Deep depth profiling using gas cluster secondary ion mass spectrometry: Micrometer topography development and effects on depth resolution

Abstract

Secondary ion mass spectrometry using the argon cluster primary ion beam enables molecular compositional depth profiling of organic thin films with minimal loss of chemical information or changes in sputter rate. However, for depth profiles of thicker organic films (> 10 μm of sputtered depth) we have observed the rapid formation of micron-scale topography in the shape of pillars that significantly affect both the linearity of the sputter yield and depth resolution. To minimize distortions in the 3D reconstruction of the sample due to this topography, a step-wise, staggered sample rotation was employed. By using polymer spheres embedded in an organic film, it was possible to measure the depth resolution at the film-sphere interface as a function of sputtered depth and observe when possible distortions in the 3D image occurred. In this way, it was possible to quantitatively measure the effect of micron-scale topography and sample rotation on the quality of the depth profile.

Introduction

Argon gas cluster ion sources have revolutionized molecular depth profiling in time-of-flight secondary ion mass spectrometry (ToF-SIMS). Ar cluster bombardment allows sputtering of samples such as conjugated polymers that were previously not feasible using SF₅⁺ or C₆₀⁺ sources,^1–3 and the large cluster size with low incidence energy per atom combine to minimize damage accumulation^4–6 and topography development^7–9. These benefits allow for molecular compositional depth profiling of organic films with depth resolutions approaching a few nanometers, while preserving the sputter rate and chemical information even at sputter depths greater than 10s of micrometers.^10–12 Although the strength of ToF-SIMS depth profiling lies in its ability to characterize nanometer features, the cluster source gives it the flexibility to provide mass-specific 3D images of chemical components inside very thick films such as biological samples,¹³ biomaterials,¹⁴ and drug delivery systems¹² where tens and even hundreds of micrometers of material need to be removed. This advantage potentially provides an alternative to visualization techniques that require making a cross section of soft films where structural damage could occur during sectioning. However, accurate 3D reconstruction of such ultrathick systems requires understanding of complex processes that govern parameters such as sputter rate, damage accumulation, and topography generation at depths that have not been explored in detail to date.

While ion bombardment artifacts such as sputter yield transients at surfaces^{15, 16} and sputter-induced roughening^17–19 can be mitigated for shallow depth profiles (< 10 nm to 1 μm), deep depth profiling of ultrathick films (10 μm and above) has brought to our attention new artifacts that have only recently been characterized. First, commercial ToF-SIMS instruments have analysis beams that are fixed at a 45° incidence angle which leads to a lateral shift in the secondary ion image as a function of probed depth.²⁰ In other words, 1 μm of sputtering leads to an equivalent lateral shift in the image due to the shift in the beam position as material is removed. The result of this is that a micrometer-sized sphere embedded in a film would appear as a diagonally stretched spheroid in the 3D SIMS image. Second, the sputter beam is also fixed at 45° which leads to a lateral shift in the sputter crater with depth, forming a wedge-shaped crater.¹¹ Since the two ion beams are orthogonal to each other, this results in the reduction of the analysis area with depth which places a physical limit on the maximum analyzable depth. In addition, the gas cluster source creates a pressure wave that results in a very low angular distribution for the sputtered material of less than 30° from the surface.^{21, 22} Inside a wedge-shaped crater, the sputtered flux will redeposit along the crater wall and affect the depth resolution through resputtering.¹¹ Lastly, the crater bottom roughness encountered in deep depth profiling is on the order of a few micrometers,^{11, 12} which is thought to induce non-uniform sputter yields and degrade depth resolution, but no studies have yet fully addressed the impact of this effect on the quality of the 3D SIMS image or explored approaches for its mitigation.

In our previous report, 3D ToF-SIMS imaging was used to measure the size and spatial distributions of active pharmaceutic particles inside ultrathick drug delivery films.¹² These are important physical characteristics that can affect the performance, dosage, and stability of the end product,^23–26 and the ability to reliably provide this metrology inside a solid matrix is currently needed to guide formulation choices and manufacturing processes. ToF-SIMS was able to visualize the particles inside this film to a sputtered depth of roughly 150 μm, but micron-scale topography at the crater bottom contributed to non-uniform sputter yields and the sizing of the particles became problematic. In this work, to investigate the source of the image distortion, (43 ± 5) μm spheres made of polystyrene (PS) were embedded inside a film of gelatin to simulate particles in a drug delivery film. The edges of the spheres will be used to identify how the micron-scale roughness would affect parameters such as sputter yield and depth resolution, and ultimately contribute to distortions in the 3D images. In addition, since roughness is known to contribute to sputter yield changes in thin films and approaches such as sample rotation^27–29 exist to mitigate this effect, this approach will be employed for deep depth profiling to examine its effect on roughness and mitigation of depth resolution.

Experiment

Sample Preparation.

10 cm Si(100) wafers purchased from Virginia Semiconductors^* (Fredericksburg, VA) were diced into 15 mm × 15 mm square pieces using the Disco DAD341 dicing saw (Tokyo, Japan) equipped with a 15 μm diamond impregnated metal blade, and cleaned sequentially in a sonicated bath for 10 min in methylene chloride, acetone, and methanol to serve as substrates. Porcine gelatin (175 g bloom Type A) and pullulan from Sigma-Aldrich (St. Louis, MO) were dissolved in ultrapure water at a concentration of 4 wt% and cast onto individual Si pieces. They were degassed and allowed to dry in a low vacuum desiccator overnight. (43 ± 5) μm polystyrene (PS) microspheres with a density of 1.07 gm/mL were purchased from Cospheric, LLC (Santa Barbara, CA), and a very small amount was mixed with the gelatin solution to create films with embedded spheres, which were also dried in a desiccator before use. Individual spheres on Si wafers were prepared by mixing a small amount of the spheres in a 0.1 wt% gelatin solution, and spun cast at a speed of 209 rad/s (2,000 rpm) for 60 s. 25 mm × 75 mm × 1 mm slides of thermally annealed polyethylene terephthalate glycol (PETG) from Curbell Plastics (Phoenix, AZ) were used as received. All samples were analyzed within 5 days of production and stored at room temperature.

Sample Analysis.

ToF-SIMS depth profiles were performed using an IONTOF IV (Münster, Germany) equipped with a 30 keV Bi₃⁺ liquid metal ion source for analysis and a 20 keV Ar_2600±1000⁺ source for sputtering (referred to as GCIB, gas cluster ion beam), both sources striking the sample surface at an angle of 45°. Dual beam depth profiling was performed in non-interlaced mode with 1 scan of analysis with a resolution of (256 × 256) pixels, 10 scans of sputtering, and 2 s of charge compensation per cycle, where both the analysis and sputter rasters were kept inside a (500 × 500) μm area. The corresponding ion doses were 1.9 × 10⁹ ions/cm² (0.12 pA) for Bi₃⁺, and between 2.1 × 10¹⁴ ions/cm² to 2.6 × 10¹⁴ ions/cm² (5.1 nA to 6.4 nA) per cycle for the cluster source due to day-to-day fluctuations in the beam current. Sample rotation was performed manually with the 6-axis stage equipped with the instrument, and was rotated by 90° after every 8 × 10¹⁵ ions/cm² of sputter dose after stopping data acquisition. In other words, dual beam depth profiling was performed up to a sputter dose of 8 × 10¹⁵ ions/cm², the data acquisition stopped, the sample rotated, and a new depth profiling started. This was repeated until the target dose was achieved. Since rotation did not always occur at the rotational axis, sample alignment needed to be performed using both the optical and secondary ion images with PS spheres acting as fiducial markers. Depth resolution measurements were made first by obtaining the yz image slices of PS (m/z 91.054, C₇H₇⁺) in the SurfaceLab software, plotting the intensity versus distance curve from the linescan, fitting the curve using a cubic function to capture any potential asymmetry of the profile, and measuring the 16%−84% distance.

Secondary electron microscopy images of the crater topography were acquired using a FEI Apreo (Hillsboro, OR) system with an incident electron energy of 1 keV with a current of 0.1 nA for analyzing the outermost morphology of the insulating surfaces. The instrument is equipped with a set of Trinity in-lens detectors which can offer better topographic and crystal orientation contrast, but a standard Everhart-Thornley detector (ETD) situated in a conventional position was used due to uncontrolled deflection of secondary electrons from sample charging. Working distance was held at around 8 mm to enhance topographical contrast and ETD detection efficiency. Images were acquired at 3072 × 2188 pixels and integrated 16 times using a dwell time of 100 ns/pixel. Crater depth and topography measurements were performed using a Dektak XT stylus profilometer (Bruker Corporation, Tucson, AZ) equipped with a 2.5 μm radius stylus tip, having a vertical resolution of 1 nm, and a lateral resolution of 16 nm. A 2 mm line scan was used to ensure that a large enough region outside the crater was measured to provide sufficient area for leveling. The depth of the crater bottom was estimated by averaging the 5 lowest points in each line profile. For all analyses, values are averages of at least three measurements and the error represents the standard deviation unless otherwise noted.

Results and Discussion

Sputter yields.

In this report, gelatin films (and pullulan for a smaller set of studies) was used to mimic the composition of oral drug delivery systems that are comprised primarily of water-soluble polymers such as polyethylene oxide, methylcellulose, polyvinyl pyrrolidone, gelatin, and pullulan.³⁰ Together with (43 ± 5) μm PS spheres, they were used to simulate particles in a drug delivery film, providing a model system for evaluating the quality of 3D images.

The depth profiles of three different matrices are shown in Figure 1. The profiles all showed a linear increase in depth as a function of dose up to about 42 μm in sputtered depth (or roughly 7 × 10¹⁶ ions/cm²), beyond which a sudden change in sputter rate was experienced by gelatin and pullulan. This inflection point in the sputter rate was most likely due to topography reaching some critical threshold, and this will be discussed in detail in the following sections. Sputter rates for the different materials were determined from the depth profiles. Gelatin showed a high sputter yield of (67 ± 5) nm³/ion, or (0.026 ± 0.002) nm³/atom, up to a depth of roughly 42 μm (7 × 10¹⁶ ions/cm²). For pullulan, the sputter yield was very similar at (76 ± 4) nm³/ion, or (0.029 ± 4) nm³/atom, up to a depth of 40 μm (7 × 10¹⁶ ions/cm²). The thickness of the cast pullulan film was less than 50 μm, and so the nonlinear region of the sputter rate could not be obtained. For the linear regions, these numbers were comparable to literature reported values for argon clusters of the same energy per atom.^31–34 It is important to note that sputter yields can change quite significantly between materials and also are a function of analysis parameters including primary ion beam type, cluster size, and energy.^{32, 35} In addition, there is an indication that the ratio of the analysis to the sputter ion dose per cycle can also affect these numbers (Graham and Muramoto, unpublished results).

Figure 1.

Sputtered depth vs. ion dose for gelatin with PS spheres (○), pure pullulan (Δ), pure thermally extruded PET (□), and gelatin with PS spheres using staggered rotation (●). The sputter yield of pure gelatin was very similar to that containing PS spheres, and was not included to facilitate viewing. Due to significant roughness of gelatin and pullulan, the sputtered depths were estimated from the average of 5 lowest points along the crater bottom measured using a profilometer, with 3 scans per crater.

As a comparison to spun cast films, thermally annealed polyethylene terephthalate (PET) was profiled resulting in a sputter yield similar to gelatin at (67 ± 5) nm³/ion but did not exhibit a change in the rate beyond a ion dose of (1 × 10¹⁷ ions/cm²). The reason for this linearity is most likely due to differences in topography, with the annealed polymer experiencing significantly less topography generation due to well-organized nanostructures and polymer chains as opposed to cast polymer films that contain disorganized and ill-defined polymer chains.³⁶ Therefore, the onset of the inflection point in sputter rate is likely delayed rather than not existing, since the extent of topography for this polymer was also observed to increase with sputter dose (more discussion to follow).

The sputter yield of the embedded PS spheres calculated using the equation below was (63 ± 10) nm³/ion, slightly lower than the reported 90 nm³/ion measured for a flat film of PS sputtered using a 20 keV cluster source.¹ The uncertainty in the calculation represents errors propagated from the actual size distribution of the sphere and also from the standard deviation of the ion dose that was required to sputter through different spheres.

$$Y=\frac{volume}{ion}=\frac{\frac{4}{3}\pi r^3}{\frac{2}{3}{\mathrm{ɸ}}_T{f}_A{f}_{z}A}=\frac{2\pi r^3}{{\mathrm{ɸ}}_T{f}_A{f}_{z}A}$$ (1)

In equation 1, r is the known radius of the spheres, ɸ_T is the total ion dose in ions/nm² for the entire sputtered crater, A is the raster area in nm², ƒ_A is the area fraction or coverage of the embedded sphere relative to the raster area obtained from a region of interest (ROI) analysis, and ƒ_z is the depth fraction or depth of the embedded sphere relative to the sputtered depth obtained from a 3D reconstructed image, such as from the one presented in Figure 2. The denominator represents the number of ions injected into an ellipsoid as seen in the 3D reconstructed image, and was determined as follows:

$$n_{ellipsoid}={\mathrm{ɸ}}_T\frac{v}{d}={\mathrm{ɸ}}_T\frac{\frac{4}{3}\pi \left(\frac{f_{A}A}{\pi }\right)\left(\frac{f_{z}d}{2}\right)}{d}=\frac{2}{3}{\mathrm{ɸ}}_T{f}_A{f}_{z}A$$ (2)

where v is the volume of the ellipsoid from the 3D reconstructed image, and d is the sputtered depth of the crater. A product of the three semi-axes are required to calculate an ellipsoid (i.e., 4πabc/3), so in the numerator, the arguments in the parentheses represent ab (area of ellipse equals πab, determined from the ROI) and c (determined from the 3D reconstructed image), respectively. Moreover, measurements of craters in gelatin with and without the spheres yielded very similar sputter yields, indicating as well that the yields of PS spheres and gelatin were relatively similar.

Figure 2.

3D reconstruction images of a ToF-SIMS depth profile, produced using the ‘3D Reconstruction’ menu in the SurfaceLab software, displaying the (a) xy, (b) yz, and (c) xz slices of an embedded PS sphere. The information was presented as a series of stacks, and coordinates can be displayed to determine z1 and z2, as well as the total number of slices Z. A 20 pixel square was used in the xy slice for signal integration. The total ion dose used for this data set was 7 × 10¹⁶ ions/cm², and the signal for the spheres typically survived roughly 3.5 × 10¹⁶ ions/cm² as seen in (b) and (c). The xz slice in (c) shows the lateral shift in image due to the shift in analysis beam position as material was sputtered away.²⁰

Spheres embedded inside an organic film can be sputtered away³³ by minimizing the angular dependence of the sputter yield³⁷, but the sputter yield of the spheres by themselves sitting on a Si surface could not be determined. This was because the spherical surface experienced localized regions of differential sputtering. As can be seen in the SEM images in Figure 3, these spheres were not completely sputtered away after an ion dose of 3 × 10¹⁶ ions/cm², which was sufficient for their removal when embedded in gelatin as can be seen in Figure 2. Even after an ion dose of 5 × 10¹⁶ ions/cm², a dose expected to be sufficient to remove about 35 μm of material as suggested by the profiles in Figure 1, not much sputtering seemed to have taken place and a majority of the structure still remained. In addition, the molecular ion was no longer detectable at this point, probably due to the distortion of the extraction field along the sample contour and deflection of secondary ions to outside of the analyzer acceptance angle.^{38, 39}

Figure 3.

Secondary electron images at a tilted view (45°) showing the PS sphere on a Si wafer sputtered within a 500 μm × 500 μm GCIB raster, with ion doses of (a) 4 × 10¹⁵ ions/cm², (b) 1 × 10¹⁶ ions/cm², (c) 2 × 10¹⁶ ions/cm², and (d) 5 × 10¹⁶ ions/cm². The cluster ions are coming from the right at 45°. Different spheres were exposed to different ion doses. The dark area on the substrate is a thin film of gelatin used to adhere the spheres to the surface.

Of particular interest was the formation of micron-sized pillars from within the spheres protruding towards the GCIB source. Small protrusions seen in the sputtered surface in Figures 3a and 3b indicate some formation of a cap which then shelters the underlying material from erosion, which seems to lead to the development of large pillars. This topography growth through masked sputtering is very similar to those seen in studies of cluster induced topography such as the micron-sized “Hershey kisses” through the creation of a Si-C cap in the case of C₆₀ sputtering of Si,⁴⁰ and although not micron-sized, the elaborate design of In nano-cones by preferential GCIB sputtering of the InP substrate around the In nanoparticles which acted as masks.⁴¹ Therefore, in the case of PS spheres, it is possible that the seeds for pillar formation could have formed from the presence of inorganic and chemical impurities inside or on the surface of the spheres during their manufacture.⁴² In addition, several reports in the focused ion beam (FIB) community have noted the formation of “nano drops” (i.e., bubbles or voids) in the sub-surface region of the crater bottom following a high ion dose.^{43, 44} The hemispherical surface of the voids could theoretically serve as seeds for pillar formation due to the surface having angles that may not be amenable for sputtering.

Crater bottom topography.

Similar topography was observed in gelatin during the inspection of the crater bottom with SEM, which showed the formation of non-uniformly distributed micron-sized pillars across the entire bombarded surface (Figure 4a). The majority of the pillars were quite large and had a conical shape that increased in size from the tip to the base of the structures. Though the image may misleadingly represent the pillars to be taller than the depth of the crater, a profilometer trace revealed the tallest structure to be roughly half of the crater depth (d = 42.4 ± 2.1 μm at a GCIB ion dose of 7 × 10¹⁶ ions/cm²). Their angle was 45° to the surface plane, and was consistent with the GCIB incidence angle, as can be seen in Figure 4b where the tilted SEM view from the upstream side of the crater shows the tips of the pillars. Investigation of the surface surrounding the main crater area (i.e., areas that have received substantially lower ion dose) revealed the presence of smaller pillars from which the larger micro-pillars are thought to have originated. They emerged shortly after the gelatin had been exposed to cluster ions, and are not thought to be initially present on or inside the matrix. Additionally, at these length-scales, their formation is not thought to be due to any surface diffusion of atoms or molecules,⁴⁵ or to surface ripple formation through stress induced surface lifting^{46, 47} which are the dominant forces for the formation of nano-topography under ion bombardment.

Figure 4.

Secondary electron images of gelatin at a tilted view (45°) showing the crater in gelatin from the (a) downstream and (b) upstream sides, where the surface was bombarded with a GCIB ion dose of 7 × 10¹⁶ ions/cm². The numbers 1 through 6 mark the position of the embedded PS spheres. 3D constructed image of sphere 1 is shown in Figure 2.

In Figures 4a and 4b, the numbers represent the location of the embedded PS spheres. In their positions were the presence of circular humps, suggesting the presence of spheres even though the secondary ion signal for PS had completely decayed (sphere 1 is also shown in Figure 2 in the 3D reconstructed image). Profilometer measurements revealed that their heights varied an average of 10 μm to 15 μm from the crater bottom, which were consistent with the raised heights of partially buried spheres in the virgin film (seen in areas outside of the craters in subsequent figures). Therefore, these are simply raised humps in the virgin film being translated down as the film was sputtered, and their larger diameters of roughly 60 μm to 80 μm seem to support this claim.

It is also worth noting the apparent material contrast between the pillars and the surrounding gelatin. This is probably due to the extraction field created by the pillars,⁴⁸ which allows more secondary electrons to be emitted, but at the same time directs the beam of electrons towards a certain direction. This is demonstrated as bright pillars in the upstream electron micrograph (Figure 4b), and as darker pillars in the downstream image (Figure 4a). This also means that what appear to be pillars in the image but do not have contrast against gelatin are more likely to be a different type of structure on the surface.

To see the evolution of topography with increasing sputter ion dose, electron micrographs taken with a 45° sample tilt at different doses are shown in Figure 5. On gelatin, the morphology of the topography included micro-pillars and what appeared to be a similar structure lying flat on the crater bottom (which resembled bubbles). Both were seen to appear at a dose of less than 2 × 10¹⁶ ions/cm², and were seen to generally increase in both number and size with dose. Despite the incredibly rough topography, both the secondary ion and sputter yields remained constant up to an ion dose of 7 × 10¹⁶ ions/cm². Although elucidating the mechanism of pillar formation is not within the scope of this report, the primary pathway seems to be caused by the presence of impurities in the film given their spatial distribution. High magnification images of the peak of the pillars are presented in Figure S1 (supporting information) but could not support this theory. Another pathway could be due to particle growth from preferential sputtering of differently organized polymer chains. This is because the crater bottom of a thermally annealed PET polymer, which should contain more organized nanostructures, showed a different morphology. As can be seen in Figure 6, the structure was quite different and resembled fingernails, suggesting that the organization of polymer chains could affect morphological outcome as well as their number and spatial distribution. In terms of number, only a small fraction of the crater bottom was covered with topography even after a dose of 1 × 10¹⁷ ions/cm², which may explain the linearity of the sputter rate beyond this ion dose. The morphology of features in pullulan (a cast film similar to gelatin) was very similar to that of gelatin, and can be seen in Figure S2 (supporting information), as well as more images showing the evolution of topography in PET (Figure S3, Supporting Information).

Figure 5.

Secondary electron images of gelatin at a tilted view (45°) showing the crater bottom topography after GCIB ion doses of (a) 2 × 10¹⁶ ions/cm², (b) 4 × 10¹⁶ ions/cm², (c) 7 × 10¹⁶ ions/cm², and (d) 1 × 10¹⁷ ions/cm². The cluster ions are coming from the right. Gelatin films contain PS spheres and can be located from the circular humps on the surface. Each of the craters were formed on fresh areas of the sample.

Figure 6.

Secondary electron images of PET at a tilted view (45°) showing the crater bottom topography after GCIB ion doses of (a) 1 × 10¹⁷ ions/cm². A magnified view of the topographical feature is shown in (b). The cluster ions are coming from the right. Craters were formed on different areas of the sample.

For gelatin, the transition to a nonlinear sputter yield occurred at an ion dose of 7 × 10¹⁶ ions/cm², with minimal sputtering seen beyond 1 × 10¹⁷ ions/cm². Corresponding topographical images in Figures 5c and 5d did not show much difference in terms of the type and coverage of the morphological features, but a comparison with those in PET suggest that coverage of the micro-pillars could be the primary driving force in reducing the sputter rate. These topographical features exist because they don’t sputter as well as the crater bottom. Therefore, as these features populate the surface, less area exists for sputtering which would then reduce the overall sputter rate.

Studies of thin films have shown that nanometer-scale surface roughening is a result of a complex interplay between ion beam conditions and the sample;^{7, 49–54} the energy and incidence angle of the ion beam can either mitigate or worsen roughening, and in the sample, the presence of grain structures and boundaries can lead to differential sputtering. Sample rotation during sputtering has been shown to reduce roughening by averaging out the angular dependence of the sputter yield,^{27–29, 50, 55} and this was evaluated in this study to see whether the observed micrometer-scale topography could be reduced. As can be seen in Figure 7, even though the crater bottom was incredibly rough, a step-wise rotation of 90° after every 8 × 10¹⁵ ions/cm² of ion dose led to the decrease of the large micro-pillar formation, and allowed the sputter rate to stay linear beyond an ion dose of 7 × 10¹⁶ ions/cm² (Figure 1).

Figure 7.

Secondary electron images of gelatin at a tilted view (45°) showing the crater bottom topography after GCIB ion doses of (a) 1.7 × 10¹⁶ ions/cm², (b) 3.4 × 10¹⁶ ions/cm², (c) 6.8 × 10¹⁶ ions/cm², and (d) 1 × 10¹⁷ ions/cm², with the sample rotated 90° every 8 × 10¹⁵ ions/cm² of sputter dose. The numbers with arrows indicate the sequence and direction of the GCIB beam, with the respective craters being rotated a total of (a) two, (b) four, (c) eight, and (d) twelve times. The inset in (a) is a magnified view of the pillars. The craters were formed on different areas of the sample.

Careful inspection of the micrographs indicated that the micro-pillars were rather quick to form. In Figure 7a, there were numerous pillars with an average height of (0.8 ± 1.0) μm after only 8 × 10¹⁵ ions/cm² of ion dose, inside a crater that was (d = 11.9 ± 0.2) μm deep. Additional sputtering after sample rotation created a new set of pillars in the direction of the rotated GCIB source, while sputtering away the previously formed pillars and replacing them with small bubbles or blisters which seemed quite similar to artifacts of ion irradiation mentioned previously, such as the “nano drop” formation in FIB⁴⁴ or bombardment of inorganic surfaces using inert gases^{56, 57}. After four rotations (Figure 7b), the bubbles were seen to grow in size, possibly due to argon incorporation in the subsurface region, but the height of the pillars remained the same. This same trend was observed after twelve rotations (Figure 7d), and showed that micro-pillars can be sputtered away by the cluster source from a different angle, suggesting that they are not materials of different composition but formed strictly from masked sputtering of some sort. Additionally, it seems the presence of bubbles minimally contributed to the non-uniform sputter yields, despite the roughness (root mean square) of the crater bottom increasing from R_rms = 0.4 μm to 3.5 μm from Figure 7a to 7d, respectively. This increase in roughness would be expected to degrade the depth resolution, which was analyzed in detail in the following section.

Depth resolution of embedded PS spheres.

It is expected that depth profiling with micron-sized topography will lead to a significant degradation of depth resolution. In this report, the top and bottom edges of the PS spheres (or leading and trailing edges, respectively) were used to monitor the depth resolution, or apparent interface sharpness (given a non-planar interface), to more accurately assess the effect of micrometer topography on the distortion of 3D images. The sharpness of these interfaces were measured from the linescan of the yz slices of the spheres (such as the one shown in Figure 2), but for the rotated samples, the yz slices (i.e., datasets) needed to be joined together since the acquisition had to be stopped every time before manually rotating the sample. An example of a joined dataset can be seen in Figure S4, Supporting Information. The figure also includes the anticorrelating gelatin images to show that the spheres were completely sputtered away.

Figure 8 compares the interface sharpness at the leading and trailing edges, with and without sample rotation. The figure outlines the relationship between interface sharpness with sputtered depth, at the same time revealing interesting properties about the system. For example, the leading-edge sharpness was obtained as deep as 40 μm due to the spheres being embedded at different depths throughout the film. On the other hand, the trailing-edge sharpness could be obtained as early as 15 μm, a discrepancy caused by the (43 ± 5) μm spheres partially protruding from the film surface and the depth calibration made using the crater depth. This has affected a few of the measured lengths of the particles along the z-axis and is something that needs to be addressed for particles of unknown size.

Figure 8.

Plot showing the depth resolution (μm) as a function of sputtered depth (μm) for PS spheres embedded in gelatin. Depth resolutions (16%−84%) at the leading-edge are represented by circles, with rotation (○) and without rotation (●), while squares represent depth resolutions at the trailing edge, with rotation (□) and without rotation (■). Each datapoint represents an individual measurement.

Interestingly, the leading-edge sharpness was not seen to change much with depth, averaging around (5.8 ± 1.5) μm for both the rotated and non-rotated samples. This was an indication that the crater bottom was already sufficiently rough towards the beginning of the depth profile and any subsequent increase in number or size of micro-pillars with sputtered depth were not responsible for significantly affecting the interface sharpness. There did seem to be a trend downward (i.e., better depth resolution) for the non-rotated sample, but this was most likely caused by the decreasing sputter yield. Since the depth profile was constructed assuming a linear yield, a lowering of the yield towards the tail-end of a profile would give the false perception of a sharper interface. And this was supported by a corresponding increase in trailing-edge interface sharpness for the non-rotated sample. Conversely, the rotated sample displayed a constant sputter rate and the trailing-edge sharpness remained constant at around (7.1 ± 1.6) μm.

At this time, it is not yet clear whether fewer or more frequent staggered rotations would lead to better depth profiles in terms of crater bottom topography, depth resolution, and linearity of sputter yield for even deeper craters. Given the large sputter dose applied per cycle, it might also be better to use continuous rotation, although this may complicate things by requiring the sample to be centered on the rotational axis and having to align the images during post processing. However, one of the benefits of continuous rotation for deep profiles is being able to circumvent the formation of a wedge-shaped crater,¹¹ which can prevent the redeposition of the sputtered flux onto the crater bottom due to the low take-off angle imposed by gas cluster ion sources²¹, and potentially prevent further degradation of depth resolution.

Conclusions

For deep depth profiles of organic films using the gas cluster ion source (> 10 μm of sputtered depth), micron-scale topography was seen to form rather quickly and was shown to significantly affect both the linearity of the sputter yield and depth resolution. Although a step-wise, staggered sample rotation was shown to reduce the formation of micro-pillars and preserve the sputter yield, topography on the crater bottom still remained in the form of micron-scale blisters which contributed to severely degraded depth resolutions. The use of polymer spheres, which could be sputtered away when embedded inside an organic matrix, showed that it was possible to consistently measure the sharpness of the interface as a function of sputtered depth and to observe when possible distortions in the 3D image occurred. In this way, it was possible to quantitatively measure the effect of sample rotation on the quality of a depth profile, similar to delta layers for thin films.

Although trying to elucidate the mechanism of micron-scale topography generation was outside the scope of this investigation, it was noted that samples with different morphologies such as films and spheres could display similar types of micro-structures, and different types of polymers such as gelatin and PET could show different structures at the crater bottom. The formation of blisters also may be due to the incorporation of argon into the matrix. Therefore, different systems could behave differently, and techniques used in this investigation may not apply to all systems. Future investigations will examine other film materials, and probe even deeper into ultrathick films to evaluate the effect of rotation on depth resolution.