Gas-cluster ion sputtering: Effect on organic layer morphology

Abstract

Analysis of the surface of thin Irganox 1010 films before and after sputtering with an argon gas-cluster ion beam was performed with AFM and XPS to determine the effect that Zalar rotation has on the chemistry and morphology of the surface. The analysis is based on the change in roughness of the surface by comparing the same location on the surface before and after sputtering. The ion beam used was an ${\mathrm{Ar}}_n^{+}$ of size n = 1000 and energy 4 keV. The XPS analysis agreed with previous results in which the ion beam did not cause measurable accumulation of damaged material. Based on the AFM results, the Irganox 1010 surface became rougher as a result of ion sputtering, and the degree of roughening was quantified, as was the sputter rate. Furthermore, Zalar rotation during ion sputtering did not have a significant effect on surface roughening, surprisingly.

I. INTRODUCTION

Gas-cluster ion sputtering (GCIS) has allowed many instances of soft and sensitive samples to be cleaned in vacuum without leaving behind a damaged chemical structure due to ion bombardment.¹ This new method has been applied to a wide variety of systems including polymers,^2,3 biological samples such as rat brain tissue,⁴ synthetic soil,⁵ and paint sample cross sections,⁶ all of which are highly sensitive to sputter-induced chemical damage from conventional monatomic ion sputtering. Although GCIS has been able to nondestructively depth profile and clean these and other “soft” samples, there is still the possibility of GCIS-induced morphology change to the sample.

To date there has been a wide variety of work, both theoretical and experimental, on “soft-sputtering,” as GCIS is now commonly called. Molecular dynamics simulations have been used to model the interactions of gas clusters with surfaces,^7,8 forming a theoretical framework on which to build our understanding of experimental observations. Major insights point to a comparable damage cross section relative to the sputter cross section, such that, in simple terms, all damaged upper-layer material is also removed, leaving behind an undamaged new surface layer. With negligible primary-ion penetration to deeper layers, damage accumulation ahead of the newly exposed surface layer is also eliminated.

Experimental soft-sputtering research has focused on understanding sputter yields of various materials, including what has now emerged as the soft-sputtering standard Irganox 1010,⁹ industrially common polymers such as polypropylene¹⁰ and polystyrene,¹¹ and mixed polymers.¹² A common and logical consideration in these studies is to understand general trends in sputter yield, and to establish a “universal sputtering curve.”^3,7 To understand what can affect the sputter rate, multiple instrumental parameters such as incident angle,¹³ cluster size and energy,³ and even comparing argon-based clusters to other compositions of gas species have been studied.¹⁴

There has been less work on understanding how GCIS changes the morphology of the samples. Irganox 1010 has been used as a standard material for soft-sputtering and to establish GCIS rates,¹⁵ yet there has been little work published to understand how the surface structure is changed by cluster sputtering. Sputter-induced changes in morphology have been demonstrated in mixed polymer blends,¹⁶ on copper surfaces,¹⁷ with choice of ion species.^18,19 These first studies have shown that sputtering by some cluster sources can increase sample roughness.

Aside from soft-sputtering, a wide variety of methods to intentionally change polymer and organic layer surface roughness has been attempted, including solvents,²⁰ hot press,²¹ graft treatments,²² plasma,²³ and abrasion.²⁴ Unlike some methods, soft-sputtering has the unique advantage of not significantly modifying the chemical composition of the organic surface. In this article, we study and quantify changes to the chemical environment of the Irganox 1010 surfaces as a result of argon GCIS with x-ray photoelectron spectroscopy (XPS), and the corresponding changes in the surface morphology by atomic force microscopy (AFM).

II. EXPERIMENT

All samples were prepared simultaneously by controlled thermal evaporation of Irganox 1010 [pentaerythritol tetrakis (3,5-di-tert-butyl-4-hydroxyhydrocinnamate)] (Sigma Aldrich) onto cleaned silicon wafers as described by Shard et al.⁹ Its structure is shown in Fig. 1(c). A tungsten evaporation boat containing the compound was heated to 180 ± 5 °C in an Edwards Auto360 thermal evaporator at 1 × 10⁻⁷ mbar. Approximately 650 nm layer of Irganox 1010 was deposited at a rate of 25.5 nm/min. Postdeposition ellipsometry using a Gaertner Scientific Corporation (Skokie, IL) model L117 single-wavelength ellipsometer, with He–Ne laser (λ = 632.8 nm) at an incident angle of 70°, was used to measure film thickness. AFM measurements confirmed the layer thickness. The AFM thickness measurements were based on methods used by Ton-That et al.²⁵ with some modifications. To determine film thickness, the AFM tip was engaged to the surface, then displaced further into the surface by 8.12 μm, then scanned laterally to scratch away the film with the slow-scan axis disabled, then similarly scratched again with the tip moving perpendicular to the previous scratch, forming the “+” pattern of removed material. The + pattern was used to determine if there was any damage to the silicon substrate by comparing the point where the horizontal and vertical lines crossed (scratched twice) to the scratched sections of the + where the two lines did not cross.

Fig. 1.

XPS spectra of the samples during treatment (a) carbon 1s spectra with atomic percent next to spectra, (b) oxygen 1s spectra, and (c) the chemical structure of Irganox 1010.

A Bruker (formerly Digital Instruments, Camarillo, CA, USA) Multimode IV AFM was later used to image the scratched samples. The same portion of the organic layer film was imaged by AFM before and after sputtering to directly compare the effect of GCIS. The same AFM tip was used to eliminate any possible tip-shape effects. The nominal radius of the tip was a maximum of 10 nm. Silicon tips were used in tapping mode, with tips having a nominal resonance frequency of 300 kHz (NCHV-A). The AFM images were collected with the tip moving at 0.275 Hz in the fast direction (1375 nm/s) and 0.00027 Hz in the slow direction (1.3 nm/s). Imaging of the same spot before and after sputtering was crucial so as to directly compare the individual morphological changes caused by GCIS.

The samples' chemical compositions were monitored while sputtering with a Thermo Scientific (East Grinstead, UK) K-Alpha⁺ XPS located in the Surface Analysis Facility at the University of Delaware. An x-ray spot size of 100 μm was used to reduce x-ray-induced damage, and a low-energy electron flood gun was used to stabilize sample charging. To further reduce possible effects of x-ray-induced damage, a new XPS analysis location, not previously exposed to x-rays, but still within the sputter area, was analyzed each sputter cycle.¹¹ The x-ray source was the K-alpha of aluminum at $h\nu =1486.7\mathrm{eV}$. There was no elemental contamination found by XPS at any point in the sample preparation and treatment.

The energy of the argon-cluster sputter gun was set to 4 keV, with the cluster size set to nominally 1000 argon atoms, resulting in an elliptical spot profile of 300 ± 100 μm at the sample, the ion source was positioned to be 60° to sample normal. The sputter beam was rastered over a 1.85 mm by 3.69 mm area, in a nonsynchronous square wave. The sample was thinned by sputtering to a thickness of approximately 200 nm, determined by ellipsometry and confirmed by AFM scratching. The AFM was also used to observe finer effects of the gas-cluster beam. Some samples were held stationary during sputtering, while Zalar rotation²⁶ was employed with a rotation rate of 0.04 Hz on other samples. The raster rate of the sputter gun was 333 Hz in the short direction and 3 Hz in the long direction, with 300 line scans per raster area. The sputter rate was determined to be 35.9 ± 0.8 nm³ per ion and in close agreement with 36.9 nm³ per ion from previous studies of Irganox 1010 samples made in the same way.^9,15 Care was used to ensure that the sample was positioned over the center of rotation, specifically the area that was used to determine topographical changes.

Irganox 1010 is known to have a relatively high glass-transition temperature of 50 °C²⁷ and previous studies²⁸ have acknowledged that Irganox 1010 can anneal at room temperature. To determine if the sample roughness was affected by the delay between the measurements and sputtering, the relaxation rate of the surface was found to be 5 × 10⁻⁶ ΔRMS (nm)/min under ambient conditions, a negligible rate for the purposes of these studies wherein the samples were measured by AFM within hours of sputtering. This was determined by measuring the roughness over several weeks.

III. RESULTS AND DISCUSSION

A. XPS

Figure 1(a) shows the carbon 1s XPS spectra for each step of the process of making and modifying the Irganox 1010 films. The solid spectrum, shown for reference, is of the pure Irganox 1010 powder. The dash line spectrum down is of an Irganox 1010 film prior to sputtering. The dot line spectrum is the same film after sputtering. Each spectrum was energy calibrated to methylene carbon at 284.6 eV. As shown in Table I, and by visual inspection, there is no significant change to the line shape after any step of the film preparation or modification. It has been shown that ester carbon (289.0 eV) is most susceptible to sputter-induced damage,²⁹ yet it remained unchanged throughout the experiments (average 4.3 ± 0.4% n = 3). The intensity of the π-π* shakeup does not change significantly with sample preparation (average 1.1 ± 0.2% n = 3). Figure 1(b) shows the corresponding oxygen 1s XPS spectra. Like the carbon 1s spectra, the atomic percentages of oxygen vary only slightly between the samples and are in reasonable agreement with the bulk chemical structure shown in Fig. 1(c). The bulk atomic percentages are 85.9% carbon and 14.1% oxygen. The oxygen 1s peak for each step of sample preparation exhibits no significant changes in the line shape, and was an unresolved triplet for the three types of oxygen present: O–C=O at approximately 532.0 eV; C–OH at approximately 532.7 eV; and O–C=O at approximately 533.4 eV.

Table I.

Carbon 1s chemical state relative concentrations (all values have a 1% uncertainty).

Sample	$\mathrm{C}-\underset{\_}{\mathrm{C}}-\mathrm{C}$ (284.6 eV)	$\mathrm{C}-\underset{\_}{\mathrm{C}}-\mathrm{O}$ (286.1 eV)	$\mathrm{O}-\underset{\_}{\mathrm{C}}=\mathrm{O}$ (289.0 eV)	$\pi -{\pi }^{\ast }$ shakeup (290.8 eV)
Powder	80.1%	14.5%	4.4%	1.0%
Film	79.4%	14.8%	4.8%	1.0%
Film postsputter	81.6%	13.2%	3.8%	1.4%
Bulk values	83.5%	11.0%	5.5%	—

The XPS carbon 1s peak envelope was fitted with four chemical components: 284.6 eV for methylene, methyl and aromatic carbon (C–C–C); 286.1 eV for the phenol and ether carbons (C–C–O); 289.0 eV for the ester carbon (C–C=O); and finally 290.8 eV for the π−π* shakeup. Within experimental error, the relative concentrations of these carbon chemical states were unchanged by sample preparation or sputtering, as shown in Table I and Figs. 1(a) and 1(b).

Figure 2 shows an example fitted carbon 1s spectrum of the powder sample. Four component peaks, as shown in Fig. 2, are labeled from left to right as the $\pi -{\pi }^{\ast }$ shakeup, the ester carbon (C–C=O), the ether/phenol carbon (C–C–O), and finally the methyl/methylene carbon (C–C–C). The phenol/ether component peak full-width-at-half-max (FWHM) was allowed to float but was constrained to be the same as that of the methyl/methylene component peak (≤1.5 eV). The ester component was left unconstrained and had an FWHM of 1.0 eV. The $\pi -{\pi }^{\ast }$ shakeup component was also left unconstrained and had an FWHM of 1.8 eV. We chose not to include a component for the so-called β-carbon (C–CO₂). The positions of the component peaks were left unconstrained for all of the fitting attempts. All spectra were fitted with a Shirley background averaged over five data points, all peaks are fit with 80:20 Gaussian:Lorentzian peak mixing.

Fig. 2.

Fitted XPS spectra of the carbon 1s region of the Irganox 1010 powder.

There was no appreciable chemical damage to the organic layer due to sputtering as evidenced by the consistency of XPS spectra. Although not shown here, the evaluation of XPS difference spectra indicates a reduction of FWHM by ∼5% after sputtering. This change is attributed to the adventitious carbon present on the pre-sputtered samples, later removed by argon-cluster sputtering.

B. AFM

To ensure the sputtered position was tracked carefully, the sample had a macropattern etched into the sample by hand, establishing distinct orientation on the sample surface. As seen in Fig. 3, microfeatures were then used to ensure that the position was maintained before and after sputtering. Figures 3(a)–3(d) show the same area with increasing magnification. Figure 4 depicts the changes in the samples due to sputtering and Zalar rotation. Figures 4(a) and 4(c) are AFM images of the organic layer surfaces prior to sputtering, showing a typical RMS roughness of 0.3–0.4 nm. Figure 4(b) was obtained in the same spot as in Fig. 4(a), after being sputtered while not being rotated, showing a typical threefold increase in RMS roughness. Figure 4(c) is the AFM image of the same sample and spot as in Fig. 4(d), obtained following sputtering while being rotated (Zalar rotation). The effect of Zalar rotation is not significantly different from nonrotated samples.

Fig. 3.

AFM images of thin films of Irganox 1010 (a) 50 × 50 μm², (b) 20 × 20 μm², (c) 5 × 5 μm², and (d) 2 × 2 μm² collection area. The black arrows and squares show the subset that was scanned with finer resolution. The vertical scale is constant through all images.

Fig. 4.

AFM images of the thin films before [(a), (c)] sputtering, (b) after sputtering without rotation, and (d) after sputtering with rotation (Zalar rotation). Image positions were maintained such that (a) and (b) are the same and (c) and (d) are the same areas. The vertical color scale is consistent to all images.

Figure 5 shows distributions of pixel heights in representative AFM images. The distributions are in groups of two per plot, allowing the AFM images of Fig. 4 to be compared. Figure 5 shows the distributions of the pixel heights of the nonrotated (a) and rotated (b) samples prior to and after sputtering, showing the general effect of increased RMS roughness upon sputtering, with an insignificant difference due to Zalar rotation. This is further illustrated in Fig. 5(c), showing a comparison of samples rotated and nonrotated after sputtering. The broadening of the pixel height distribution is a clear indication that the samples became rougher due to sputtering even if Zalar rotation was used. Any change in surface roughness by sputtering is significantly less than the depth resolution that is typically achieved by either XPS (10 nm) or ToF-SIMS (6 nm);¹⁵ therefore, the increase in roughness would not affect the depth resolution of depth profiles.

Fig. 5.

Zero-centered pixel height distribution comparisons of AFM images from Fig. 4, showing before and after sputtering a nonrotated sample (a); before and after sputtering a rotated sample (b); and after sputtering a rotated and nonrotated sample (c). The RMS values in Fig. 4 are the same as the standard deviations of the distributions in this figure.

The RMS roughness of the surface was increased from 0.30 ± 0.05 nm (n = 3) to 0.89 ± 0.06 nm (n = 3) as a result of sputtering for the nonrotated samples, and from 0.40 ± 0.05 nm (n = 3) to 0.99 ± 0.05 nm (n = 3) if the samples were rotated during sputtering. The change in the root mean squared deviation of pixel height is shown in Figs. 5(a) and 5(b), and there is a clear development of craters within the sputtered area as seen in Figs. 4(b) and 4(d).

The general postsputter topographies seen in Figs. 4(b) and 4(d), nonrotated (4b), and Zalar rotation (4d) are similar to those found in plasma sputtering using corona discharge with common polymers,^30,31 as well as with plasma treatment with various gases.³² This shows that there might be a common pathway to morphological changes for reactive-ion etching, as described by Greenwood;³³ ion bombardment, as described by Yamada et al.;³⁴ and for cluster ion sputtering of organic layers, as shown here. Recent studies have shown³⁵ that GCIS can produce a classic example of ridge and valley structure that is common with hard ion sputtering;³⁶ we attribute the absence of these features to the low sputter yield but more work is required to prove it.

IV. SUMMARY AND CONCLUSIONS

Gas-cluster ion sputtering with ${\mathrm{Ar}}_n^{+}$ (n = 1000, 4.0 keV) causes no discernible degradation of Irganox 1010 films from a chemical perspective but does modify the surface roughness in a significant way. GCIS soft-sputtering was able to thin samples of Irganox 1010 by 450 nm causing the RMS roughness to change by approximately threefold. Zalar rotation had no significant effect on decreasing the morphologic changes caused by GCIS.