlying material. Coating techniques such as chemical vapour deposition (CVD) have been utilized to coat particles. During a CVD reaction the chemical reactants are allowed to coincide 3 Author to whom any correspondence should be addressed.
in the gas phase making this technology dependent upon reaction time, exposure and temperature [1–5]. The nature of CVD can lead to films that have a thickness of 1 µm or more [6, 7], are non-conformal and highly granular [8] or are deposited around aggregates of particles rather than on primary particles [9–11]. Various CVD technologies such as metal– organic chemical vapour deposition (MOCVD) and plasmaenhanced chemical vapour deposition (PECVD) have also been applied to particles [12–15]. However, granular coatings are commonly observed [12] and uniform performance has not been achieved due to particle agglomeration during processing [13]. Atomic layer deposition (ALD) provides a unique method for depositing ultrathin films on surfaces [16–18]. ALD uses sequential surface reactions to coat substrates with high conformality and precise thickness control at the atomic scale.
0957-4484/05/070375+07$30.00 © 2005 IOP Publishing Ltd Printed in the UK
S375
L F Hakim et al
1
To pump
2
3
1 8
To N2 supply
4
6 5
7
7
Figure 1. Schematic diagram of ALD-FBR apparatus: (1) pressure transducers, (2) sintered-metal filter, (3) reaction column, (4) vibro-motors, (5) spring supports, (6) pneumatic valve, (7) reactant containers, (8) mass flow controller.
Coating schemes have been developed for Al2 O3 ALD using a sequence of surface reactions [19–26]. Using doses of trimethylaluminium (TMA) and water in an ABAB. . . sequence alumina films can be deposited. The growth of ultrathin and conformal Al2 O3 nanolayers using sequential surface reactions is based on the binary reaction: 2Al(CH3 )3 + 3H2 O → Al2 O3 + 6CH4
(1)
Al2 O3 ALD is achieved by splitting this binary reaction into two separate half-reactions [19–22]:
2. Experimental details
(A) AlOH∗ + Al(CH3 )3 → AlOAl(CH3 )∗2 + CH4
(2)
(B) AlCH3∗ + H2 O → AlOH∗ + CH4
(3)
where ∗ indicates a surface species. During each dose, the reactants completely saturate the active sites on the substrate, making this a self-limiting and self-controlling process. The thickness of the film deposited after sequential exposures of Al(CH3 )3 and H2 O is nominally 1.1 Å per coating cycle at 450 K [19]. The growth of ultrathin films on nanoparticles has been successfully demonstrated at small scale [27–30] for several ALD systems. In these studies, in situ FTIR spectroscopy was used to monitor the surface chemistry during the sequential exposures. Metal and ceramic micron-sized particles have been coated with alumina using ALD in a fluidized bed S376
reactor [31–33]. Bulk processing of ultrafine particles allowed testing the improved properties of coated particles. Enhanced oxidation resistance and increased surface reactivity are some examples of these superior characteristics. As the interest in the use of nanoparticles continuously grows the need for bulk processing also rises. The present work demonstrates that primary zirconia nanoparticles can be conformally and uniformly coated at large scale with Al2 O3 using ALD in a fluidized bed reactor.
A schematic diagram of the atomic layer deposition fluidized bed reactor (ALD-FBR) is shown in figure 1. The ALD reactor is a 4.6 cm inside diameter (ID) stainless steel column with a 20 µm pore size porous metal disc as the gas distributor. A 316 L porous metal filter element (1.9 cm ID × 15.24 cm long; 0.5 µm pore size) was used at the inside top of the reactor column to prevent particles from leaving the system. Due to their small size, zirconia nanoparticles fluidize as aggregates and remain in the bed during the coating process. The fluidization system was capable of operating at reduced pressures with mechanical vibration. Reduced pressure was achieved in the fluidization column by controlling the opening of a Nupro® LD series diaphragm valve placed at the inlet of an Alcatel 2063 vacuum pump. Vibration was
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
Figure 2. (a) and (b). Transmission electron micrographs of uncoated zirconia nanoparticles.
applied to the system using two synchronized vibro-motors from Martin Engineering having variable amplitude. Vibration frequency was controlled using an ACS 140 speed controller from ABB Drive and Power Products. The vibro-motors were attached to a custom-built vibration platform. Four spring supports on this platform gave homogeneous vertical vibration to the entire system. The fluidization column was attached to this platform via a mounting assembly. Details are shown in figure 1. The fluidized system was controlled and monitored using LabView® . A script recorded the pressure drop across the bed of powder and allowed control of the fluidizing gas flowrate and all the system valves. High-purity nitrogen was used as the fluidizing gas. Its flowrate was controlled by an MKS® 1179 series mass flow controller. MKS® 902 series piezo transducers were located below the distributor plate and at the outlet of the fluidization column to measure the pressure drop across the bed of powder. During operation the system was vibrated to improve the fluidization of nanoparticles. A vibration frequency of 20 Hz and amplitude of 3.3 mm were used. A vacuum trap with an activated alumina insert from Laco Technologies was located at the inlet of the pump to prevent damage to the vacuum system. Particles were obtained from Nanoproducts Corp. They were 26 ± 3 nm in primary particle size and had a BET surface area of 40.9 ± 1.9 m2 g−1 . The bulk density of the powder was 320 kg m−3 and the density of primary nanoparticles was 5890 kg m−3 . Zirconia nanoparticles were not perfectly spherical and showed high agglomeration tendency as observed in figures 2(a) and (b). Some nanoparticles may show sintered regions to some extent. The resulting structures are also nanosized. This condition is not the result of the fluidized bed ALD processing. In order to perform the alumina-ALD reactions, TMA obtained from Sigma Aldrich and deionized water were used as reactants. A series of pneumatically activated SwagelokTM diaphragm valves controlled the automatic and sequential dosing of reactants during the coating cycles. The reactants were fed via their vapour pressures and the system was kept at low pressure (∼1 Torr) at all times. The entire reactor was encased by a clamshell-type furnace from Marshall Furnace Co. The reaction temperature was 450 K. The feeding lines were also heated to approximately 350 K using heavy-duty heating tape. This helped to elevate the temperature of gases before reaction as well as reducing the deposition of gaseous species on the internal walls of the feeding lines.
Prior to loading, powders were sieved using a 40 mesh screen. This screening was intended to eliminate large stationary aggregates (>420 µm) formed during shipping and storage. Particles were then dried at 400 K and 0.5 Torr for 2 h inside the reactor and under a continuous flow of nitrogen. The pressure inside the reactor was monitored at all times. As observed in figure 3, the half-reactions during a coating cycle can be easily identified by the characteristic dosing pressures of TMA and water. After each reactant dose, the system was flushed with inert gas to eliminate unreacted species as well as any methane formed during reaction. The reactants were fed for enough time so saturation of all active sites occurred for every dose. This time was calculated based upon the surface area of the nanoparticles and the sample mass utilized for each batch. Typical batch volumes for ALD processing of nanoparticles were of the order of 120 cm3 . The coating cycle was automatically repeated until the desired film thickness was achieved.
3. Results and discussion 3.1. Fluidization behaviour of zirconia nanoparticles Since the ALD coating was performed in a fluidized bed reactor, understanding the fluidization of zirconia nanoparticles is a key step in improving the coating process. In general, due to high cohesive forces, nanoparticles fluidize as aggregates [34–38]. This means that properties of aggregates, rather than those of primary particles, will determine the fluidization properties of nanoparticles. Aggregate properties such as size, density and sphericity will determine fluidization properties such as bed expansion and minimum fluidization velocity. A fluidized bed offers the advantage of high heat and mass transfer coefficients due to constant solid recirculation and excellent fluid–solid contacting. In an ALD system, this characteristic would help achieve fast saturation of active sites on the particle surface. Therefore, fluidization needs to be assured at all times during processing. Figure 4 shows a plot of pressure drop across the bed of particles versus superficial gas velocity. This plot is used to determine the minimum fluidization velocity of particles. Incipient fluidization is achieved when the pressure drop remains constant. For the case of zirconia nanoparticles, this value is about 0.035 cm s−1 . This shows that zirconia nanoparticles fluidize despite their small size and high cohesive forces and that this system is suitable for ALD processing. 3.2. Test for conformality, uniformity and composition of films A Philips CM 10 transmission electron microscope (TEM) was used to analyse the conformality of alumina films deposited on zirconia nanoparticles. TEM analysis was performed at 100 kV. Figures 5(a) and (b) show two micrographs of coated nanoparticles after 50 cycles. From visual inspection, the growth rate is calculated to be in the range of 1–1.4 Å/cycle. It is important to mention that the growth rate of films may vary with the size and geometry of the substrate. For particles with a high ratio of curvature, more active sites on the surface are exposed to the gas phase S377
L F Hakim et al
1.4 H2O Dose
1.2 TMA Dose
Pressure (Torr)
1 0.8 0.6
N2 Purge
N2 Purge
0.4 0.2 0 0
50
100
150
200
250
300
350
Time (sec) Figure 3. Pressure profile inside the ALD fluidized-bed reactor during a coating cycle.
0.4
Pressure Drop (kPa)
0.35 0.3 0.25 0.2 0.15 0.1 0.05 0
0.01
0.02
0.03
0.04
0.05
Superficial Gas Velocity (cm/s) Figure 4. Pressure drop across the fluidized bed versus superficial gas velocity for zirconia nanoparticles.
reactants. This is explained because less steric effect occurs during the adsorption step of the reaction, thus higher growth rates are expected. The TEM micrographs show extremely conformal and uniform films around nanoparticles. In figure 5(a), particles appear aggregated due to high cohesive forces commonly observed in nanoparticles. An important observation is the fact that films deposited on different nanoparticles are clearly differentiated. This can be observed on the sharp edge formed between coated nanoparticles. Figure 5(b) shows individually coated primary nanoparticles with a very uniform film throughout their surface. In order to further test the homogeneity of the deposited films, a high-resolution microscopy analysis was performed. A Hitachi S-5200 field emission scanning electron microscope (FESEM) was used. This equipment was coupled with a Princeton Gamma Tech Spirit energy dispersive spectrometer (EDS) with a Li doped Si detector. The EDS analysis was performed at 15 kV. Analyses were performed by Geoff S378
Courtin and Dr William Kroenke at the Center for MicroEngineered Materials at the University of New Mexico. Figures 6(a) and (b) show two sets of FESEM images on the left and their corresponding aluminium-EDS signal on the right. As observed in the images, the aluminium-EDS signal matches the shape of the coated particles and is distributed fairly uniformly throughout the substrate. The non-granular nature of ALD films allows the uniform deposition of alumina on nanoparticles. A Nicolet 750 Magna-IR Fourier transform infrared (FTIR) spectrometer was used to determine the composition of the deposited films. The vibrational mode of bulk alumina is located at about 930 cm−1 . As observed in figure 7, no features appear at this frequency for the uncoated zirconia sample. Al2 O3 bulk vibrational modes appear for coated particles after 50 and 70 cycles. The frequency of the vibrational modes for the coated particles matches the one of A-16 SG® alumina powder obtained from Almatis. This is a direct confirmation of the composition of the alumina films. The alumina vibrational
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
Figure 5. (a) and (b). Transmission electron micrographs of alumina-coated zirconia nanoparticles after 50 coating cycles.
a
spectrum appear. This behaviour is expected for conformally and uniformly coated particles [29, 30].
b
3.3. Effect of coating on particle size distribution and surface area
Figure 6. (a), (b) FESEM images and aluminium-EDS signal for alumina-coated zirconia nanoparticles.
modes for coated particles completely match the one for the reference alumina powder and no features from the zirconia
3.0
Bulk Alumina Feature 2.5
2.0
1.5
Infrared Absorbance
After 70 cycles After 50 cycles Alumina powder Uncoated Zirconia
As has been studied by several researchers [34–39], nanoparticles fluidize as dynamic aggregates of several hundreds of microns in size. If aggregates, rather than individual nanoparticles, were coated, this large size would be shown in a measurement of the aggregate size distribution. A 3225 Aerosizer® from TSI was used to determine the size distribution of dynamic aggregates of zirconia nanoparticles before and after coating (figure 8). This system only detects particles in the micron-sized range. Therefore, individual nanoparticles do not appear in this distribution. As observed in the plot, the size of dynamic aggregates remains fairly unchanged after the coating process, meaning that no aggregates are being coated. These are dynamic aggregates of coated particles and not coated aggregates. If aggregates of nanoparticles were coated and glued together, the size distribution after coating would drastically shift to the
1.0
1200
1100
1000
900
800
700
600
0.5 500
-1
Frequency (cm ) Figure 7. FTIR spectra of uncoated, reference alumina powder and alumina-coated zirconia nanoparticles after 50 and 70 cycles.
S379
L F Hakim et al
1.0
Uncoated Coated (50 cycles)
Normalized Frequency
0.8
0.6
0.4
0.2
0.0 0
1
2
3
4
5
6
7
8
9
10
Size (microns) Figure 8. Particle size distribution of uncoated and alumina-coated zirconia nanoparticles.
50.0 44.4
Surface Area (m2/g)
40.0 34.3
30.0 23.1 20.8
20.0 11.5
11.2
10.0
0
2
4
6
8
10
Film Thickness (nm) Figure 9. Prediction of change in surface area with increasing film thickness for different primary particle sizes. 26 nm ( ), 50 nm (), 100 nm ().
right. Measured nanoparticles remain aggregated only due to cohesive London–van der Waals forces [40, 41]. An additional method to verify that nanoparticles are coated individually is the determination of the surface area. If nanoparticles were sintered together during the coating process, the surface area would drastically decrease. Nevertheless, there is an expected change in the surface area of nanoparticles as the thickness of the film increases. This change occurs due to the increase in the particle diameter as well as the change in the effective density. Figure 9 shows the prediction in the change of surface area for three different sizes of zirconia nanoparticles as the thickness of the alumina film increases. The surface area of zirconia nanoparticles was measured before and after 70 coating cycles using a Quantachrome Autosorb® -1. The experimental surface areas for uncoated and coated samples are 38.1 ± 4.0 and 33.2 ± 7.9 respectively. The prediction mentioned above shows an expected change of S380
14.8% in the surface area after 70 cycles. The experimental change is 12.9%, lower than the predicted value, showing that nanoparticles are not being glued together during the coating process. It is important to mention that the prediction shown in figure 9 assumes spherical particles. However, as observed in TEM micrographs shown in figure 2, zirconia nanoparticles are not perfectly spherical due to sintering that occurs to some extent during the manufacturing process. Deviations from the predicted change in surface area can be attributed to this precoating condition. 3.4. Conformal coating of primary nanoparticles Probably the biggest challenge in trying to individually coat primary nanoparticles is to overcome their natural tendency for agglomeration. Due to cohesive London–van der Waals forces [41] nanoparticles aggregate into large structures that
Conformal nanocoating of zirconia nanoparticles by ALD in a fluidized bed reactor
can achieve several hundreds of microns in size. These aggregates are held together by a combination of interparticle forces [34, 37]. Processing ALD coating of nanoparticles in a fluidized bed offers an advantage over other reactor configurations. When interparticle forces are minimized a behaviour called dynamic aggregation is observed during fluidization of nanoparticles [34]. Dynamic aggregates partially break apart and reform due to constant solid recirculation and gas flow through the bed of particles. Even though this breakage is partial during a single collision event, due to recirculation and frequent impacts, the entire surface of all the particles will be eventually exposed. If the same gas atmosphere is kept during this time (the time it takes to expose the entire surface of all particles) the surface will be saturated and individual nanoparticles can be coated. Two competing mechanisms exist during the adsorption of gas precursors into the surface of nanoparticles. On one hand, dynamic aggregation allows for exposing the surface of the particle if processed for long enough times. On the other hand, diffusion of reactants inside aggregates can reach the surface of individual particles despite the fact that they are aggregated. However, the conduits formed between nanoparticles are extremely narrow and the diffusion mechanism would be very slow [42–44]. This means that surface adsorption of gas precursors will predominantly take place during aggregate breakage. This mechanism also assures that nanoparticles are not glued together due to coating as the adsorption and reaction of gas precursors occur only when nanoparticles are not aggregated. This implies that in the absence of dynamic aggregation coating of primary nanoparticles may not be feasible.
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]
4. Conclusions
[30]
Primary zirconia nanoparticles were conformally coated with alumina ultrathin films using atomic layer deposition (ALD) in a fluidized bed reactor. The composition of alumina nanolayers was confirmed by FTIR spectroscopy. Highly conformal and uniform films are observed via TEM and high-resolution FESEM-EDS analysis. The particle size distribution and surface area of nanoparticles are not affected by the coating process. Primary nanoparticles are coated individually due to their dynamic aggregation behaviour.
[31]
[34] [35]
Acknowledgments
[36]
This work was funded by the National Science Foundation (grant NER-0210670), the Department of Education GAANN Program in Functional Materials and the University of Colorado Engineering Excellence Fund. The authors thank Tom Giddings for his support in the TEM analysis. The authors would also like to thank Geoff Courtin and Dr William Kroenke at the Center for MicroEngineered Materials at the University of New Mexico for performing the FESEM-EDS studies on coated nanoparticles.
[32] [33]
[37] [38] [39] [40] [41] [42] [43] [44]
Buijnsters J G et al 2002 Diamond Relat. Mater. 11 1760–8 Hong B et al 1997 Diamond Relat. Mater. 6 55–80 Kellerman B K et al 1997 Surf. Sci. 375 331–9 Laikhtman A et al 1999 Diamond Relat. Mater. 8 725–31 Takatsuka T et al 2000 Appl. Surf. Sci. 162/163 156–60 Itoh H, Watanabe N and Naka S 1988 J. Mater. Sci. 23 43–7 Itoh H, Hattori K and Naka S 1989 J. Mater. Sci. 24 3643–6 Toda Y and Kato A 1989 Ceram. Int. 15 161–6 Sanjurjo A, Lau K and Wood B 1992 Surf. Coat. Technol. 54 219–23 Wood B J et al 1991 Surf. Coat. Technol. 49 228–32 Lau K H, Sanjurjo A and Wood B J 1992 Surf. Coat. Technol. 54 234–40 Karches M, Morstein M and Rudolf von Rohr P 2003 Surf. Coat. Technol. 169/170 544–8 Kim J, Han G Y and Chung C-H 2002 Thin Solid Films 409 58–65 Sanchez I et al 2001 Powder Technol. 120 134–40 Shin H S and Goodwin D G 1994 Mater. Lett. 19 119–22 George S M, Ott A W and Klaus J W 1996 J. Phys. Chem. 100 13121–31 Goodman C and Pessa M 1986 J. Appl. Phys. 60 R65–R81 Suntola T 1992 Thin Solid Films 216 84–9 Ott A W et al 1997 Thin Solid Films 292 135–44 Ott A W et al 1996 Appl. Surf. Sci. 107 128–36 Dillon A C et al 1995 Surf. Sci. 322 230–42 Higashi G S and Fleming C G 1989 Appl. Phys. Lett. 55 1963–5 Fan J F, Sugioka K and Toyoda K 1991 Japan. J. Appl. Phys. 2 30 1139–41 Kumagai H and Toyoda K 1994 Appl. Surf. Sci. 82/83 481–6 Ritala M et al 1996 Thin Solid Films 286 54–8 Soto C and Tysoe W T 1991 J. Vac. Sci. Technol. A 9 2686–95 Ferguson J D et al 2004 Appl. Surf. Sci. 226 393–404 Ferguson J D, Weimer A W and George S M 2002 Thin Solid Films 413 16–25 Ferguson J D, Weimer A W and George S M 2000 Thin Solid Films 371 95–104 Ferguson J D, Weimer A W and George S M 2000 Appl. Surf. Sci. 162/163 280–92 Wank J R, George S M and Weimer A W 2004 J. Am. Ceram. Soc. 87 762–5 Wank J R, George S M and Weimer A W 2004 Powder Technol. 142 59–69 Wank J R et al 2004 Coating fine iron particles with an oxidation-resistance alumina nanolayer using AlD in a fluidized bed reactor Fluidization XI, Present and Future of Fluidization Engineering (Brooklyn, NY: ECI International) Hakim L F et al 2005 Powder Technol. submitted Nam C H, Pfeffer R and Dave R N 2004 Fluidization XI 11th Int. Conf. on Fluidization (Naples: Engineering Conferences International) Wank J R, Weimer A W and George S M 2001 Powder Technol. 121 195–204 Yao W et al 2002 Powder Technol. 124 152–9 Zhu C et al 2004 Powder Technol. 141 119–23 Matsuda S, Hatano H and Tsutsumi A 2001 Chem. Eng. J. 82 183–8 Hamaker H C 1937 Physica 4 1058–72 Visser J 1989 Powder Technol. 58 1–10 Norberg P, Patersson L-G and Lundstrom I 1994 Vacuum 45 139–44 Philip J R 1995 Chem. Eng. Sci. 50 793–802 Shiundu P M, Williams P S and Giddings J C 2003 J. Colloid Interface Sci. 266 366–76
S381
