TEM Sample Preparation of Nanoparticles in Suspensions
Understanding the Formation of Drying Artefacts
- TEM Sample Preparation of Nanoparticles in Suspensions - Understanding the Formation of Drying Artefacts (© fotoliaxrender - Fotolia.com)
- Fig. 1: Size distribution of nanoparticles in a single state (top) and in an aggregated state (bottom). A and E show a detail of two micrographs containing gold (Au) nanoparticles coated with citrate originating from the same stock. The original dataset consisted of 25 images (from 5 different TEM grids) using a systematic uniform random sampling grid. Figure A and E represent 0.3% of the total recorded area. B and C show the result after binarization and figure C and D show the elliptical modelling of the binarized data by ImageJ. Finally, the size distribution graphs in D and H, based on the modelled ellipses of the entire dataset, show considerable differences. The aggregated state has a mean almost twice the mean determined for the single state data and a very high standard deviation. Also, much nearly 10 times less counts were achieved.
- Fig. 2: Wetting, colloidal stability and microfluidal forces enacting upon nanoparticles in a slowly evaporating droplet. A cartoon of an aqueous droplet evaporating on a hydrophobic substrate is shown on the left: the contact angle remains constant ("Theta" A = "Theta" B = "Theta" C) while the surface area with the substrate is decreasing. Hydrophilic substrates (right) will induce a declining angle ("Theta" A > "Theta" B > "Theta" C) but with a constant surface area. Two capillary forces are active: the coffee ring effect transports colloids outward while the Marangoni flow counteracts this effect. In practice, the Coffee ring effect is superior in aqueous solutions. All capillary forces are independent of the substrate hydrophobicity.
- Fig. 3: Wetting and coffee ring effects in a dried nanoparticle suspension on a TEM grid. Citrate-coated Au nanoparticles suspended in water were dried on a commercially available, Formvar coated, single slot TEM grid. The majority of the nanoparticles were deposited in a ring-like fashion (Coffee-ring effect, quarter ring shown, arrow). Furthermore, the entire drop, including particles, moved top left during the drying (wetting effect), leaving a large portion of the Formvar film empty.
Nanomaterials are of significant economic interest with a global market value of roughly 20 billion €, which is expected to rise to 2 trillion € by 2015 . For commercial use, the European Union provides the following recommended definition: "‘Nanomaterial' means a natural, incidental or manufactured material containing particles, in an unbound state or as an aggregate or as an agglomerate and where, for 50% or more of the particles in the number size distribution, one or more external dimensions is in the size range 1 nm-100 nm" .
This description differs from earlier definitions by the insertion of a number size distribution concept. This number size distribution aspect has a direct impact on the characterization technique of nanoparticles dedicated for consumer goods: transmission electron microscopy (TEM) becomes an attractive method for the characterization of the size distribution of colloidal nanoparticles since it can provide this parameter. Furthermore, another strong argument for its use is the easy and straightforward sample preparation: a small drop of colloidal suspension (usually about 5 µl) is pipetted onto a TEM grid and simply allowed to dry at room temperature. The grid can then be directly observed in a TEM once the medium is evaporated.
Automated Image Analysis
In standard brightfield TEM, the recorded micrographs (fig 1A) will depict darker objects (the particles) on a bright background. The automated image analysis consists of two major steps . First the digital micrograph is converted to a binary image (fig 1B). Second, morphological descriptors are extracted from each particle in the binary image (fig 1C) . An automated measurement following binarization makes a high-throughput characterization of size distribution possible (fig 1D). However, the removal of the liquid medium, the drying step during sample preparation, does not happen without the introduction of artefacts: drying effects cause the accumulation and aggregation of particles (fig 1E). Aggregated nanoparticles produce overlapping projections that cannot be properly separated after binarization (fig 1F) and will create misleading configurations (fig 1G) causing a biased quantification (fig 1H).
Furthermore, aggregation prevents the use of unbiased sampling schemes (adding further systematic bias) and considerably decreases the efficiency and thereby slows down the characterization process and increases costs.
Mechanisms Controlling Drying Effects
The consequences of drying a drop of colloidal suspension for TEM analysis are twofold. First, the drying effects of colloidal suspensions are mainly controlled by colloidal interactions, which impact on the stability of the colloids, i.e. the formation of aggregates. Second, wetting and microfluidics in an evaporating drop of suspension causes inhomogeneous distribution of the colloids on the TEM grid, i.e. reducing sampling efficiency. Therefore, it is essential to understand the underlying forces during drying in order to achieve usable and reproducible TEM samples.
Whether a particle system is stable or not is defined by the net force of attractive and repulsive individual forces between the particles. The main attractive component is the van der Waals interaction. The repulsion in aqueous systems is primarily due to the interaction between the electric double layers surrounding charged particles, which is influenced by the ion concentrations: high ion concentrations weaken the repulsive magnitude whereas low concentrations strengthen the repulsive effect.
The concentration of nanoparticles steadily increases in an evaporating droplet. This increases the probability of particle collisions which, in turn, allows attractive van der Waals interactions to overcome the repulsive forces between particles and result in the aggregation of particles. At the same time, the equally increasing concentration of ions will reduce the effect of the repulsive electric double layer.
The hydrophobicity of the surface influences the droplet morphology during the drying process. On a hydrophobic surface, a constant contact angle (Theta) will be maintained with decreasing contact area (fig. 2, left), while on a hydrophilic surface a constant contact area will be maintained with a decreasing contact angle on a hydrophilic surface (fig. 2, right) . Since TEM grids are usually coated with hydrophobic polymer films, such as Parlodium or Formvar, the droplet will strive for minimizing its contact area with the substrate. The drop, including the suspended colloids will be conveyed onto a decreasingly small fraction of the film. Wetting explains the observation that colloids are not homogenously deposited over the TEM grid but rather converge into a small area, while other parts of the grid remain nearly entirely free of nanoparticles (fig. 3). Therefore, the wetting of colloidal suspensions on an untreated hydrophobic surface impedes the use of systematic random sampling schemes, since these schemes assume a random distribution of events.
Microfluids Coffee Ring Effect
Differential evaporation rates across the drop induce a capillary flow from the center of the droplet towards its edges, resulting in a net transport of dispersed material to the edge (fig. 2, right), which is known as the coffee-ring effect . The deposition of the nanoparticles near the edge of the droplet causes substantial aggregation of a major part of the objects (fig. 3 denoted by the arrow). This effect is, besides wetting, a major cause of the non-uniform particle deposition of nanoparticles in TEM sample preparation. Furthermore, the nanoparticles may undergo a size-based separation, with smaller particles accumulating at the edge of a drop and larger particles at the center  introducing a systematic bias, even if the image processing algorithms can cope with the aggregation effects.
Since surface tension is temperature dependent, a surface tension gradient will follow from the temperature gradient: the surface tension is the highest in the polar (central) region of the drop and is the lowest near the edges (fig. 2, right). This gradient is the source of a recirculating inward directed flow, known as Marangoni flow. In practice, the Marangoni flow is weak in water droplets [6, 8], but is of significant influence in organic solvent systems such as octane . Therefore, in processes using organic solvents (such as the Stöber-Fink-Bohn process for silica particles ) particles will accumulate near the centre of the drop rather than near its edges. Note that the capillary flows (Coffee-ring effect and Marangoni flow) occur independently of the hydrophobicity of the substrate.
The colloidal morphology can influence the deposition in a drying drop. For example, ellipsoids were found to form loosely packed configurations on the air-water interface , producing a surface viscosity that overcomes the bulk viscosity, thereby facilitating ellipsoid resistance to radially outward flows. Spheres deform the air-water interface to considerably lower extend  and experience a much weaker interparticle attraction than ellipsoids .
Automated image analysis routines provide the size distribution of thousands of particles in a matter of seconds. Together with the straightforward TEM sample preparation, these routines open up new possibilities for a low-cost, high-throughput size distribution of nanoparticles as recommended by the European Commission. The prerequisite is that colloidal stability is maintained during the sample preparation. A drop of evaporating medium is a complex, difficult-to-control, non-equilibrium system where a multitude of forces causes non-uniform deposition and aggregation. In the best case, such artifacts reduce the efficiency of the process. Much worse, a systematic bias is introduced and the characterization becomes inaccurate and misleading. Therefore, the possible introduction of artefacts during drying processes of colloidal suspensions needs to be taken into account and the development of optimized protocols that can reduce or even overcome the involved forces needs to be encouraged.
 The European Commission, in Second Regulatory Review on Nanomaterials (2012)
 The European Commission, Official Journal of the European Union, vol. L 275/38 (2011)
 Igathinathane C. et al.: Computers and Electronics in Agriculture (63) 168-182 (2008)
 Schneider C. A. et al.: Nature Methods 9, 671-675 (2012)
 Picknett R. G. and Bexon, R.: Journal of Colloid and Interface Science 61, 336-350 (1977)
 Deegan R. D. et al.: Nature 389, 827-829 (1997)
 Weon B. M. and Je J. H.: Physical Review E 82 (2010)
 Hu H. and Larson R. G.: Journal of Physical Chemistry B 110, 7090-7094 (2006)
 Stober W. et al.: Journal of Colloid and Interface Science 26, 62 (1968)
 Loudet J. C. et al.: Physical Review Letters 94 (2005)
 Loudet J.C. et al.: Physical Review Letters 97 (2006)
 Yunker P.J. et al.: Nature 476, 308-311 (2011)
Dr. Benjamin Michen
Dr. Sandor Balog
Prof. Barbara Rothen-Rutishauser
Prof. Alke Petri-Fink
Dr. Dimitri Vanhecke (corresponding author via e-mail request button below)
Adolphe Merkle Institute
University of Fribourg