Preparation of Nanoparticles for ToF-SIMS and XPS Analysis

This paper presents a variety of sample preparation methods for surface analysis of nanoparticles. Since the physicochemical properties of a specific NP will define both the optimal method for sample preparation (e.g., drop-casting vs. spin coating) and the best procedure for that method (for example, requiring different substrates or solvents), the suitability of the method used should be validated via alternative analytical methods and optimized if necessary. The results seen in this publication are consistent with previously published literature in showing the need for consistent protocols and procedures for sample preparation as well as the need for quality checks to ensure that the sample preparation and purification methods are appropriate, successful, and do not damage the nanoparticles^{22,33,34,35,36}.

Sampling and storage methods for NPs have not been addressed here, as they are described in detail in various other references^{14,15,16,17,18,34,37,38,39}. Naturally, great care should be taken that the samples analyzed are representative of the overall nanoparticle distribution and suitable sampling methods developed and validated. Storage conditions have also been shown to strongly affect nanoparticle properties over a period of months and should therefore be carefully considered. As an example, we recommend that nanoparticles should be stored in small amounts in sealed containers away from light, ideally below 4 °C. It is also crucial that storage, sampling and sample preparation is consistently performed according to validated procedures as well as is being documented in detail. This documentation should include the metadata from the NPs themselves, such as provenance information and storage conditions⁴⁰. Tools such as electronic lab notebooks (ELNs) may be useful for consistent documentation of procedures and NP metadata, as well as enabling the production of data according to the FAIR principle (Findable, Accessible, Interoperable, and Reusable).

Accurate and correct surface analysis of NPs firstly requires a suitable choice of substrate. We have used cleaned Si wafers as substrates because they are readily available, durable, easily cleaned, conducting and sufficiently flat, however depending on the goals of the analysis the oxide surface layer can be a drawback, as the adventitious hydrocarbons on the substrate cannot be differentiated from those on the nanoparticles. When necessary, other materials such gold or polymeric coatings on Si wafers, Si₃N₄ wafers, or HOPG (highly oriented pyrolytic graphite) can be used^19,20,21,22. The first step in sample preparation described in this paper is cleaning the Si wafer, shown as a schematic in Figure 1. The efficacy of the cleaning process can be verified by a variety of methods including XPS, as shown in Figure 2. The main contaminant (adventitious carbon) is typical for samples stored in air and is significantly reduced after the cleaning process. Additionally, hydroxylating the wafer surface via UV or ozone treatment avoids the coffee-ring effect from deposition from aqueous suspension by enhancing wettability and leading, therefore, to a more homogeneous distribution of the nanoparticles as shown in Figure 3. Alternative wet chemical cleaning methods for Si wafers may be used as needed; here only a reproducibly clean surface is required rather than the complete removal of all organic contaminants or the oxide layer. If the protocol is paused between the cleaning and suspension deposition steps, the wafer should be treated again under plasma or UV/ozone and the suspension deposited ideally within 15 min of treatment.

The suspension of 60 nm Au-Ag core-shell nanoparticles shown in section 2.2 contained a significant amount of sodium citrate as a stabilizer, which is a common occurrence in nanoparticle suspensions. For accurate analysis of these particles and their surface properties, particularly via XPS, as much stabilizer should be removed as possible, as it attenuates the signal from the nanoparticles and causes charging effects. In order to establish the optimal purification method for these nanoparticles, shown as SEM micrographs in Figure 4, they were either dialyzed in ultrapure water or purified using centrifugation and re-dispersion in triplicate. Although dialysis would seem a gentler method and centrifugation and re-dispersion more likely to cause agglomeration and aggregation of the particles, the SEM images show significant deformation and damage of the Au-Ag nanoparticles after dialysis (Figure 4B), while the centrifuged/re-dispersed particles are still intact (Figure 4C). This is particularly remarkable with metallic nanoparticles; our hypothesis is that there is an optimum amount of sodium citrate that enables some stabilization of the solution while not interfering with the signal for the nanoparticles, and removal of too much stabilizer causes damage to the nanoparticles. A previous report shows that there is an optimum number of centrifugation cycles for removal of most of the sodium citrate; exceeding this number causes some NP aggregation³³. In this study, nine dialysis cycles (a total of 36 h) were required to obtain similar citrate concentration; however, this method resulted in a higher amount of aggregation than centrifugation as well as causing a decrease in surface functionalization. These results demonstrate the importance of verifying each step in the preparation procedure for each different type of nanoparticle, particularly with unknown samples.

The 60 nm Au-Ag core-shell nanoparticles used in this example are suitable for drop-casting due to their electrical conductivity, because charging effects are not an issue and a thick spot can be generated by repeated deposition using relatively little equipment. This thicker layer has the advantage of giving more reproducible measurements, and casting from a more concentrated suspension can save time by reducing the number of deposition steps. The deposition can be influenced by the substrate wettability; poor wetting can produce a thick nanoparticle spot which is advantageous for conductive samples, while good wetting can produce a more homogeneous nanoparticle layer, which can be useful for both conductive and insulating samples. As described in the protocol, drop-casting of nanoparticle suspensions usually requires repeated applications to obtain a thick layer with full coverage; this should be verified using XPS, but may also be quickly and easily verified using optical microscopy. Figure 5 shows the evolution of droplet coverage in a drop-casting of Au-Ag core-shell nanoparticles from aqueous solution; in this case, 13 drop-casting steps are required to achieve full coverage. Drop casting is particularly suitable for conductive particles, or the ones where charging effects can be adequately compensated. As with the other methods described in this publication, drop-casting should be optimized for each sample as different NP materials will have different properties concerning information depth and concentration and film thickness limits. It is important to avoid too thick films which can cause stacking of organics in turn inhibiting the NP signal.

A homogeneous and good quality coating helps to ensure consistent and reproducible results. In addition to the suspension concentration, solvent, and spin-coating parameters, the quality of spin-coated suspensions can also be negatively influenced by the presence of dust or other large macro- or microscopic particles. Figure 6 shows the improvement in spin-coating quality of a nanoparticle suspension after filtration with an 0.45 µm syringe filter. The filter should be selected to ensure that it does not remove nanoparticles from the suspension. The three different suspension concentrations described in the protocol (90, 9.0 and 0.9 mg/mL of 135 nm PS-PTFE core-shell nanoparticles) were spin-cast under the same conditions and analyzed using SEM and XPS. The top image and spectrum in Figure 7 show the film cast from the 90 mg/mL suspension, which shows a thick and gapless multilayer coverage in the SEM image as well as a notable absence of Si peaks in the CPS spectra, indicating no contribution of the substrate to the spectrum. This sample is ideal for XPS or ToF-SIMS analysis; additionally, the smaller F1s peaks from the shell of the particles can be clearly seen in the absence of a large signal from the substrate. The second sample cast from the 9.0 mg/mL suspension shows the particles in small single-layer agglomerates, which do not completely cover the surface. This sample is too thin and inhomogeneous for XPS or ToF-SIMS analysis. Furthermore, quantitative analysis can be impaired due to the contribution of adventitious carbon on the substrate even after careful cleaning; at the very least, such an effect must be considered in the uncertainty budget of the measurement. This sample would, however, be ideal for SEM or TEM analysis of particle size distribution using image analysis software, as the particles exist in a single layer and in a sufficient number (within the image) to provide a statistically significant evaluation. The sample cast from the lowest concentration (0.9 mg/mL) does not provide either continuous coverage or sufficient particle density to make it suitable for analysis of either surface chemistry or particle size distribution. A reliable quantitative analysis is not at all possible due to the dominant influence of the substrate.

Al₂O₃-TiO₂ core-shell NPs with either a PDMS or glycerol outer layer were prepared via drop-casting from suspension as well as from powder using the “stick-and-go” method in order to compare the effects of the different preparation methods on the sensitive outer layer. The samples were analyzed with ToF-SIMS, where in the spectra was analyzed using Principal Components Analysis (PCA). PCA is a statistical technique for reducing the dimensionality of large data sets by creating new uncorrelated variables (the principal components), which maximize the variance in the data^{41,42,43,44,45}. The separation of different sample sets on the principal component graph allows the results to be more easily analyzed and grouped. On the PCA scores plot in Figure 8B, which shows the discrimination power of each data set in comparison to all other data sets (i.e., between different sample sets) the two samples prepared from powder show very different scores, while the samples prepared from dispersion show very similar scores. The loading plots shown in Figure 8C indicate the relationship between variables, i.e., which peaks contribute the most to the respective principal components. All principal components are sorted according to their contribution to the observed difference between the data sets, i.e., PCA1 contributes the most to the observed separation of the different data sets. PC1 is dominated by the presence (PDMS-coated NPs prepared from powder) or absence (all other samples) of PDMS peaks, while PC2, the factor accounting for the second largest variation within the data sets, enables the differentiation of the Al₂O₃ and the organic capping on the NPs. This indicates that the measured spectra of NPs prepared from suspension are very similar and suggests that the PDMS and glycerol layers may have been removed or damaged by preparation from suspension, from either the suspension itself or the drying process, with dominating signals from the Al₂O₃ or TiO₂.

While pressed pellets can provide advantages for preparation of powdered samples such as ease of handling and stability in ultrahigh-vacuum instruments (including the ability to sputter without dislodging NPs in the high-vacuum chamber), the high forces involved may also damage sensitive nanoparticles, as has already been seen with other preparation methods. A suitable protocol should be prepared and validated.

In the case of NP dispersions, cryofixation of drop-cast sample suspensions avoids coffee ring effects (because of the instantaneous fixing of the NP suspension and therefore elimination of drying effects) as well as the preservation of larger structures present in the suspension. Additionally, the application of adhesive tape is avoided. This in turn is reflected in reduced signals, which may be attributed to salts, contaminants, or other artifacts of the sample preparation procedure in the respective mass spectra as shown in Figure 9. The main advantage of cryofixation is the ability to conserve “as is” the chemical space around the nanoparticles and/or the chemical entity of the particle agglomerates or heteroagglomerates as well as their correlation to biological features within tissues or single cells or even the co-localization to intracellular compartments, without disruption from sample handling steps such as drying, drop-casting, etc⁴⁶_’⁴⁷. We have demonstrated the applicability of the cryofixation technique within the current paper and have highlighted the advantages of cryofixation for TiO₂ nanoparticles. We stress that cryofixation is particularly suitable for the analysis of biological samples due in their natural state without the dislocation of chemicals due to sample preparation artifacts. For more in-depth information about fixation techniques for biological samples the reader is referred to literature^{19,25,27,48,49}.

Table 1: Comparison of various methods for surface analysis.

Table 2: Comparison of different sample preparation methods.

Figure 1: Cleaning process for Si wafers. Please click here to view a larger version of this figure.

Figure 2: XP spectra of Si wafer before and after cleaning. Survey before (gray) and after (red) cleaning, showing the decrease of the carbon amount from 13 at% to 2 at%. The spectra were obtained with a Kratos Supra DLD (Manchester, UK) with a monochromatic Al Kα radiation. The samples were fixed with double-adhesive tape on the sample holder, pass energy was 80 eV, step width 1 eV, dwell time 500 ms. The “hybrid lens mode” was used. The X-ray spot size was 300 x 700 µm². A flood gun was used for charge compensation. For quantitative analysis, the software package UNIFit 2020⁵⁰ was used, using the peak areas of the corresponding photoelectron peaks corrected with a Tougaard background and normalized with Scofield factors, inelastic mean free pathways and the transmission function. Please click here to view a larger version of this figure.

Figure 3: Effect of UV/Ozone cleaning on homogeneity of particle dispersion in the drop-casting of PTFE-PMMA core-shell nanoparticles from aqueous suspension. The wafers cleaned with UV/ozone show a significant decrease in coffee-rings, as well as better adhesion of the particles to the surface. Please click here to view a larger version of this figure.

Figure 4: Treatment options for removing impurities (e.g., stabilizers) from nanoparticle suspensions SEM images showing the effect of dialysis (top right) and centrifugation and re-dispersion in triplicate (bottom right) on 60 nm Au-Ag core-shell nanoparticles. The nanoparticles are clearly damaged by the dialysis, while centrifugation has no visible affect. All scale bars are 100 nm. Please click here to view a larger version of this figure.

Figure 5: Optical microscope images from drop-casting of 60 nm diameter Au-Ag core-shell nanoparticles from aqueous suspension onto silicon wafers, showing sufficient coverage after 13 drops. Please click here to view a larger version of this figure.

Figure 6: Spin-coated nanoparticle suspension, before (left) and after (right) filtration with an 0.45 µm syringe filter. The improvement in quality after filtration can clearly be seen. Please click here to view a larger version of this figure.

Figure 7: SEM images and XPS spectra of PMMA-PTFE core-shell nanoparticles spin-cast at various concentrations, showing the effect of substrate peaks (from insufficient coverage) on the XPS spectra. Please click here to view a larger version of this figure.

Figure 8: Principal Component Analysis (PCA) score plot, derived from ToF-SIMS spectra of glycerol- and PDMS-coated Al₂O₃-TiO₂ core-shell NPs. (A) Schematic of NP structure; (B) Scores and (C) Loading plots after ToF-SIMS analysis of drop-cast (dispersion) and ”stick-and go“ (powder) preparation methods. PC1 represents peaks correlating to PDMS fragments; PC2 separates samples with an organic coating (samples prepared from powder) from Al₂O₃ peaks seemingly without surface coating. Spectra were measured in positive mode on an IONTOF ToF-SIMS IV instrument (ION-TOF GmbH, Münster, Germany) in the spectrometry mode (HCBU) with a 25 kV Bi­₃⁺ ion beam with a maximum dose density of 10¹² ions/cm². A field of view of 150 x 150 µm was scanned in sawtooth mode with 125 x 125 pixels. Please click here to view a larger version of this figure.

Figure 9: Section of ToF-SIMS mass spectra of TiO₂ NPs. (A) prepared from powder with the “stick and go” method and (B) after cryofixation of the NP dispersion. A ToF-SIMS instrument (ION-TOF V; Ion-TOF GmbH, Münster, Germany) was used for mass spectrometry analyses with a pulsed 30 keV Bi₃⁺ liquid metal ion gun (LMIG, direct current (dc), 16 nA). Each spectrum was acquired by scanning the ion beam over a sample area of 500 × 500 µm. Positive secondary ions were acquired in the mass range 0–1,200 Da using 10⁶ Bi₃⁺ pulses. Please click here to view a larger version of this figure.