Three-dimensional chemical imaging of embedded nanoparticles using atom probe tomography

Single crystal MgO(100) substrates were cleaned in acetone/methanol in an ultrasonic bath prior to placement in the implantation chamber. Implantation was carried out using 2 MeV Au²⁺ ions using a 3.0 MV tandem accelerator located in the EMSL at the Pacific Northwest National Laboratory (PNNL). The Au²⁺ ions were implanted at 7° to the normal at a dose in the range of 7 × 10¹⁶–1.4 × 10¹⁷ Au²⁺ cm⁻² at a substrate temperature of 975 K and, subsequently, the samples were annealed in a tubular furnace for 10 h at 1275 K followed by furnace cooling. The samples for APT and STEM analysis were prepared using a standard lift-out method in a dual-beam FIB system (FEI Helios Nanolab). Annular milling was used to obtain the APT tips and a low-kV clean up step towards the end was used to ensure minimal Ga impurities, if any. The details of sample preparation are beyond the scope of this paper and the readers are recommended to follow Thompson et al [14]. The tips are mounted on a pre-sharpened silicon micro-tip array, which is further introduced into the LEAP 3000X HR system. This system utilizes a green laser with a femtosecond pulse and the analysis chamber is maintained under UHV conditions. The experiments are typically carried out at  ∼ 50 K specimen temperature and the voltage is increased to continuously achieve a constant evaporation rate of about 0.5%. Initial experiments were carried out to optimize the laser pulse energy to provide artifact free images. The laser pulse energies were systematically varied from 0.3 to 0.8 nJ, at 0.1 nJ increments. Five million data points were collected at each pulse energy and the data is reconstructed to verify features such as matrix stoichiometry, the peak shapes and distribution. It was observed that the data obtained at 0.3 nJ yields stoichiometric MgO and hence could be optimum within the experimental range. The reconstructed data was analyzed to get the composition of the individual Au nanoparticles within the MgO matrix, and the distribution of Au, Mg, and O was studied as a function of particle size. Scanning transmission electron microscopy analysis was carried out at FEI Company, Hillsboro, OR. on a representative atom probe sample and the elemental profiles have been obtained for Mg, Au and O using energy dispersive x-ray spectroscopy (EDX). The size of the nanoparticles from the TEM analysis was compared with that from APT.

The implantation conditions have been optimized previously [15, 16], to ensure in situ nanocluster formation with the help of HRTEM analysis. One of the key observations was the need to carry out implantation at elevated temperature and post-implantation annealing at 1000°C to obtain well-defined nanoparticles, in situ. This was also required to ensure that the radiation damage due to the high-energy implantation was reduced significantly. Wang et al [15] have discussed the formation of quantum anti-dots as a result of vacancy clustering in Au–MgO systems. Diffraction analysis was also carried out to understand the crystallographic relationship between the Au clusters and the MgO matrix; however, the authors have not presented any compositional analysis in these studies. As the chemistry of these clusters often controls the final properties, we have analyzed the Au-implanted MgO samples with APT.

Zoom In Zoom Out Reset image size

A representative 3D reconstructed image, 230 × 220 × 400 nm³, of the Au–MgO sample is shown in figure 2. The iso-concentration surface analysis revealed pronounced clustering behavior for the Au ions, as shown by the red colored particles, whereas the same was not observed in the case of Mg or O species. It is apparent from the figure that the Au clusters were not present in the top 200 nm under the surface of the MgO. Under the given implantation conditions, the Au concentration is supposed to follow a broad Gaussian profile, with a peak around 300–400 nm. Since the tips were made to include the unimplanted region between the surface and implanted region we did not observe any Au evaporation in the initial stages during APT analysis of these tips. The number density of Au particles was found to increase towards the peak Au concentration in the implanted region as did the average cluster size. STEM-EDX maps from a similar, representative, APT tip are presented in figure 3. Both APT and STEM-EDX results clearly demonstrate that the dimensions of the Au nanoparticles are similar at  ∼ 8–12 nm.

Zoom In Zoom Out Reset image size

While STEM-EDX analysis is powerful, the very nature of the TEM hinders the precise chemical analysis of the Au nanoclusters, as a result of shadowing effects through the sample thickness. The complications increase further in attempting to obtain tomographic images of such small features in TEM. However, layer-by-layer evaporation during APT analysis enables the observation of clearly resolved Au nanoparticles under appropriate evaporation conditions. The 2D projection of 3D APT reconstruction is provided in figure 4(a), only 10% Au-iso-concentration surfaces are shown here for reasons of clarity. Each individual iso-concentration surface chemically identifies the nanoparticle and composition variations are analyzed with the help of proximity histograms and 1D concentration profiles. Iso-concentration surfaces are user-defined envelopes corresponding to a specific concentration threshold within the concentration space and hence could be treated, more than just as graphical objects, as real features in the sample for further analysis. The proximity histogram represents the concentration of various species as a function or proximity from a chosen interface and hence is extremely sensitive to interfacial segregation and intermixing at phase boundaries [17]. Often the proximity histograms are obtained for a selected isosurface. The 1D concentration profile takes all ions within a chosen region of interest (ROI) and displays the % composition at a calculated distance in the direction of the axis of the ROI [17, 18]. While the iso-concentration surfaces help us identify the nanoparticles within the matrix, 1D concentration profiles are utilized to analyze the composition of the nanoparticles. 10% Au-iso-concentration surfaces have been outlined to represent the nanoparticles. The 1D concentration profiles of the individual nanoparticles are obtained by choosing a cylindrical ROI (1 × 1 × 25 nm³) passing through the center of each nanoparticle. Selected profiles from nanoparticles with small, medium and larger dimensions are provided in figure 4(b). It should be noted that the sizes of the clusters are identified in 'nm³'(i.e. volume) as not all of them have the same shape. It is immediately evident from the figure that the amount of Au increases with increasing particle size, with a peak Au concentration for the largest particle at  ∼ 50%. It can also be observed from the plots that the concentration of Au gradually increases and hits the peak, resulting in a plateau at  ∼ 50%. It is also clear from the plot that the size of the Au particles increased with increasing Au concentration in the MgO matrix, i.e. the size of the particles also follows a Gaussian, similar to the Au²⁺ dose.

Zoom In Zoom Out Reset image size

With the advent of laser pulsing in atom probe tomography, there has been an increasing interest in analyzing semiconductor and oxide materials [19–21]. While a sizable number of studies are available on semiconducting materials, the reports on oxides are relatively sparse [22]. Mazumder and Tsukada [23, 24] have reported the evaporation characteristics of MgO in thin film and bulk crystal forms. They have suggested a possibility of more than just thermal effects from the laser pulse for the evaporation of atoms during atom probe experiments. It is proposed that the accumulation of holes near the tip apex can significantly reduce the field evaporation barrier. In the present work, we have seen a significant influence of the laser pulse energy on the stoichiometry of MgO. While 0.3 nJ pulse energy resulted in almost stoichiometric MgO (Mg:O—52.36:46.56 at.%), the experiments carried out at higher pulse energies resulted in unusual, complex Mg_xO_y species and their clustering. Hence, most of the analysis is focused on the data set collected at 0.3 nJ. All the Mg isotopes have been clearly identified with a full-width half-maximum mass resolving power of 770 and a full-width tenth-maximum mass resolving power of 350, indicating that the compositions obtained are reliable (mass spectrum not shown here). As the focus of the present study is to understand the ability of APT to analyze embedded nanoparticles, the details of MgO evaporation characteristics are not discussed here.

Embedded nanoparticle analysis is not completely new to the APT community as one of the most widely studied systems using APT is the oxide dispersion strengthened (ODS) steels, where oxide nanoparticles are dispersed in a steel matrix [25]. Researchers have often used APT data to obtain the chemistry of the nanoscale oxide precipitates in the case of ODS steels. One of the major challenges in the case of ODS systems is the ability to precisely determine the composition of the ultra-small oxide particles while still ensuring that there is no influence of the matrix overlapping effects. The presence of matrix elements in these nanoparticles was often ascribed to the overlap of the trajectories of the various evaporating elements from the atom probe specimen. After diligent efforts by different groups, it was noted that some of the trajectory overlap related artifacts may not influence the core of clusters of size more than 4 nm [22]. We utilize this information to explain the observed intermixing of Au, Mg and O in the nanoparticles reported in this manuscript.

As the majority of Au nanoparticles analyzed in the present work are larger than 4 nm, it is logical to accept the composition of the nanoparticles obtained using APT analysis. Further, we have also optimized experimental parameters such as laser pulse energy to obtain a stoichiometric matrix (MgO) composition. However, it is premature to propose an exact field evaporation mechanism without further studies. Although the laser pulse energy of 0.3 nJ suggests mixing of Au and MgO in the particles, the characterization using pulse energies lower than 0.3 nJ could not be established because of the lower limit achievable with the green laser. Studies are currently in progress to carry out both APT analysis using significantly lower laser pulse energies with a UV laser and HRSTEM on the same tip to understand whether there is intermixing of Au and MgO in the particles. It will be interesting to compare the results from the present study, using a green laser, with those obtained from UV-laser pulsing. This will help decipher plausible mechanisms of field evaporation in the case where conducting nanoparticles are present inside a dielectric matrix and also clarify the reasons, if any, for the presence of Mg and O in the nanoparticles. At this point, it is logical to propose that the laser-assisted atom probe tomography analysis has the capability to analyze embedded nanoparticles and provide three-dimensional chemical images with unprecedented spatial resolution and chemical sensitivity. We also emphasize that there exists a need for detailed experiments to establish the right parametric space and to include complementary analytical techniques such as, primarily, HRTEM and STEM.

A portion of the research was performed using EMSL, a National Scientific User Facility sponsored by the Department of Energy's Office of Biological and Environmental Research and located at Pacific Northwest National Laboratory (PNNL). Funding support was provided by the Chemical Imaging Initiative at PNNL. The authors would like to acknowledge help from M Nandasiri and R Sanghavi during the course of this project.