# Accuracy of surface strain measurements from transmission electron microscopy images of nanoparticles

- Jacob Madsen
^{1}Email authorView ORCID ID profile, - Pei Liu
^{2}, - Jakob B. Wagner
^{2}, - Thomas W. Hansen
^{2}and - Jakob Schiøtz
^{1}

**3**:14

https://doi.org/10.1186/s40679-017-0047-0

© The Author(s) 2017

**Received: **8 May 2017

**Accepted: **5 October 2017

**Published: **25 October 2017

The Correction to this article has been published in Advanced Structural and Chemical Imaging 2017 3:16

## Abstract

Strain analysis from high-resolution transmission electron microscopy (HRTEM) images offers a convenient tool for measuring strain in materials at the atomic scale. In this paper we present a theoretical study of the precision and accuracy of surface strain measurements directly from aberration-corrected HRTEM images. We examine the influence of defocus, crystal tilt and noise, and find that absolute errors of at least 1–2% strain should be expected. The model structures include surface relaxations determined using molecular dynamics, and we show that this is important for correctly evaluating the errors introduced by image aberrations.

## Keywords

## Background

The surface lattice strain in nanostructures as a topic of research has gained increased interest in recent years due to its significant impact on many material properties. As an example, surface strain is a possible tunable parameter that can be used to optimize the adsorption energies of surfaces for a particular catalytic reaction [1]. Platinum-based oxygen reduction catalysis is improved by weakening the binding of adsorbed oxygen intermediates by 0.1 eV, this can be achieved by a 2% compressive strain [2]. Strain in nanoparticles can be generated by a variety of sources: particle size, shape, twinning, by the lattice mismatch between metals in multimetallic core–shell nanoparticles or it can be induced by the supporting substrate [3]. Characterizing the influence of these effects requires a technique capable of measuring structural information at atomic resolution.

High-resolution transmission electron microscopy (HRTEM) has become a routine tool for determining the structure of materials at an atomic scale [4]. TEM is particularly attractive due to the ability to map local strain. However, TEM images are the result of a complex diffraction and aberration-limited imaging process, and hence considerable care needs to be shown when extracting quantitative information.

An approach to overcome this is to iteratively compare experimental images with simulations [5, 6]; imaging parameters and model structure of the sample are refined until the simulated and experimental image match. This method has been successfully applied to determine various structures including surfaces. Another solution is to reconstruct the exit wave from a focal series, to eliminate the effect of aberrations [7]. However, the additional complexity added by such methods has limited their use. Instead an often used approach is to obtain the atomic positions directly from the experimental images. The positions of the intensity extrema within the image depend on imaging conditions, orientation and sample thickness, hence they do not necessarily coincide with the atomic positions. However, in the periodic part of a solid, a constant spatial relationship can still be assumed between the image and the atomic positions. This assumption breaks in areas with thickness variations, defects and in particular in the vicinity of surfaces and interfaces [8] and thus a systematic assessment of the accuracy is needed for these cases.

A first investigation to determine the accuracy with which surface strain could be determined was undertaken by Marks [9]. Image simulations were used to compare actual relaxations, in the input structural models, with apparent relaxations, measured from the corresponding simulated images. He found that there was a linear relationship between apparent and real strain, with a constant outward shift of about 5%. He also demonstrated that the true positions of atomic columns at the surface could be determined within 0.2 Å, corresponding to 5% of the lattice parameter of gold. This investigation was done before the invention of the spherical aberration corrector, which today has made it feasible to measure surface relaxations on the order of a few percent.

Newer investigations on the accuracy of strain analysis directly from HRTEM images have focused on interfaces in heterostructures [8, 10–12]. The error in such cases was found to be as low as 0.5% [13, 14]. Using a new technique based on Fourier transforming several overlapping sliding windows, it has been demonstrated that picometric precision and accuracy of interatomic distances can be achieved for measurements inside periodic solids [15]. However, these studies do not investigate surfaces and generally assume a uniform thickness. Moreover, in all these cases the strain distributions were fundamentally 2D, i.e. the atomic columns were mainly displaced in the plane perpendicular to the zone axis. This is different from nanoparticles where the true 3D strain is projected as a 2D image.

The literature has several examples of studies using aberration-corrected microscopy that includes measurements of strain in nanoparticles, and in the vicinity of surfaces, these measurements are often backed by comparison with a simulation that approximates the experimental structure and microscope conditions [16–19]. The general conclusion is that the erroneous surface strain due to imaging aberrations is much smaller in aberration-corrected images than the 5% found by Marks. However, these studies lack a systematic analysis of the sensitivity to experimental variables.

In the present work, we evaluate the accuracy of strain analysis directly from simulations of aberration-corrected HRTEM images focusing on surfaces of nanoparticles. The simulated objects are gold nanoparticles, which in addition to being a topic of research in their own right, provides a model structure that has different exposed surfaces and a linear thickness gradient. We examine the influence of four different effects: defocus, particle size, crystal tilt and noise, and we investigate what accuracy can be expected under which imaging and sample conditions.

## Methods

### Image simulation

#### Model and temperature effects

In this study the ideal crystals were relaxed using molecular dynamics (MD) with an empirical potential. The interactions between the atoms were calculated with the charge-optimized many body (COMB) potential [23]. The potential parameters were fitted with a high priority for surfaces and nanoparticles, and hence reproduce the experimental surface relaxations of gold quite well. For an infinitely extended {111} surface, the potential predicts a 1.2% surface expansion of the top layer, which is close to the experimental value of 1.3% [24]. For {100} surfaces an inward relaxation of 1.1% is predicted. There is no corresponding experimental value; however, the prediction is close to 1.2% [25] and 1.51% [26] calculated with density functional theory.

The effect of finite temperature is included using the frozen phonon approximation [27]. This is a semi-classical model based on the assumption that a single high-energy electron passing through the specimen at about half the speed of light can only probe a single frozen “snapshot” of the vibrating crystal. The image is produced by averaging incoherently over many snapshots where the atoms are slightly displaced from their equilibrium positions. The frozen phonon model has been shown to be numerically equivalent to the full quantum-mechanical treatment of the inelastic phonon scattering process [28]. The snapshots are typically determined using the Einstein approximation; however, we chose to use random steps from a constant temperature MD simulation using Langevin dynamics at 300 K [29]. We only used steps after the initial equilibration and the simulation was run for long enough to properly represent the thermal distribution of the atomic positions. We found that the simulated images are converged when \(\sim 40\) snapshots are included in the averaging.

During a MD simulation the projected atomic positions follow a 2D normal distribution. The standard deviation of this distribution is around 0.05 Å or approximately 2% of the distance between the columns. The standard deviation of the distributions is not identical for all columns. It can be approximately 30–40% larger for some surface and corner atoms (see Additional file 1: Figure S3). We find that the difference between the mean relaxed positions and the mean positions obtained from a thermal average is just a constant thermal expansion of the entire crystal.

#### Diffraction and objective lens aberrations

The exit waves were simulated with the multislice algorithm [30] using the QSTEM code [31]. This code has been interfaced with Python and utilizes the atomic simulation environment [32] for setting up model structures, providing a single environment for building models, simulating and analysing images. The code is publicly available.^{1} We have also made code available for directly recalculating and analysing a selection of the results from this paper.

The electrostatic potential of the sample was generated using the independent atom model with the parametrizations of Rez et al. [33]. The potential was generated on a 3D grid before slicing, allowing for accurate simulations of tilted samples. Aberrations due to the objective lens were included by Fourier space multiplication with the contrast transfer function (CTF). The effect of a finite source size and energy spread (i.e. partial spatial and temporal coherence) was included in the Quasi-coherent approximation where envelopes are applied to the wave function [34]. The imaging process is illustrated in Fig. 1c.

The microscope conditions were modelled after an image aberration-corrected FEI Titan microscope operated at 300 kV. Unless otherwise stated, the third-order spherical aberrations were set to \(C_{\text{s}} = -10\) \(\upmu\)m and all other aberrations except for defocus were set to zero. Other aberrations are generally not negligible in aberration-corrected microscopy; however, we chose to neglect them in order to keep the degrees of freedom limited. We tested the stability of our results to inclusion of additional aberrations, in particular twofold astigmatism on the order of 5–10 nm and 5th-order spherical aberrations on the order of 2.5 mm. While some results change slightly, we found that inclusion of additional aberrations does not change our conclusions in significant ways.

The focal spread was \(\Delta = 2.9\) nm and the convergence angle was set to 15 mrad. The sampling used for the simulations was at least 0.05 Å/pixel, and when needed the large simulated images were downsampled using bilinear interpolation.

#### MTF and thermal magnetic noise

*q*is the spatial frequency and the parameters are taken as \(a=0.58\), \(b=2.5\) \({\AA }\) and \(c=5.9\) \({\AA }\).

#### Finite electron dose

*N*collected by the

*i*th detector pixel is given by

*D*is the dose in electrons per area,

*δ*is the sampling and

*I*

_{ i }is the probability for an electron hitting the

*i*’th pixel. The signal-to-noise ratio of the whole image is given by [41]

### Strain analysis

There are several different approaches for obtaining strain directly from HRTEM images. The methods can broadly be classified into three different types: direct measurement of interatomic distances in real space [43, 44], extraction of the lattice by comparison to a template [45] and analysis in Fourier space [46]. The results of the different approaches are similar inside periodic structures, but can differ in the presence of defects [44]. In this paper the real space method is used, since it has the most straight forward interpretation for surfaces, where the results of Fourier space analysis are very opaque. A comparison between real and Fourier space analysis, using geometric phase analysis (GPA), is provided as supplementary information (see Additional file 1: Figure S4).

The most critical step in the real space approach is to determine the positions of the lattice points. There are several ways of defining these positions. However, the simplest way is to define them as the position of the intensity extrema, assumed to correspond with an atomic column. If the lattice points do not correspond to single intensity peaks, they can instead be found using a cross-correlation of the image with a template motif [8]. The intensity extrema are found at sub-pixel accuracy by fitting a 2D function, usually a polynomial or a Gaussian, to the neighbourhood of each peak and setting the derivatives to zero [44]. It is also possible to define the lattice positions from the centre of mass of the intensity distributions [47]. The methods agree if the intensity distributions are symmetric. However, this is not necessarily the case close to asymmetries in the lattice, such as an interface. A comparison of the two methods of measuring the atomic positions is included as supplementary information. The conclusion is that the methods lead to slightly different errors; however, the magnitude of the errors is essentially the same.

The peak pairs algorithm [44] is the most popular method for finding strain from a set of 2D lattice points from HRTEM images. For the calculation of strain at every lattice point, the peak pairs algorithm uses only two lattice vectors. We have found that an approach using a larger number of lattice vectors is significantly more stable in the presence of noise. For an fcc crystal in the [48] zone axis, this method uses the four nearest and two second nearest neighbours to find the strain at any lattice point in the bulk. Another advantage of this method is that it allows us to determine the strain for lattice points at all surfaces and corners, which is not possible with the standard implementation of the peak pairs algorithm. The routines used for strain analysis, including a rudimentary implementation of GPA, are implemented in Python and made available as open source.^{2}

*r*is the residual term, \(v_i\) and \(w_i\) are vectors containing the ideal and actual lattice vectors, \({\mathbf{A}}\) is the affine transformation and \(\Vert \cdot \Vert\) denotes the Euclidean norm. The orientation and elastic strain matrices can be extracted from \({\mathbf{A}}\) via a left-sided polar decomposition of the deformation gradient

*x*- and

*y*-direction

## Results

### Influence of relaxations and temperature effects

When image simulations are used to estimate errors due to aberrations, it is a common practice to use a model of an unrelaxed crystal, under the assumption that the errors caused by these aberrations are insensitive to the small difference between the unrelaxed and relaxed crystal [16–19]. Our results demonstrate that this assumption is invalid in general.

The origin of the errors is deviations from the constant spatial relationship between the image and the underlying projected potential. The peaks are generally more asymmetric for both the image resulting from a relaxed crystal and from a thermal average of crystals, and these small irregularities in the symmetry of adjacent intensity peaks can cause large measurement errors, as illustrated in Fig. 4. All results in the following sections will be based on simulations where temperature effects are included. We also note the \(\sim 50\)% reduction of the image contrast due to thermal vibrations, making the influence of temperature on the image contrast approximately as important as the MTF and thermal magnetic noise.

### Influence of defocus

A defocus of 8.5 nm results in planar strain errors smaller than 1% everywhere, while a defocus of 12.5 nm causes significant errors at the {100} facets. Due to the sign and location of these errors, they could easily be mistaken for real surface relaxations. The errors generally stay small for columns not at the surface; however, at larger defocus some errors start to appear, generally following the thickness gradient.

For the {100} facet the error is almost zero up to a defocus of 8.5 nm, across all three particle sizes. Meanwhile the error for the {111} facet never becomes smaller than 1% for the 4 nm particle, which is approximately the same magnitude as the actual relaxations. For both facets and all sizes, the errors stay below 2% up to a defocus of \(\sim 11\) nm, where the mean error increases sharply at the {100} facets. The mean error does not increase as drastically for the {111} facets. On the other hand, the standard deviation does increase. This is mainly due to the thickness variation along these facets.

### Influence of tilt

It is unavoidable that the sample will be slightly tilted relative to the ideal zone axis. Figure 8 shows the distribution of errors in the planar strain for increasing tilt, \(\alpha\), around the \(\Omega _1\)-axis. At tilt \(\alpha =1.0^{\circ }\), the errors have changed very little compared to the untilted crystal, though the appearance of the image have changed in the central part of particle, this is due to an effective diminishing of the projected potential, as have been reported elsewhere [49]. The errors stay small up to a tilt \(\alpha =2.0^{\circ }\), but increase sharply in the centre of the nanoparticle between \(\alpha =2.0^{\circ }\) and \(\alpha =3.0^{\circ }\). The error introduced by tilt is very dependent on the height of the atomic columns, since the length of the footprint of the projection of a tilted column increases linearly with its height. Only one direction of tilt is shown; however, the trends are similar for other tilt directions. One other tilt directions is included as Additional file 1: Figure S6.

Figure 9 shows the effects of tilt on the errors in the measured surface relaxations for a defocus \(\Delta f = 8.5\) nm. The tilt has a relatively limited impact on the measured surface relaxations. The mean and standard deviation of error changes by at most 1% over the entire tilt range. The effects of tilt on the strain measurements are very dependent on defocus. For example at a defocus \(\Delta f = 14.5\) nm, the mean surface relaxation error changes by more than 6% at the {100} facets, a plot showing this is shown in Additional file 1: Figure S7.

### Influence of noise

The evolution of the object visibility with respect to the sampling and dose is shown in Fig. 10a. At a dose of \(10^2\) \(e^-/{\AA}^2\) the object is barely visible, while the images are essentially unaffected by noise at \(10^5\) \(e^-/{\AA}^2\).

*i*’th lattice point measured from the

*k*th noisy image and \(\epsilon _{\infty ,i}\) is the corresponding measured strain without noise. Since the automatic polynomial fitting can fail at low doses, extreme outliers have been removed before taking the average. Figure 11 shows the distribution of the MAE across a nanoparticle, there is a fairly large difference between the MAE for different lattice points, varying by a factor of three between the centre of the particle and a corner. The reason for this is mainly that the strain at surfaces is determined on the basis of fewer surrounding lattice points. The strain at corner atoms is determined on the basis of just three neighbours, while the measurements in the centre rely on twice that number of neighbours.

## Conclusion

We looked at the accuracy of surface strain measurements from HRTEM images of nanoparticles. We showed that the practice of using simulations based on ideal sample models to calibrate strain measurements is problematic, since the predicted errors from such simulations do not in general reflect the errors for an identical model that includes relaxations.

In general, the impact of the interaction between tilt, thickness and defocus on the final strain measurement is very complicated. However, we observe that if the defocus is small enough, the errors in the measured surface relaxations due to image aberrations can be kept at less than 2%, even for visually obvious tilts. This is significantly larger than the 0.5% that have been found for strain measurements inside periodic solids [13]. The main reason for the larger error is the asymmetry in the peaks close to surfaces.

In order to obtain measurements with small errors, the defocus should not be chosen solely to maximize contrast, since this will also cause large errors due to aberrations. The choice of defocus has to balance delocalization and contrast; if the defocus is too small the contrast will suffer, while if defocus is too large the image aberrations will be the main source of error.

For a dose of \(10^3\) \(e^-/{\AA}^2\), the optimal defocus for the gold nanoparticles is somewhere around 8.5 nm; at this defocus the errors in the surface relaxations are below 2% and the expected noise error is 1.2% with a standard deviation 0.8%.

## Notes

## Declarations

### Author's contributions

JM did the simulations. The data analysis was done by JM with inputs from the other authors. PL performed the experimental work that inspired this article. The manuscript was written by JM with contributions from the other authors. All authors read and approved the final manuscript.

### Acknowledgements

Not applicable.

### Competing interests

The authors declare that they have no competing interests.

### Availability of data and materials section

The source code and algorithms used in this article are available in the repositories https://github.com/jacobjma/PyQSTEM and https://github.com/jacobjma/structural-template-mapping.

### Consent for publication

We consent for this manuscript to be published under the Creative Commons Attribution 4.0 International License.

### Ethics approval and consent to participate

Not applicable.

### Funding

We gratefully acknowledge funding through Grant 1335-00027B from the Danish Council for Independent Research.

### Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Open Access**This 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.

## Authors’ Affiliations

## References

- Mavrikakis, M., Hammer, B., Nørskov, J.K.: Effect of strain on the reactivity of metal surfaces. Phys. Rev. Lett.
**81**(13), 2819–2822 (1998). doi:10.1103/PhysRevLett.81.2819 View ArticleGoogle Scholar - Hernandez-Fernandez, P., Masini, F., McCarthy, D.N., Strebel, C.E., Friebel, D., Deiana, D., Malacrida, P., Nierhoff, A., Bodin, A., Wise, A.M., Nielsen, J.H., Hansen, T.W., Nilsson, A., Stephens, I.E.L., Chorkendorff, I.: Mass-selected nanoparticles of PtxY as model catalysts for oxygen electroreduction. Nat. Chem.
**6**(8), 732 (2014). doi:10.1038/nchem.2001 Google Scholar - Sneed, B.T., Young, A.P., Tsung, C.-K., Romanov, A.E., Shalkovskii, A.G., Wang, J., Ma, Y., Xia, Y., Mao, S.X., Lee, B., Shevchenko, E.V., Liu, Z., Salmeron, M., Somorjai, G.A., Somorjai, G.A., Yang, P., Stamenkovic, V.R.: Building up strain in colloidal metal nanoparticle catalysts. Nanoscale
**7**(29), 12248–12265 (2015). doi:10.1039/C5NR02529J View ArticleGoogle Scholar - Smith, D.J.: Characterization of nanomaterials using transmission electron microscopy. In: Kirkland, A.I., Hutchison, J. (eds.) Nanocharacterization, 2nd edn, pp. 1–29. Royal Society of Chemistry, London (2015). doi:10.1039/9781782621867-00001 Google Scholar
- Möbus, G., Schweinfest, R., Gemming, T., Wagner, T., Ruhle, M.: Iterative structure retrieval techniques in HREM: a comparative study and a modular program package. J. Microsc.
**190**(1–2), 109–130 (1998). doi:10.1046/j.1365-2818.1998.3120865.x View ArticleGoogle Scholar - Van Aert, S., De Backer, A., Martinez, G.T., den Dekker, A.J., Van Dyck, D., Bals, S., Van Tendeloo, G.: Advanced electron crystallography through model-based imaging. IUCrJ
**3**(Pt 1), 71–83 (2016). doi:10.1107/S2052252515019727 View ArticleGoogle Scholar - Bals, S., Aert, S.V., Tendeloo, G.V., Ávila-Brande, D.: Statistical estimation of atomic positions from exit wave reconstruction with a precision in the picometer range. Phys. Rev. Lett.
**96**(9), 096106 (2006). doi:10.1103/PhysRevLett.96.096106 View ArticleGoogle Scholar - Du, K., Phillipp, F.: On the accuracy of lattice-distortion analysis directly from high-resolution transmission electron micrographs. J. Microsc.
**221**(1), 63–71 (2006). doi:10.1111/j.1365-2818.2006.01536.x View ArticleGoogle Scholar - Marks, L.D.: Direct atomic imaging of solid surfaces - 1. Image simulations and interpretation. Surf. Sci.
**139**(1), 281–298 (1984)View ArticleGoogle Scholar - Guerrero, E., Galindo, P., Yáñez, A., Ben, T., Molina, S.I.: Error quantification in strain mapping methods. Microsc. Microanal.
**13**(05), 320–328 (2007). doi:10.1017/S1431927607070407 View ArticleGoogle Scholar - Tillmann, K., Lentzen, M., Rosenfeld, R., Peralta, P., McCartney, M.R., Drucker, J., Picraux, S.T.: Impact of column bending in high-resolution transmission electron microscopy on the strain evaluation of GaAs/InAs/GaAs heterostructures. Ultramicroscopy
**83**(1–2), 111–128 (2000). doi:10.1016/S0304-3991(99)00175-8 View ArticleGoogle Scholar - Rosenauer, A., Gerthsen, D., Potin, V.: Strain state analysis of InGaN/GaN—sources of error and optimized imaging conditions. Phys. Status Solidi
**203**(1), 176–184 (2006). doi:10.1002/pssa.200563519 View ArticleGoogle Scholar - Chung, J., Rabenberg, L.: Effects of strain gradients on strain measurements using geometrical phase analysis in the transmission electron microscope. Ultramicroscopy
**108**(12), 1595–1602 (2008). doi:10.1016/j.ultramic.2008.05.010 View ArticleGoogle Scholar - Hÿtch, M.J., Plamann, T.: Imaging conditions for reliable measurement of displacement and strain in high-resolution electron microscopy. Ultramicroscopy
**87**(4), 199–212 (2001). doi:10.1016/S0304-3991(00)00099-1 View ArticleGoogle Scholar - Klinger, M., Polívka, L., Jäger, A., Tyunina, M.: Quantitative analysis of structural inhomogeneity in nanomaterials using transmission electron microscopy. J. Appl. Crystallogr.
**49**(3), 762–770 (2016). doi:10.1107/S1600576716003800 View ArticleGoogle Scholar - José Yacamán, M., Ascencio, J.A., Canizal, G.: Observation of surface relaxation surface steps and surface reconstruction in gold nanorods. Surf. Sci.
**486**(1), L449–L453 (2001). doi:10.1016/S0039-6028(01)01048-2 View ArticleGoogle Scholar - Fujita, T., Guan, P., McKenna, K., Lang, X., Hirata, A., Zhang, L., Tokunaga, T., Arai, S., Yamamoto, Y., Tanaka, N., Ishikawa, Y., Asao, N., Yamamoto, Y., Erlebacher, J., Chen, M.: Atomic origins of the high catalytic activity of nanoporous gold. Nat. Mater.
**11**(9), 775–780 (2012). doi:10.1038/nmat3391 View ArticleGoogle Scholar - López-Haro, M., Yoshida, K., del Río, E., Pérez-Omil, J.A., Boyes, E.D., Trasobares, S., Zuo, J.M., Gai, P.L., Calvino, J.J.: Strain field in ultrasmall gold nanoparticles supported on cerium-based mixed oxides. Key influence of the support redox state. Langmuir
**32**(17), 4313–4322 (2016)View ArticleGoogle Scholar - Du, K., Ernst, F., Pelsozy, M.C., Barthel, J., Tillmann, K.: Expansion of interatomic distances in platinum catalyst nanoparticles. Acta Mater.
**58**(3), 836–845 (2010). doi:10.1016/j.actamat.2009.09.061 View ArticleGoogle Scholar - Martienssen, W., Warlimont, H.: Springer Handbook of Condensed Matter and Materials Data, 1st edn, p. 1119. Springer, Berlin (2005)View ArticleGoogle Scholar
- Nichols, R.J., Nouar, T., Lucas, C.A., Haiss, W., Hofer, W.A.: Surface relaxation and surface stress of Au(111). Surf. Sci.
**513**(2), 263–271 (2002). doi:10.1016/S0039-6028(02)01510-8 View ArticleGoogle Scholar - Walsh, M.J., Yoshida, K., Kuwabara, A., Pay, M.L., Gai, P.L., Boyes, E.D.: On the structural origin of the catalytic properties of inherently strained ultrasmall decahedral gold nanoparticles. Nano Lett.
**12**(4), 2027–2031 (2012). doi:10.1021/nl300067q View ArticleGoogle Scholar - Liang, T., Shan, T.-R., Cheng, Y.-T., Devine, B.D., Noordhoek, M., Li, Y., Lu, Z., Phillpot, S.R., Sinnott, S.B.: Classical atomistic simulations of surfaces and heterogeneous interfaces with the charge-optimized many body (COMB) potentials. Mater. Sci. Eng. R Rep.
**74**(9), 255–279 (2013). doi:10.1016/j.mser.2013.07.001 View ArticleGoogle Scholar - Vook, R.W., Ouyang, S., Otooni, M.A.: X-ray measurement of the (111) surface relaxation in gold. Surf. Sci.
**29**(1), 277–290 (1972). doi:10.1016/0039-6028(72)90084-2 View ArticleGoogle Scholar - Yu, B.D., Scheffler, M.: Physical origin of exchange diffusion on fcc(100) metal surfaces. Phys. Rev. B
**56**(24), 15569–15572 (1997). doi:10.1103/PhysRevB.56.R15569 View ArticleGoogle Scholar - Singh-Miller, N.E., Marzari, N.: Surface energies, work functions, and surface relaxations of low-index metallic surfaces from first principles. Phys. Rev. B
**80**(23), 235407 (2009). doi:10.1103/PhysRevB.80.235407 View ArticleGoogle Scholar - Loane, R.F., Xu, P., Silcox, J.: IUCr: Thermal vibrations in convergent-beam electron diffraction. Acta Crystallogr. Sect. A Found. Crystallogr.
**47**(3), 267–278 (1991). doi:10.1107/S0108767391000375 View ArticleGoogle Scholar - Van Dyck, D.: Is the frozen phonon model adequate to describe inelastic phonon scattering? Ultramicroscopy
**109**(6), 677–682 (2009). doi:10.1016/j.ultramic.2009.01.001 View ArticleGoogle Scholar - Tadmor, E.B., Miller, R.E.: Modeling Materials : Continuum, Atomistic, and Multiscale Techniques, p. 759. Cambridge University Press, Cambridge (2011)View ArticleGoogle Scholar
- Goodman, P., Moodie, A.F.: Numerical evaluations of N-beam wave functions in electron scattering by the multi-slice method. Acta Crystallogr. Sect. A
**30**(2), 280–290 (1974). doi:10.1107/S056773947400057X View ArticleGoogle Scholar - Koch, C.: Determination of Core Structure Periodicity and Point Defect Density Along Dislocations. Arizona State University, Phd (2002)Google Scholar
- Bahn, S.R., Jacobsen, K.W.: An object-oriented scripting interface to a legacy electronic structure code. Comput. Sci. Eng.
**4**(3), 56–66 (2002). doi:10.1109/5992.998641 View ArticleGoogle Scholar - Rez, D., Rez, P., Grant, I.: IUCr: Dirac-Fock calculations of X-ray scattering factors and contributions to the mean inner potential for electron scattering. Acta Crystallogr. Sect. A Found. Crystallogr.
**50**(4), 481–497 (1994). doi:10.1107/S0108767393013200 View ArticleGoogle Scholar - Kirkland, E.J.: Advanced Computing in Electron Microscopy. Springer, Boston (2010). doi:10.1007/978-1-4419-6533-2 View ArticleGoogle Scholar
- Van den Broek, W., Van Aert, S., Van Dyck, D.: Fully automated measurement of the modulation transfer function of charge-coupled devices above the nyquist frequency. Microsc. Microanal.
**18**(02), 336–342 (2012). doi:10.1017/S1431927611012633 View ArticleGoogle Scholar - Uhlemann, S., Müller, H., Hartel, P., Zach, J., Haider, M.: Thermal magnetic field noise limits resolution in transmission electron microscopy. Phys. Rev. Lett.
**111**(4), 046101 (2013). doi:10.1103/PhysRevLett.111.046101 View ArticleGoogle Scholar - Haider, M., Hartel, P., Müller, H., Uhlemann, S., Zach, J.: Information transfer in a TEM corrected for spherical and chromatic aberration. Microsc. Microanal.
**16**(04), 393–408 (2010). doi:10.1017/S1431927610013498 View ArticleGoogle Scholar - Hÿtch, M.J., Stobbs, W.M.: Quantitative comparison of high resolution TEM images with image simulations. Ultramicroscopy
**53**(3), 191–203 (1994). doi:10.1016/0304-3991(94)90034-5 View ArticleGoogle Scholar - Thust, A.: High-resolution transmission electron microscopy on an absolute contrast scale. Phys. Rev. Lett.
**102**(22), 220801 (2009). doi:10.1103/PhysRevLett.102.220801 View ArticleGoogle Scholar - Lee, Z., Meyer, J.C., Rose, H., Kaiser, U.: Optimum HRTEM image contrast at 20 kV and 80 kV-Exemplified by graphene. Ultramicroscopy
**112**, 39–46 (2011). doi:10.1016/j.ultramic.2011.10.009 View ArticleGoogle Scholar - De Graef, M.: Introduction to Conventional Transmission Electron Microscopy, p. 718. Cambridge University Press, Cambridge (2003)View ArticleGoogle Scholar
- Lee, Z., Rose, H., Lehtinen, O., Biskupek, J., Kaiser, U.: Electron dose dependence of signal-to-noise ratio, atom contrast and resolution in transmission electron microscope images. Ultramicroscopy
**145**, 3–12 (2014). doi:10.1016/j.ultramic.2014.01.010 View ArticleGoogle Scholar - Bierwolf, R., Hohenstein, M., Phillipp, F., Brandt, O., Crook, G.E., Ploog, K.: Direct measurement of local lattice distortions in strained layer structures by HREM. Ultramicroscopy
**49**(1), 273–285 (1993). doi:10.1016/0304-3991(93)90234-O View ArticleGoogle Scholar - Galindo, P.L., Kret, S., Sanchez, A.M., Laval, J.-Y., Yáñez, A., Pizarro, J., Guerrero, E., Ben, T., Molina, S.I.: The Peak Pairs algorithm for strain mapping from HRTEM images. Ultramicroscopy
**107**(12), 1186–93 (2007). doi:10.1016/j.ultramic.2007.01.019 View ArticleGoogle Scholar - Zuo, J.-M., Shah, A.B., Kim, H., Meng, Y., Gao, W., Rouviére, J.-L.: Lattice and strain analysis of atomic resolution Z-contrast images based on template matching. Ultramicroscopy
**136**, 50–60 (2014). doi:10.1016/j.ultramic.2013.07.018 View ArticleGoogle Scholar - Hÿtch, M.J., Snoeck, E., Kilaas, R.: Quantitative measurement of displacement and strain fields from HREM micrographs. Ultramicroscopy
**74**(3), 131–146 (1998). doi:10.1016/S0304-3991(98)00035-7 View ArticleGoogle Scholar - Huang, P.Y., Kurasch, S., Alden, J.S., Shekhawat, A., Alemi, A.A., McEuen, P.L., Sethna, J.P., Kaiser, U., Muller, D.A.: Imaging atomic rearrangements in two-dimensional silica glass: watching silica’s dance. Science
**342**(6155), 224–247 (2013)View ArticleGoogle Scholar - Larsen, P.M., Schmidt, S., Schiøtz, J.: Robust structural identification via polyhedral template matching. Model. Simul. Mater. Sci. Eng.
**24**(5), 055007 (2016). doi:10.1088/0965-0393/24/5/055007 View ArticleGoogle Scholar - Van Dyck, D.: Does crystal tilt enhance the electron interaction? Microsc. Microanal.
**4**(4), 428–434 (1998). doi:10.1017/S1431927698980412 View ArticleGoogle Scholar - Gonzalez, R.C., Woods, R.E.R.E., Eddins, S.L.: Digital Image Processing Using MATLAB®, 2nd edn. Prentice Hall, New Jersey (2009)Google Scholar