Contact mechanics of highly porous oxide nanoparticle agglomerates
- 1.9k Downloads
Efficient nanopowder processing requires knowledge of the powder’s mechanical properties. Due to the large surface area to volume ratio, nanoparticles experience relatively strong attractive interactions, leading to the formation of micron-size porous structures called agglomerates. Significant effort has been directed towards the development of models and experimental procedures to estimate the elasticity of porous objects such as nanoparticle agglomerates; however, none of the existing models has been validated for solid fractions below 0.1. Here, we measure the elasticity of titania (TiO\(_2\), 22 nm), alumina (Al\(_2\)O\(_3\), 8 nm), and silica (SiO\(_2\), 16 nm) nanopowder agglomerates by Atomic Force Microscopy, using a 3.75 \(\upmu\)m glass colloid for the stress–strain experiments. Three sample preparations with varying degree of powder manipulation are assessed. The measured Young’s moduli are in the same order of magnitude as those predicted by the model of Kendall et al., thus validating it for the estimation of the Young’s modulus of structures with porosity above 90 %.
KeywordsYoung’s modulus Porous agglomerates Atomic force microscopy Oxide particles Nanoparticles Particle characterization Instrumentation
The mechanical properties of nanopowders are critical for the optimization of their processing (Iwadate and Horio 1998; Bika et al. 2001; Rong et al. 2004). These properties are crucial in gas-phase processes involving agglomerate collision such as in flame aerosol reactors (Kruis et al. 1998; Madler et al. 2006), lung nanoparticle uptake studies (Limbach et al. 2005), and nanopowder fluidization (Matsuda et al. 2004). The processing method of fluidization, where the powder is suspended in a gas stream moving upwards, is known to enhance fluid to solid contact by powder dispersion (Jung and Gidaspow 2002; van Ommen et al. 2012; Dadkhah et al. 2012; Quevedo et al. 2010; Shabanian et al. 2012), which is beneficial for heat and mass transfer, and widely used in gas–solid reaction, granulation, and particle coating, drying, and mixing, among many other applications. Nonetheless, nanoparticles (NP) fluidize as clusters called agglomerates (Parveen et al. 2013; Zhou and Li 1999; Khadilkar et al. 2014; Liang et al. 2009), making the dynamics within the fluidized bed complex and challenging to study, mainly due to the lack of accurate characterization of the agglomerates (Salameh et al. 2012).
Nanopowders agglomerate in a stepwise fashion (Yao et al. 2002). During synthesis at high temperatures, primary particles form chemical bonds creating chain-like structures called aggregates, reaching sizes of 100s of nm. These aggregates then cluster together by physical interactions forming simple agglomerates with sizes of a few 10s of \(\mu\)m, mainly during powder storage. Finally, the simple agglomerates assemble into complex agglomerates, which can reach sizes of 100s of \(\upmu\)m. As a hierarchical process, each level has structures with particular features such as fractal dimension (de Martin et al. 2014; Hu et al. 2012). This research focuses on the agglomerate properties since these are the structures readily available from stored nanopowder and found during nanopowder processing in the gas phase.
Agglomeration arises from the relatively strong attractive interactions among NPs, which include capillary, solvation, van der Waals, and electrostatic forces (Laube et al. 2015; van Ommen et al. 2012; Castellanos 2005; Hakim et al. 2005; Seville et al. 2000; van Ommen 2012; Quevedo and Pfeffer 2010; Yan et al. 2010). Electrostatic forces diminish in the presence of humidity. In earlier studies, it was shown that physisorbed water molecules situate between the nanoparticles creating an attractive interaction described by a combination of capillary and solvation forces, which can surpass the van der Waals contribution (Salameh et al. 2012; Laube et al. 2015). During nanopowder processing, attractive forces are challenged by external separation forces such as collision (Matsuda et al. 2004). In order to estimate the magnitude of the forces acting on the agglomerates, and thus their morphological stability at the given conditions, their Young’s modulus must be known. The high porosity (>90 %) and rather large size of these structures (\(\approx\) \(100\) \(\upmu\)m) make them extremely fragile. Therefore, stress measurements to study their mechanical properties are particularly challenging.
Because of their large void fraction, nanoparticle agglomerates are expected to have a relatively low Young’s modulus. Due to their fragile nature, any type of manipulation can easily compromise the morphological integrity of the agglomerates and reproducibility of the results. Thus, all techniques considered for the study of agglomerates have to be evaluated for the degree of morphological modification and data reproducibility. Additionally, the number of available techniques to study structures in the micron-size scale with nanoscale resolution is very limited. A quite challenging property to measure is elasticity, mainly due to the structural and technical limitations just mentioned. There have been novel techniques to measure elasticity of soft microscopic objects such as microcapillaries, relying on pressure-induced deformation of microscopic deformable particles in a dilute suspension (Wyss et al. 2010), the ultrasonic pulse-echo method by measuring the velocity of ultrasonic waves in materials along a known crystal direction for isotropic, millimeter thick samples (Yoshimura et al. 2007; Schwarz et al. 1997), or the compression and indentation techniques such as atomic force microscopy (AFM). Methods that require the samples to be in the liquid phase, specifically oriented, or placed at a set location will considerably affect the original structure of the nanoparticle cluster and hinder result reproducibility.
Agglomerate elasticity has been measured before; however, those agglomerates had a significantly higher solid fraction, well-defined geometry, and/or customized formation process than those of interest in this research. In 1987, Kendall et al. (1987) measured the elasticity of ceramic NP clusters to study the effect of solid fraction, developing a model to estimate the effective Young’s modulus in terms of the volume packing, and particle interface energy, size, and modulus. Nonetheless, the experiments were limited to structures with porosity below 70 % (Kendall et al. 1987). Later on, in 1992, Kendall focused on the elasticity of spray-dried spherical agglomerates of uniformly packed 210-nm zirconia particles (Kendall and Weihs 1992), modeling the steps towards agglomerate fracture and describing the use of a nanoindenter to study agglomerate deformation, again, facing the porosity limitation. In 2001, Bika et al. (2001) presented a summary of studies done on the mechanical properties of wet and dry agglomerates, highlighting their morphological frailty, and the lack of proper measuring techniques and realistic theoretical models to obtain accurate values of the agglomerate mechanical properties. Nonetheless, all the data gathered from the literature and reviewed by Bika et al. is for agglomerates with porosity bellow 75 %.
The elasticity, represented as the Young’s modulus, of porous materials can be predicted from theoretical models found in the literature (Yoshimura et al. 2007). These models consider the agglomerate volume fraction and primary particle Young’s modulus as critical variables to determine the agglomerate Young’s modulus. However, the models of Hasselman (1962), Wang (1984), Martin and Haynes (1971), and Phani and Niyogi (1987) have fitting parameters that rely on elasticity experimental data, and thus not really predicting the value. The models of Yoshimura et al. (2007) and Yoshimura et al. (2007) require previous knowledge of the shear and bulk modulus of the porous structure, and Poisson’s ratio of the NP, which leads to a straightforward calculation of the elasticity. Nonetheless, these values are unknown for nanoparticle agglomerates. Kendall et al. (1987) developed a model with a simple expression that uses the agglomerate solid fraction and NP Young’s modulus, work of adhesion, and diameter to estimate the effective elasticity of the porous agglomerate, which can be obtained from commercial suppliers or literature. However, to the best of our knowledge, none of these models has been experimentally validated for structures with porosity above 90 % such as those seen in nanopowders.
A well-established technique to study the elasticity of soft matter is the AFM, which works by forcing an interaction between a probe and the sample. The versatility of the technique allows for the visualization of topographic characteristics to an atomic level, the quantification of interacting forces between nanosized objects, metal deposition on electroconductive substrates, and the measurement of mechanical properties of soft materials (Pimpang et al. 2013; Vakarelski and Higashitani 2006; Barcons et al. 2012; Sigdel et al. 2013; Salameh et al. 2014; Stiles et al. 2008; Farshchi-Tabrizi et al. 2006; Li and Chen 2014; Salameh et al. 2012; Rong et al. 2004; Webber et al. 2008; Tanabe and Tatsuma 2012). This includes fragile micron/nanosized systems such as muscle cells (Engler et al. 2004) and thin gels (Engler et al. 2004) among many other applications (Picas et al. 2012; Xu et al. 2007; Lin et al. 2007; Zheng and Ya-Pu 2004; Rong et al. 2004; Landolsi et al. 2013; Fotiadis et al. 2002; Rico et al. 2005; Dimitriadis et al. 2002; Salameh et al. 2012). In earlier studies, the AFM equipped with a glass colloid attached to the cantilever was used to measure the Young’s modulus of highly porous NP films (Schopf et al. 2013; Butt et al. 2005). To neglect extra phenomena such as adhesion forces and plasticity, only the approach part of the force curved was fitted to the Hertz model for elasticity estimations. However, these films differ from the fluidized agglomerates on the mechanism of formation, homogeneity, and stability, with porosity still bellow that of the complex nanoparticle agglomerates. This method is widely accepted for materials in the kPa–MPa range such as biological samples (Vinckier and Semenza 1998; Roduit C 2009; Radmacher et al. 1996).
The objective of this work is to present an experimental method to measure the elasticity of nanopowder agglomerates, which typically have a porosity above 90 %. The results are used to validate the applicability of elasticity models for highly porous structures. Three sample preparation approaches are compared to verify the conservation of the structure, and measurement accuracy and reproducibility. To preserve the original morphology of the agglomerate, the technique requiring the least manipulation during sample preparation is used to investigate hydrophilic titania (TiO\(_2\)—P25), alumina (Al\(_2\)O\(_3\)—Alu C), and silica (SiO\(_2\)—A130) nanopowders. The experimental results are compared to theoretical models from the literature, and the Kendall et al. (1987) method was found to give a descent estimation.
Properties of the nanopowders as provided by the manufacturer and obtained from TEM image analysis
\(\rho _p\) (kg/m\(^3\))
22 ± 8
16 ± 6
8 ± 2
Three sample preparation methods were tested, referred to as pressed on glass, double-sided tape, and rough substrate. For the powder pressed on glass, a small amount of the nanopowder is placed on a glass microscope slide over an area of about \(1cm^2\), and pressed using a thick piece of flat glass until leaving a homogeneous layer of powder looking uniform to the naked eye. The force exerted over the film is estimated using a scale reaching \(12 \pm 1\) N, which results in a pressure of \(9 \pm 0.5\) kPa. The double-sided tape method involves the spreading of powder over a transparent double-sided tape (Scotch) attached to a glass slide. Then, the sample is gently shaken to remove any loose powder without blowing or touching, to prevent morphological changes. Similarly, the rough surface preparation starts with the spreading of powder on the rough side of a microscope slide, with a final gentle shake to remove the excess powder. These spreading and shaking steps are repeated a few times to ensure a thick enough powder layer for AFM measurements. Due to the extent of manipulation, the pressed on glass method deliberately modifies the structure of the powder, while the sample on the rough substrate is expected to have an almost unchanged morphology. Comparably, the double-sided tape technique is believed to preserve the original structure of the agglomerates. However, the effect of the glue on the mechanical properties was questionable, and thus evaluated.
All three samples were characterized by SEM imaging. A SEM (Jeol JSM-6010 LA) was used to evaluate the general morphology of the nanopowder film on the smooth glass, rough surface, and double-sided tape. To assess the glue–powder integration, images of the tilted double-sided tape sample were taken and analyzed. The samples were slightly blown to prevent nanopowder contamination of the sample chamber. Additionally, for clear SEM imaging, the samples were coated with gold using the Auto Sputter Coater (JEOL JFC-1300).
The Young’s modulus was calculated using the JPKSPM Data Processing software by fitting the Hertz model (Hertz 1881) to the approach curves. First, the baseline was subtracted from the curves to set the point of cantilever–sample contact at zero and have displacement equal to indentation. Then, the x offset (contact point) was adjusted and the height for cantilever bending, corrected previous to the Young’s modulus estimation using the embedded “determine elasticity from indentation” software function. Retraction curves were not considered for elasticity measurements of the agglomerates; hence, only the approach curves are presented and used for the estimation of the Young’s modulus. Other contact mechanics models such as Johnson–Kendall–Roberts (JKP) (Johnson et al. 1971), Derjaguin–Muller–Toporov (DMT) (Derjaguin et al. 1975), and Maugis–Dugdale (MD) (Maugis 1992), which account for adhesion forces (Lin et al. 2007; Landolsi et al. 2013), were also considered.
Results and discussion
From the TEM pictures, it is evident that nanoparticles are found in clusters. These structures are very porous and expected to be susceptible to changes by external disturbances. Therefore, any powder manipulation and processing will dramatically modify their original morphology. Insufficient analysis and understanding of the handling effect can lead to erroneous conclusions regarding the nature of the nanoparticle clusters.
Besides the preparation method, the sample substrate could also affect the AFM measurements. From SEM images, it was seen that the powder film thickness is considerably larger than the colloid indentation depth (Fig. 3), and since the elasticity of the solid substrates is known to be orders of magnitude higher than that of the porous film, the substrates should not have an effect on the measurements. However, the possibility of glue penetration by capillary into the highly porous structures led to extra evaluation of samples placed on the double-sided tape. These samples were assessed by tilted SEM imaging, where the glue was found to immerse less than 0.5 \(\upmu\)m of the attached agglomerates (Fig. 3b). Thus, the glue, as well as the solid substrates, is expected to have negligible to no influence on the AFM measurements, leaving any measurement discrepancy to the preparation method itself.
Force curve analysis
Elasticity of different materials
The elasticity of the nanopowder depends on the particle packing, size distribution, shape, surrounding conditions, powder processing, and contact forces. Three common oxide nanopowders of different materials and primary particle sizes were studied, namely Al\(_2\)O\(_3\), SiO\(_2\), and TiO\(_2\). The packing density depends on the size distribution, powder processing, and particle shape, affecting the space available for particle rearrangement. Additionally, the interparticle forces are affected by material properties such as the Hamaker coefficient and surface hydrophobicity. Thus, each of the three powders has an elasticity which depends on the unique material properties and particle morphology.
As seen from Fig. 7, the values show a wide distribution. This distribution is typical for AFM measurements. Even in the case of perfectly flat substrates such as mica or silicon, AFM values always show wide distribution based on a different number of molecules coming into contact at each measurement (Butt et al. 2005; Farshchi-Tabrizi et al. 2006). However, in the case of porous agglomerates, the contact scenario shows even more variation between measurements since the sample surface is rather heterogeneous, leading to a wider range of measured values.
Values used for the estimation of the effective Young’s modulus
22 ± 8
8 ± 2
16 ± 6
Kendall’s model can estimate the elasticity of the highly porous sample to the right order of magnitude, which is known to be extremely challenging. The slight discrepancy between the theoretical and experimental Young’s modulus values can be attributed to the partial plasticity of the agglomerates and the rough distribution of powder throughout the film. In all three cases, the experimental value is lower than the theoretical one since plastic deformation based on aggregate rearrangements during agglomerate compression by the colloid is not accounted for in Kendall’s model. This plasticity of the system must be too small (compared to the spring constant of the cantilever) for the experimental equipment and data analysis software to find the elastic Herzt model unsuited. Furthermore, the spread of the measured values also represents the range of agglomerate properties found throughout the film.
The parameters needed to calculate the elasticity for the different sample preparation methods are presumably known for the rough substrate and double-sided tape, and assumed to be the same; however, the porosity of the pressed on glass sample is unknown. A porosity of \(91\pm 5\) % was back-calculated from the Kendall model for TiO\(_2\) pressed on glass sample using the AFM measurements as the effective Young’s modulus (Fig. 6). This means that even after the squeezing step, the pressed powder shows a highly porous morphology, which from the SEM images (Fig. 3c) seems appropriately described by the estimated value.
Other theoretical models to compute the Young’s modulus were considered (Adachi et al. 2006; Wagh et al. 1991; Pabst et al. 2006; Kupkova 1993; Yoshimura et al. 2007; Choren et al. 2013). Nevertheless, some of them include fitting parameters that require experimental data (Hasselman 1962; Wang 1984; Martin and Haynes 1971; Phani and Niyogi 1987; Choren et al. 2013), which defeats the purpose of the analytical calculation for this study, and those from Yoshimura et al. (2007) use as parameters properties of the porous material that are still unknown due to technical limitations similar to those encountered for the Young’s modulus (Kovacik 2001). Alternative models, listed in Choren et al.’s review (Choren et al. 2013), which only depend on agglomerate porosity and Young’s modulus of the nonporous material estimate moduli in the gigapascal (GPa) range (see Fig. S4, Supporting Information), more than four orders of magnitude higher than the experimental values obtained from the AFM.
The research presented in this paper describes a method to experimentally determine the Young’s modulus of structures with porosity higher than 90 %. The focus of the study is on nanoparticle agglomerates, which are a few hundred micrometers in size and very fragile, formed due to strong attractive interactions among the primary particles. The experiments are done by AFM on five different samples including three materials (Al\(_2\)O\(_3\), SiO\(_2\), TiO\(_2\)) using the double-sided tape sample preparation method, and three sample preparation methods (pressed on glass, rough surface, double-sided tape) for one of the nanopowders (TiO\(_2\)). The results validate the applicability of Kendall et al. model to predict the elasticity of nonspherical highly porous structures. A more detailed analysis on the extrapolation of Kendall’s model to low solid fractions and/or irregularly shaped particles will lead to a better understanding of the solid fraction’s effect on the effective elasticity of porous structures. The proposed experimental technique can be used for validation of current or future models.
The research leading to this paper has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013)/ERC Grant, agreement No. 279632.
- Bergstrom L (1997) Hamaker constants of inorganic materials. Adv Colloid Interface Sci 70:125–169. doi: 10.1016/S0001-8686(97)00003-1, URL:http://www.sciencedirect.com/science/article/pii/S0001868697000031
- Bika DG, Gentzler M, Michaels JN (2001) Mechanical properties of agglomerates. Powder Technol 117(12):98–112. doi: 10.1016/S0032-5910(01)00318-7, URL:http://www.sciencedirect.com/science/article/pii/S0032591001003187
- Derjaguin B, Muller V, Toporov Y (1975) Effect of contact deformations on the adhesion of particles. J Colloid Interface Sci 53(2):314–326. doi: 10.1016/0021-9797(75)90018-1, URL:http://www.sciencedirect.com/science/article/pii/0021979775900181
- Fotiadis D, Scheuring S, Muller SA, Engel A, Muller DJ (2002) Imaging and manipulation of biological structures with the AFM. Micron 33(4):385–397. doi: 10.1016/S0968-4328(01)00026-9, URL:http://www.sciencedirect.com/science/article/pii/S0968432801000269
- Hertz H (1881) Ueber die berhrung fester elastischer krper. J Reine Angew Math 92:156–171Google Scholar
- JPK (2009) Determining the elastic modulus of biological samples using atomic force microscopy. JPK Instrum. www.jpk.com
- Kendall K, Weihs TP (1992) Adhesion of nanoparticles within spray dried agglomerates. J Phys D 25(1A):A3. http://stacks.iop.org/0022-3727/25/i=1A/a=002
- Khadilkar A, Rozelle PL, Pisupati SV (2014) Models of agglomerate growth in fluidized bed reactors: critical review, status and applications. Powder Technol 264(0):216–228. doi: 10.1016/j.powtec.2014.04.063, URL:http://www.sciencedirect.com/science/article/pii/S0032591014003817
- Kruis FE, Fissan H, Peled A (1998) Synthesis of nanoparticles in the gas phase for electronic, optical and magnetic applicationsa review. J Aerosol Sci 29(56):511–535. doi: 10.1016/S0021-8502(97)10032-5, URL:http://www.sciencedirect.com/science/article/pii/S0021850297100325
- Mackenzie JK (1950) The elastic constants of a solid containing spherical holes. Proc Phys Soc Sect B 63(1):2–11. http://stacks.iop.org/0370-1301/63/i=1/a=302
- Madler L, Lall AA, Friedlander SK (2006) One-step aerosol synthesis of nanoparticle agglomerate films: simulation of film porosity and thickness. Nanotechnology 17(19):4783. http://stacks.iop.org/0957-4484/17/i=19/a=001
- Maugis D (1992) Adhesion of spheres: The jkr-dmt transition using a dugdale model. J Colloid Interface Sci 150(1):243–269. doi: 10.1016/0021-9797(92)90285-T, URL:http://www.sciencedirect.com/science/article/pii/002197979290285T
- Norman LD (1971) Computer simulation of particulate systems. UNT Digital LibraryGoogle Scholar
- Radmacher M, Fritz M, Kacher C, Cleveland J, Hansma P (1996) Measuring the viscoelastic properties of human platelets with the atomic force microscope. Biophys J 70(1):556–567. doi: 10.1016/S0006-3495(96)79602-9, URL:http://www.sciencedirect.com/science/article/pii/S0006349596796029
- Shabanian J, Jafari R, Chaouki J (2012) Fluidization of ultrafine powders. Int Rev Chem Eng 4:16–50Google Scholar
- van Ommen JR (2012) Manufacturing core-shell nanoparticles by atomic layer deposition in a fluidized bed. In: International symposium on chemical reaction engineeringGoogle Scholar
- Vinckier A, Semenza G (1998) Measuring elasticity of biological materials by atomic force microscopy. FEBS Lett 430(12):12–16. doi: 10.1016/S0014-5793(98)00592-4, URL:http://www.sciencedirect.com/science/article/pii/S0014579398005924
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.