The advent of cluster of polyatomic ion beams, including SF₅⁺ and C₆₀⁺, for ToF-SIMS analysis represented a watershed for the analysis of organic materials with ToF-SIMS. Increased secondary ion yields of higher mass molecular ions and more chemically characteristic fragments particularly aided the analysis of chemically complex samples such as biological specimen.

The polyatomic beams also provided a means of overcoming the static limit that had previously applied to the analysis of organic samples. Detection of molecular species beyond the sample surface became possible heralding in the new era of molecular depth profiling and 3D molecular imaging. Gas cluster ion beams have improved these capabilities even further. The energy of the primary ion is now partitioned over several thousand atoms as it impacts the sample surface resulting in “softer” sputtering and a further reduction in sub-surface damage accumulation compared with C₆₀.

So have these new capabilities really changed how we perform SIMS analysis? Are we suddenly able to answer more challenging questions, particularly in bioanalysis, than ever before? Are we living the dream?

This paper will provide a perspective on the impact of 3D molecular imaging for SIMS analysis looking at what has been achieved, what can be done at present and where we might end up in the future.

1. Introduction

There are practical limitations to the use of ion beam sputtering for probing the sample chemistry beyond the surface region which include preferential sputtering and accumulated sputter beam damage. Both effects result in a distortion or complete loss of the true 3D chemical distribution as a function of depth

An alternative approach to achieve 3D chemical imaging of complex matrix chemistries is to utilize in situ FIB milling and sectioning in conjunction with TOF-SIMS chemical imaging, or 3D FIB-TOF tomography [1]. This can minimize or eliminate artifacts caused by sputter depth profiling such as differential sputtering and accumulated ion beam damage.

However, even with FIB polishing there remains some FIB beam-induced chemical or molecular damage that may or may not limit the detection of characteristic molecular signals for organic species. For certain specimens, it is an advantage to follow FIB polishing with cluster ion polishing to recover the characteristic molecular signals.

2. Method

The 3D chemical characterization of pure organic and metal-organic mixed composition structures was achieved utilizing 3D FIB-TOF tomography on a PHI nanoTOF II (Physical Electronics, USA) imaging mass spectrometer. The spectrometer’s large angular acceptance and depth-of-field maintain high mass resolution and high mass scale linearity even in this challenging geometry. This provides the highly desirable ability to perform artifact-free 3D chemical imaging of high aspect ratio features.

3. Results

Samples from two classes of materials were examined for 3D imaging: one metal-organic mixed matrix composition and one mixed organic phase comprised of two polymer moieties. One immediate result of the FIB-TOF imaging was the accurate depth scale determination since there was no preferential sputtering. We have collected characteristic molecular information from each sample for the purposes of 2D and 3D imaging. Cluster ion beam polishing (e.g. C₆₀⁺ or Ar_2,500⁺) was necessary to remove the FIB beam-induced damage, and the new instrument configuration facilitates cluster ion polishing. We will highlight certain aspects of the studies for presentation.

[1] A. Wucher, G.L. Fisher and C.M. Mahoney, Three-Dimensional Imaging with Cluster Ion Beams (p. 207-246) in Cluster Secondary Ion Mass Spectrometry: Principles and Applications, C.M. Mahoney (Ed.), Wiley & Sons, N.J. (2013).

Dictyostelium discoideum, a model eukaryotic amoeba (~10 µm in size), exhibits intriguing characteristics including unicellular and multicellular life stages as well as directional sensing when forming aggregation streams. While there is a good understanding of chemotaxic signaling pathways and their activators, little is known about the distribution of these molecular species within the individual D. discoideum cell. In the present work, pseudopods of chemotaxing D. discoideum are characterized using Mass Spectrometry Imaging at ~300nm spatial resolution. For the first time, asymmetric signals at the leading edge can be identified with high and ultra-high mass resolution to generate 2D and 3D molecular maps. Briefly, wild type Ax2 cells were grown on flat gold-coated silicon chips and starved for 17 hours in water. After freeze-drying, single cells during chemotaxis were analyzed (in triplicates) using a dual beam IonToF⁵ TOF-SIMS spectrometer setup with a 25keV Bi₃⁺ imaging probe and a 20keV Ar₁₅₀₀⁺ sputtering beam. Two dimensional images were collected as a function of the sample depth and allowed for the generation of 3D maps of single molecular components. Due to the complexity of the sample, molecular identification was performed using ultrahigh mass resolution C₆₀ FT-ICR SIMS on a parallel sample at lower spatial resolution. Inspection of the 2D images and molecular IDs permitted the visualization of chemotaxic activators at protruding pseudopods exhibiting directional sensing without the need of exogenous labels. Results will be shown for the molecular localization and identification at the single cell level.

The high spatial resolution of ToF-SIMS has made it a good candidate for localizing molecularly specific changes within cells and tissues. In order to maximize the information gained from ToF-SIMS imaging it is of interest to combine the chemically specific information of the SIMS with other imaging methods including optical microscopy and MALDI-MS. In order to optimize the correlation between images it is important to maximize the image contrast of the SIMS images. Producing high contrast ToF-SIMS images requires collecting enough counts to see image features as well as maintaining good beam focus. The count rate of secondary ions is severely limited by the ionization efficiency of the ToF-SIMS process. In order to increase the count rate the user has several options including changing the primary ion, increasing the number of scans taken, or summing the signal from a depth profile (Summed depth profile). Collecting a summed depth profile is a tempting option since one could assume that the result would be a significant gain in counts from the summed signal from each “slice” of the profile. However, even argon cluster beams that are know to minimize damage during depth cause significant loss in higher mass lipid signals during ToF-SIMS depth profiles. In order to optimize the conditions for ToF-SIMS image collection we compared using summed depth profiles to sputtering for a given time followed by imaging to a fixed primary ion dose. It was found that a light sputter for 2 sec (1.3e13 ions/cm2) to 10 sec (6.5e13 ions/cm2) followed by analyzing to a primary ion dose of 1e12 ions/cm2 produced images with comparable contrast, but higher lipid signals compared with summed depth profiles. It was also noted that a light sputter improved image contrast compared to images acquire using no pre-sputter. This could be a result of removing contaminants from the surface that degrade the image contrast.

As Mass Spectrometry Imaging (MSI) has progressed from the development stage to delivering impact in real biological and medical studies, the need to combine the MSI data with other imaging techniques has become increasingly apparent. Combining ToF-SIMS images with complementary imaging modes, such as optical micrographs, secondary electron images, Raman images and MALDI-MS images, is often essential to provide a reliable, unambiguous interpretation of the data. Analyzing multi-modal image data, however, can be a formidable challenge due to the increased volume and complexity of the data.

Image Fusion is a potentially valuable approach to maximizing the value of multi-modal imaging. The goal of image fusion is to combine the strengths of complementary techniques to create a single high information content image, without introducing artefacts in either the spatial or spectral domains. A wide variety of fusion algorithms have been developed for use in both medical diagnostics and optical remote sensing. These Image Fusion algorithms fall into two major classes, those that operate in the spatial domain (i.e. Intensity Hue Saturation (HIS), Principal Components Analysis (i.e. PCA), Brovey Transform (BT)) and those that operate in a frequency domain (Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT), High Pass Filter (HPF)). Although these Image Fusion algorithms can produce dramatic improvements in image sharpness and contrast, they can also lead to significant artefacts and care must be taken to ensure reliable results.

We have tested a range of fusion algorithms on synthetic MSI images and real MSI images of both model samples and animal tissues. Our results indicate that MSI is highly sensitive to fusion artefacts, making many of the approaches that have been used in optical sensing ill-suited for SIMS images. In order to minimize artefacts and produce reliable results, the algorithms must be adapted to account for the unique characteristics of each imaging mode. Of the methods in the literature, PCA Image Fusion is the most readily adapted for use with ToF-SIMS. We have developed methods for adapting PCA fusion for optimal use with ToF-SIMS and Raman images. Combinations of PCA with discrete wavelet transformation or discrete cosine transformation have provided the most successful route to minimizing fusion artefacts. PCA image fusion can be a valuable technique for reducing noise and improving image contrast and sharpness of ToF-SIMS and MALDI-MS images. With appropriate attention to the unique characteristics of each spectrometry, this can be done without significant artefacts or distortion of the spectral detail.

Combining images from multiple techniques sounds like a wonderful idea, but how easy is it? Can we analyse the same sample? What about spatial resolution differences? How do we register (align) multiple images? Once we have our registration, how do we combine the information in each image to best effect? In this contribution we will seek to understand the issues, compare and contrast different approaches, and show the outcomes.

Using ToF-SIMS and FTIR as examples of hyperspectral modalities, together with fluorescence microscopy, we will explore the problem space for image alignment. Once our data is aligned, we show how the techniques of multivariate statistics, non-parametric methodologies and machine learning can bring different facets of the data space to light.

Overall, we will highlight how combining the outputs from different hyperspectral modalities and microscopies can create an outcome greater than the sum of the parts.

Image fusion has garnered attention recently for improving the visual quality of SIMS images by merging the data with data from a source which is capable of providing higher resolution. The improved resolution could be due to better imaging focus, an inherent increase in the number of pixels, or higher signal intensity. Image fusion using pan-sharpening, a subset which works to preserve the color integrity of the SIMS image, and thus the chemical information, has been previously applied to pairs of SEM and SIMS images acquired using the same instrument.

The ultimate goal is to make image fusion with SIMS a universal approach so that any source of high resolution data could be used. One of the most important assumptions that must be made to perform image fusion is that the two images are of the same exact area. In order to create a universal technique so that external sources of high resolution images could be used, image registration needs to be implemented. Image registration is non-trivial, especially when it comes to correctly and accurately fusing SIMS images with those acquired with a different modality. In order to create a simple-to-use and reproducible system for image registration, the Insight and Registration Toolkit (ITK) has been implemented for preprocessing an image pair before fusion. Image pairs with known offsets were successfully registered before performing image fusion to validate the method. By successfully implementing image registration, the possibilities of fusing SIMS images with high resolution image sources, such as optical microscopy and fluorescence, now exist.