Macromolecular Sample Sputtering by Large Ar and CH4 Clusters

Sep 8, 2015 - This article reports the latest developments of our theoretical studies of gas cluster bombardment of model macromolecular samples using...
0 downloads 0 Views 8MB Size
Article pubs.acs.org/JPCC

Macromolecular Sample Sputtering by Large Ar and CH4 Clusters: Elucidating Chain Size and Projectile Effects with Molecular Dynamics Arnaud Delcorte* and Mathieu Debongnie Bio & Soft Matter (BSMA), Institut de la Matière Condensée et des Nanosciences (IMCN), Université Catholique de Louvain, 1 Croix du Sud box L7.04.01, B-1348 Louvain-la-Neuve, Belgium ABSTRACT: This article reports the latest developments of our theoretical studies of gas cluster bombardment of model macromolecular samples using molecular dynamics simulations. Here, we perform a detailed comparison of the effects of the sample molecular weight, the Ar cluster incidence angle (45° vs 0°), and the cluster nature (CH4 vs Ar) on the soft sputtering of polymeric samples. The results of Ar cluster-induced sputtering and fragmentation at 45° incidence for molecular targets with three different molecular weights (282, 1388, and 14002 amu) indicate a pronounced influence of that parameter beyond 1000 amu, which is explained by the extra energy needed to form fragments from longer chains and to overcome mechanical entanglement. An excellent agreement is found between the computed statistics of sputtering and the available experimental data for similar molecular weights. The variance of the sputtering and polymer fragmentation results with changing beam parameters is explained via the microscopic analysis of the interaction in the simulations. The influential physical quantities are identified, namely, the energy (density) deposited in the impact region, the projectile velocity, and the geometry of the impact. The lower sputtering efficiency of CH4 molecular clusters results mainly from the extra energy spent in covalent bondbreaking and vibrational excitation of the cluster constituents.



concept of desorption, will be called “soft sputtering” in the remainder of this article. The main aspects of cluster SIMS of organics and polymers were reviewed in 2013.27 To shed light on the physical processes of large cluster bombardment of surfaces, theoretical studies using molecular dynamics (MD) simulations have proven to be very useful. The main results have been discussed in review articles.28−30 In particular, several studies were devoted to the explanation of the effects of the cluster energy and nuclearity in the case of Ar cluster bombardment of various solids.31−35 The effect of the incidence angle on the sputtering of organic solids has been investigated to some extent by Czerwinski et al. for small molecules (solid benzene) and two different Ar cluster sizes (Ar366 and Ar2953).36 Those authors analyzed the difference between small and large cluster sputtering, with the former exhibiting a plateau between 0° and 45°, while the latter showed a pronounced maximum around 45°. The observed trends were explained by energy deposition in the surface region and angle-dependent blocking of the sputtered flux by the large projectile. It appeared later that these effects were both influenced by the mass of the projectile atoms.37 Recently, Postawa et al. also studied the sputtering of organic solids with two different molecular weights, n-octane (114 amu) and β-

INTRODUCTION Organic surface analysis based on desorption induced by large light-element clusters dates back to the work of Mahoney et al. in the 1990s, when they used massive supersonic water/glycerol clusters to produce secondary ion mass spectra of peptides.1 The massive cluster impact approach was later developed into a range of techniques, such as desorption electrospray ionization and related techniques2,3 and electrospray droplet impact,4 operating with water or alcohol clusters at atmospheric pressure and in a vacuum, respectively. At the other end of the spectrum in terms of elemental mass, massive gold clusters, such as Au400, were shown to stimulate the emission of large numbers of phenylalanine and biomolecular ions.5,6 In an intermediate mass range, beams of large argon clusters were originally developed for materials modification and smoothing,7,8 but their ability to desorb intact molecules9−11 and erode samples with minimal degradation of the subsurface12,13 rapidly attracted the attention of the community of secondary ion mass spectrometry (SIMS). Today, large gas clusters have become a standard for organic sample depth profiling in SIMS.14−16 With projectiles such as 5−10 keV Ar500−5000, it is possible to softly etch the most sensitive organic surfaces, including biological materials17,18 and organic electronics multilayers.19−21 In particular, they proved to be successful where other molecular projectiles, such as SF5+ 22 and C60+,23,24 were showing their limits.25,26 This situation of sputtering with minimal molecular degradation and fragmentation, closer to the © 2015 American Chemical Society

Received: July 20, 2015 Revised: September 7, 2015 Published: September 8, 2015 25868

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

identify the underlying mechanisms in the simulations of polymer sputtering. In this paper, new simulations are presented to clarify the effects of the polymer molecular weight, the nature of the projectile, and its incidence angle on the sputtering yields. First, the effect of the sample molecular weight on the sputtering is studied in detail, using three different molecular weights (282, 1388, and 14002 amu), and comparisons with recent experiments39 conducted on a series of polystyrene with increasing molecular weights are reported. Second, the data obtained with 0° and 45° incidence are compared for the solid made of intermediate chain sizes (1388 amu), in order to better understand the angular effect in such amorphous molecular samples. We show that taking the angular effect into account alleviates the former discrepancy between experiments and simulations. The yield difference between 0° and 45° bombardment is explained, at least in the first-order approximation, by the different quantities of energy placed in the surface region. Finally, the comparison of Ar cluster bombardment with methane bombardment of polymers for similar projectile masses, energies, and incidence angle allows us to analyze the mechanistic differences in their interactions with the solid and to explain some recent experimental observations.

carotene (537 amu), and they found almost no difference of sputtered quantities.38 This observation seemed at odds with some of the experimental SIMS literature39 and with our own simulations comparing kDa chains with virtually infinite polymers.40 Therefore, we felt that this point required complementary investigation with intermediate size samples. Concerning the effect of the cluster nature on the sputtering, although the mechanistic difference between heavy element and light element cluster impacts was studied to some extent in the past,30,41−45 it is not clear whether the exact nature of the cluster significantly influences the sputtering dynamics in the case of relatively light atomic (e.g., Ne, Ar) and molecular (e.g., O2, H2O, CH4) gases. The elemental mass effect in atomic noble gas clusters was already pointed out,37 but detailed comparisons with molecular clusters are still missing. Anders and Urbassek studied the influence of the intramolecular bonding on the impact and sputtering induced by molecular O2 clusters with an oxygen ice solid, by artificially varying the O− O bond strength.46 They highlighted the importance of the energy transfer in the vibrational modes of the projectile and target, reducing the sputtering yield, but did not focus distinctly on the projectile effect, since the properties (namely the O2 dissociation energy) of both projectile and target materials were modified at the same time. The interest in methane clusters, selected for the present study, springs from recent experiments where they were used, as well as H2O clusters, in the hope of improving ionization in SIMS.47−49 From the fundamental viewpoint, they allow us to compare atomic and molecular clusters, the latter having more degrees of freedom and channels to dissipate the impact energy. One point of interest is to find out whether molecular desorption might be softer using methane clusters. In the cluster sputtering literature, it has become customary to scale the sputter yields (sputtering yield volumes, Yv, or sputtered masses, Ym) by the number of atoms (or the mass) of the projectile and to plot them as a function of the projectile energy scaled in the same manner.31,37,50 In doing so, Seah51 and Cumpson52 could show that the sputter yields of many organic and inorganic materials bombarded by argon clusters of various sizes and energies were merging on a so-called “universal” curve for each considered material. A similar trend had been previously found in the simulations of Ar cluster bombardment of solid Ar31 and organic cluster bombardment of polymers.50 The matter of universality in the sputtering yields was discussed in detail by Paruch et al., who proposed a new type of scaling.53 Dividing the energy by a parameter U0, defined as the cohesive energy per atom in the solid, those authors could reduce significantly the data spread between the sputtering yields of metallic and organic solids. An in-depth analysis of the same results allowed the authors to extract mechanistic similarities of the sputtering of different systems for the same bombardment conditions in this new scaling framework.54 In our former papers on Ar cluster-induced sputtering, only normal incidence bombardment was computed using MD simulations.40,43 Even though the agreement with the experimental “universal curves” obtained at 45° was reasonable for kDa molecule sputtering, a discrepancy was found in the low-energy range. It was then speculated that the projectile incidence angle could influence the nonlinear region of the universal curves and the energy per mass unit at which the transition occurs. This parameter was only taken into account recently in the fits of the experimental sputtering curves,55 and we felt that it was important to check its effect and to clearly



COMPUTATIONAL DETAILS The MD simulation program used in this study is the SPUT code developed by the group of B. J. Garrison at Penn State University for sputtering applications.28 In the classical MD method, Hamilton’s equations of motion are numerically integrated over some time interval, providing us with the position and velocity of each particle at each time step. Forces among the atoms or particles in the system are derived from empirical interaction potentials. The polymeric targets used in this study and the methodology of coarse-graining (CG) have been described in detail in previous articles.50,56,57 They were made of united atoms of CH2 (14 amu) and CH3 (15 amu), in order to reduce computational expense. Three different chain lengths were compared in the present study, 20, 99, and 1000 united (CH2, CH3) atoms, corresponding to molecular weights of 282, 1388, and 14002 amu, respectively. The samples will be named PE282, PE1388, and PE14002 in the rest of the article. The interaction potentials were Lennard-Jones potential functions for PE−PE intermolecular (nonbonding) interactions and Morse potential functions for the intramolecular (covalent) interactions in the PE chains. Additional details about the sample potential parameters can be found in ref 58. The targets were fabricated by replicating a small amorphous cell of ∼3 × 3 × 3 nm3 in the three directions of space, in order to obtain systems that are sufficiently large to accommodate 10 keV cluster impacts.56 Because of the simplicity of the chosen approach and potentials, the resulting samples should be considered as generic models of amorphous organic samples made of linear macromolecules rather than strictly polyethylene. As was the case in our previous studies, the Ar clusters were modeled using Lennard-Jones potentials splined to the KrC repulsive potential at short interdistance.35 In order to investigate effects related to the molecular nature of the projectile, the CH4 clusters were defined at the atomistic level and their interactions modeled using the AIREBO potential for the C−C and C−H interactions, with hydrogen (1H) replaced 25869

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

Table 1. Coefficient A (amu/eV) in the Linear Regression of the Scaled Sputtering Curves, Ym/m = A(E/m)

by tritium (3H) for computational efficiency.59 A purely repulsive Molière potential was used to describe the Ar-PE interactions. A weak Lennard-Jones potential, appended to a Molière repulsive wall at short interdistance, was used between the C,H atoms of the projectiles and the CG particles of the PE samples.50 All the targets were bombarded with 45° incidence and one of them, PE1388, additionally with 0° incidence. The trajectories were run for a minimum of 25 ps, up to saturation of the sputtered flux. The targets were surrounded by a zone of rigid atoms and a Langevin heat bath region to prevent pressure waves generated by the cluster projectile impacts from reflecting off the system boundaries and to keep the sample at the required temperature of 0 K.

argon clusters PE282 (45°) PE1388 (0°) PE1388 (45°) PE14002 (45°) methane clusters PE1388 (45°)

total

intact

fragments

8.4 6.0 7.6 4.6 3.8

6.0 2.9 3.8 − 1.4

2.4 3.1 3.8 − 2.4

also various sample configurations, would be very expensive in terms of computer time. In the linear region, the fit of the PE282 data indicates a 10% larger sputtered mass than that of the reference PE1388, which is a small difference for a molecular weight that varies by a factor of ∼5 (Table 1). In addition, it is obvious from the graph that several points of the two sets of data are superimposed, especially in the high energy range. The high-energy part of the PE14002 data is sufficiently defined to observe that the yield for PE14002 is lower than that of PE1388 for most of the points, but due to the limited statistics, the difference reported in Table 1 should be considered with caution. To our knowledge, there are no data in the literature concerning the sputtering of small polyethylene chains by Ar clusters. However, as was explained in the methods section, the simulated molecular solids should be considered rather as generic models for amorphous polymers rather than accurate models for specific polymers (coarse-graining approximation, simple potentials). Therefore, we feel free to compare the simulation results to sputtering data published in the literature for other amorphous polymers and even molecular samples, as long as specific degradation or structural effects have not been observed for these organic materials upon Ar cluster bombardment. In Figure 1, along with the simulation data, the experimental data obtained in our laboratory for polystyrene (PS) oligomer standard samples with a range of MWs (from 1110 to 139 000 amu) are reported for comparison purpose.39 The incidence angle is 45° in the experiments, as in the simulations. The range of energy per amu used in the experiments is limited, but it covers the most interesting part of the evolution, where the break between the nonlinear and linear regions occurs. The effect of the sample molecular weight is significant and, in addition to the general trend of the curves, the range of yields spanned by the experimental data is the same as the one described by the simulations. In particular, the curves of PS1110 and PS1920 are very close to the one of PE1388 while the blue curve of PS10000 tends to fit the one of PE14002. Though this comparison can only be qualitative, the agreement between the model and the experiment is quite impressive and leads to two conclusions: (i) the model describes the main features of the sputter yields of organic samples, including the position of the break between linearity and nonlinearity, and the yields are similar; (ii) the molecular weight effect is predicted. Interestingly, the model also predicts only a small difference between the sputter yields of PE282 and PE1388, which suggests that the variation below a MW of 1000 amu becomes almost negligible. The calculated intermolecular binding energies per CG unit (CH2 or CH3) are 0.098, 0.091, and 0.078 eV for PE282, PE1388, and PE14002, respectively. The significantly lower binding energy reported for PE14002 is attributed to the length of the chains that tends to prevent the most efficient packing.



RESULTS AND DISCUSSION Effect of the Macromolecule Size. The sputter yields, expressed here in terms of sputtered masses, Ym, have been calculated for polyethylene oligomer samples with three molecular weights (MW = 282, 1388, and 14002 amu), and a range of Ar cluster projectiles (Ar18, Ar60, Ar250, Ar500, Ar1000, Ar1700, Ar3000, Ar5000) and energies (1−15 keV). The data corresponding to a 45° incidence angle are gathered in Figure 1. As has become customary in the literature, the data are

Figure 1. Calculated dependence of the sputtered mass per projectile mass unit on the kinetic energy per mass unit for PE282, PE1388, and PE14002 (45° incidence). The colored lines correspond to the experimental data of the sputtering of PS oligomer thin films with different molecular weights bombarded by 10 keV Arn+ clusters (n = 1300, 1698, 2778, 5000, and 6549).39

presented in terms of scaled intensity and energy, that is, sputtered mass per projectile mass unit versus energy per projectile mass unit. The shape of the obtained dependences is similar for the two lower MWs, with a linear part at high energy and a break in the curve around 0.1 eV/amu (as a reference point, 5 keV Ar1000 is 0.125 eV/amu). The data sets were fit using a linear regression of the form Ym/m = A(E/m) (0 intercept) beyond 0.1 eV/amu, and the values of A are collected in Table 1. The shape of the data is less defined for the higher MW, with sometimes large deviations from linearity at high energy and virtually no sputtering below 0.1 eV/amu. At low energy, the limited statistics is a serious issue, and a better definition of the curve, requiring more impacts but probably 25870

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

energies, in the experimental data, with a gradual decrease between PS1110 and PS61800, and a saturation beyond that MW. It is also useful to analyze the reasons behind the similar sputtered masses calculated for PE282 and PE1388. First, one can observe that the size of PE1388, around 1 nm3, is still very small in comparison with the size of the interaction volume, which should ease the desorption process. In addition, the molecules are not strongly entangled, so that desorbing one PE1388 molecule is not more energy consuming than sputtering a chunk of five PE282 molecules. For the rest, the two materials have similar physical properties and a slightly different density (0.9 kg/dm3 for PE282 versus 1.01 for PE1388). In conclusion, increased molecular weights increase the mechanical entanglement of the chains and the number of covalent bonds to be broken to release polymer material, two effects that become important only when the overall molecular sizes become significant with respect to the size of the crater (>1000 amu). Quantifying these effects in terms of a single characteristic energy that would be physically meaningful (a variation of U0 as proposed in ref 53) is not straightforward. The analysis of the fractions of molecules sputtered intact or as fragments provides a more detailed view of the emission and fragmentation induced in these two samples. Figure 3 shows

Since lower binding energies should cause higher emission yields, the computed energy differences cannot explain the emission yield decrease observed between PE1388 and PE14002. Nonetheless, this decrease can be explained by the analysis of the dynamics. The effect is illustrated in Figure 2 via

Figure 2. Microscopic views of the bombardment of PE14002 by 10 keV Ar250 (1.0 eV/amu) at 45° incidence.

snapshots of the MD for 10 keV Ar250 impinging on PE14002 with a 45° incidence. At this energy of 1 eV/amu, many bond scissions still occur in the sample, and a number of fragments are produced. However, while small fragments resulting from two bond scissions or from one bond scission near the chain end can be readily emitted, a single bond breaking in the middle of the chain only produces large chain segments that remain attached or entangled to the rest of the sample. This effect is inversely proportional to the chain size, which explains the reduction of the emission with increasing molecular weight. In rare occurrences, entire PE14002 molecules can be disentangled and sputtered intact, which leads to a step change in the sputtered mass and explains the erratic variations in the curve of Figure 1. In the particular case of Figure 2, two intact molecules are finally emitted together with a large number of fragments, which explains the unusually large sputtered mass for that event (data point with a green arrow in Figure 1). Such events should be very dependent on the exact structure of the sample and the projectile impact point, and they show the limits of our simulations with PE14002 in terms of establishing accurate sputter yields, which would require to run a larger number of simulations. At low energy per amu, when bondscissions become rare or absent, emission from the PE14002 sample is exceptional. This is mirrored by the leveling off of the PE14002 sputtered mass below 0.1 eV/amu, while the curves of PE282 and PE1388 continue to much lower energy values. The consequence of increasing molecular weight on the sputtered mass is described in greater detail, but over a small range of

Figure 3. Calculated dependence of the sputtered mass per projectile mass unit on the kinetic energy per mass unit for PE282 and PE1388 (45° incidence). The full symbols correspond to intact molecules and the open symbols to fragmented molecules.

that the total mass sputtered as intact molecules is, on average, 1.6 times larger for PE282 than for PE1388 (see also Table 1), while the mass sputtered in fragments is ∼1.6 times smaller for PE282. So, even though the total sputtered masses are almost similar, the manner in which the projectile energy is spent in the sample varies. One interesting question to elucidate is the relative importance of the energy used for bond-scissions in these samples. For instance, upon 10 keV Ar250 bombardment, 183 intact PE282 molecules and 258 fragments are sputtered. Assuming that every fragment formation requires two bondbreakings (each bond between two CG units is 3.6 eV), the total energy spent should be a maximum of 1.9 keV, i.e., ∼20% 25871

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C of the projectile energy. The same calculation done for PE1388 leads to a maximum value of 1.5 keV. The difference, less than 5% of the projectile energy, appears therefore to be marginal. For the reasons mentioned in the previous paragraph, and because the energy spent in covalent bond-scissions is far from dominant, it is natural that the total sputtered masses are close for PE282 and PE1388. A recent atomistic MD study of C60 bombardment of organic materials investigated the molecular size effect to some extent by comparing the sputtering of octane and β-carotene from their molecular solids.38 Octane is about two times smaller than PE282 while β-carotene is about 2 times larger. Despite the differences in the models and in the samples, a similar conclusion was drawn in their study, i.e., that the total sputter yield was comparable for the two samples at a given projectile energy and that the ratio of intact molecules to fragments was larger for the smaller molecules. This study, our simulations, and the experimental results suggest therefore that the MW effect occurs mainly in the range 1000−50 000 amu. Another detail provided by Figure 3 concerns the different shape of the intact molecules and fragment evolutions. While the linearity extends below 0.1 eV/amu for intact molecules, deviation from linearity already starts in the region 0.5−1 eV/ amu for the fragments, defining a region of soft sputtering which extends roughly between 0.02 and 0.4 eV/amu. Below 0.02 eV/amu, the yields per projectile amu decrease very rapidly, in a nonlinear fashion. Similar results were previously obtained at normal incidence,45 but these new simulations show that the energy window of soft sputtering is angle dependent. This dependence is investigated in the next section. Effect of the Incidence Angle. The scaled energy dependences of the sputtered mass of PE1388 upon 0° (normal) and 45° incidence are plotted together in Figure 4a. The two curves merge at high energy, but they diverge below 1 eV/amu. The experimental dependence of the sputtered mass of Irganox 1010 upon 45° Ar cluster beam incidence, taken from the data fit of Seah,55 is also represented (dashed gray line). Irganox 1010, with a MW of 1076 amu, shows a very similar energy-dependence than PE1388 for the same incidence angle. The calculated 0° dependence was shown and compared to experimental data obtained upon 45° incidence in a previous article, which highlighted the discrepancy in the low energy range.45 These new data demonstrate that the different evolutions at low energy were indeed due to the angular effect. The different scaling of the same data presented in Figure 4b will be useful for the mechanistic discussion at the end of the section. The angular dependence of sputtering upon Ar cluster bombardment has first been investigated theoretically by Czerwinski et al.36 The authors compared the angular distributions of benzene (MW = 78 amu) sputtered from a molecular sample by 14.75 keV Ar366 and Ar2953. The maximum observed at 45° for Ar2953, with much lower yields at 0°, contrasting with the plateau between 0° and 45° in the case of Ar366, is consistent with our calculations, because the lowenergy range in Figure 4a, where the divergence of the curves occurs, corresponds to Ar clusters with the largest sizes (Ar1000−5000). The ratio between the 45° and 0° calculated sputtered masses of PE1388 is shown in the inset of Figure 4b. While this ratio is close to one at high energy, it starts to increase below 1 eV/amu, mirroring the divergence observed in Figure 4a. This evolution could also be compared to recent measurements of

Figure 4. Calculated dependence of the sputtered mass of PE1388 at 0° and 45° incidence (a) per projectile mass unit, versus the kinetic energy per mass unit (the dashed gray line corresponds to Seah’s empirical equation fit to the experimental data of the Arn sputtering of Irganox 1010);51 (b) divided by the projectile kinetic energy, versus the projectile initial velocity. Clusters with the same nuclearity are connected by straight lines. The meaning of the symbols is the following: full diamonds: Ar18; open triangles: Ar60; full squares: Ar250; open diamonds: Ar500; full triangles: Ar1000; crosses: Ar1700; open circles: Ar3000; full circles: Ar5000. Inset of (a): Energy-dependence of the fraction of the mass sputtered as intact molecules. Inset of (b): Energy-dependence of the ratio of the masses sputtered under 45° and 0° incidence. The open circles and the dashed gray line correspond to Seah’s collected experimental data of the angle dependent Arn sputtering of Irganox 1010.55

the angular dependence of Irganox 1010 sputtering,55 symbolized by the gray circles and the dashed gray line in the inset of Figure 4b. Here again, the experimental results obtained on Irganox 1010 and the sputtered mass ratio calculated for PE1388 appear to be consistent, within the limits of the statistical error. The incidence angle also affects the energy-dependence of fragmentation, even though the effect is less obvious. The fraction of PE1388 sputtered intact is presented in the inset of Figure 4a, as a function of the scaled energy. The fluctuations are large above 0.3 eV/amu, with an intact fraction stabilizing between 0.4 and 0.6 beyond 1 eV/amu for both angles. There is less fluctuations in the low energy range, where the intact 25872

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

Figure 5. Time-evolution of the energy of impinging Arn clusters (divided by 1000) and of the energy deposited at different depths in the bombarded PE1388 sample. The initial energy of the projectiles is 10 keV. (a) Ar250, 0° incidence (1.0 eV/amu); (b) Ar250, 45° incidence; (c) Ar1700, 0° incidence (0.15 eV/amu); (d) Ar1700, 45° incidence.

break between the first and second regions is very clear at 45° incidence. The average velocity of the pressure wave developing in the solid upon impact was estimated from the movies of the trajectories for a series of cases. It was calculated from the time needed by the wave to reach the bottom of the sample. The results show that this velocity saturates around 9 km/s for highspeed projectiles (regions 2 and 3), but it decreases with the projectile velocity in region 1 (e.g., ∼2.5 km/s for 5 keV Ar5000), which physically substantiates the representation of the yields as a function of velocity. For low-velocity impacts, the material is able to accompany the motion induced by the projectile penetration, while for high-velocity impacts, the projectile is too fast, and the reaction of the material saturates in terms of speed. We observe that this region of saturation of the pressure wave velocity corresponds roughly to the region of saturation of the sputtered mass per unit of energy of the projectile. However, this argument does not explain the difference between the curves of the sputtered mass obtained at different angles, which confirms that other physical parameters and effects are also at play. As was mentioned before, previous articles investigating cluster bombardment of organic solids highlighted the importance of the energy distribution in the top surface region41,50 and of the blocking effect induced by large clusters upon normal incidence.36 In order to explain the calculated sputtered masses of PE1388 and their variations as a function of the Ar cluster size, energy, and angle, the energy deposition in the surface region was analyzed, in the same spirit as the methodology proposed by Garrison and co-workers in the MEDF model, which was successfully applied to predict the sputtering yields upon C60 and Au3 bombardment.41 First, the dynamics of energy deposition in the solid was analyzed by computing the energy deposited in horizontal slabs

fraction gradually tends toward unity. The value of 1 (no fragmentation) is reached at 0.2−0.3 eV/amu for 0° incidence and at ∼0.06 eV/amu for 45° incidence, illustrating the angle dependence of the “soft sputtering” energy region. A more detailed study of fragmentation, including C−H bond-breaking, would require the use of a full atomistic model,40 but the time expense would limit it to a few trajectories.60,61 In contrast with experiments, MD simulations allow us in principle to retrieve all the physical quantities of the investigated material at any time of the trajectories, in order to provide mechanistic explanations of the observations. In the remaining of this section, the physical reasons of the trends in the yields at 0° and 45° will be explored in detail. In Figure 4b, the sputtered masses were divided by the projectile energy, as suggested in ref 53. This representation allows us to see how efficiently the projectile energy is used when varying the cluster impact conditions. We also chose to plot the data as a function of the projectile velocity, because, as was discussed in ref 50, we speculate that the match or mismatch of the projectile and the material characteristic velocities and times are of importance in the explanation of the physics of the impact. As a reminder, the speed of sound in polyethylene is about 2 km/s at room temperature, and it increases with decreasing temperature, as the material contracts and atom interdistances become smaller. With this scaling, the data can also be conveniently represented in a linear plot. In Figure 4b, the two sets of data (0° and 45°) are separated, but they show similar evolutions, with a region of fast increase below 10 km/s, a saturation region between 10 and 30 km/s, and a region of slow decay above 30 km/s. The first region corresponds to the nonlinear part of the curve of Figure 4a, while the second and the third fit in the linear evolution beyond 0.1−1 eV/amu. Although the statistical variations appear to be larger in the linear representation, the 25873

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

where the data at 0° have systematically lower sputtered masses than the data at 45°, corresponds to the largest Ar clusters. This indicates the prevalence of the blocking effect identified in the article of Czerwinski et al.36 In the case of PE1388, 5 and 10 keV Ar1700 clusters still sputter some polymeric material at normal incidence, but larger clusters, such as Ar3000 and Ar5000, do not induce any sputtering. Energies beyond 10 keV were not tested with these clusters because the size of the simulation cell was not considered to be large enough. Below Ar1700, the blocking effect is of less importance in our simulations of polymer sputtering. At this point, it is useful to look at the microscopic views of the simulations and see what this surface skin of 4 nm represents with respect to the sputtered volume and the crater size. In Figure 7, the emitted molecules and fragments are traced back to their initial location in the solid, thereby showing the origin of the sputtered volume. For 10 keV Ar250 at normal incidence, the depth of the emitted volume is ∼3.2 nm and its width is 12.0 nm (Figure 7a,d). The depth is similar but the width somewhat larger at 45°, explaining the yield difference (Figure 7b,e). The emission depth is also equivalent to what was observed upon bombardment with a 10 keV C60 projectile under normal incidence.56 Finally, it is very close to the thickness of the surface slab used to establish the correlation of Figure 6. The situation changes drastically when the conditions are those of soft sputtering, i.e., when most of the emitted material is desorbed as intact molecules, without covalent bond scissions. It is illustrated in Figure 7c,f by the 7.5 keV Ar5000 bombardment of PE1388, with a 45° incidence (0.038 eV/ amu). Here, all the molecules constituting the sputtered flux originate from the topmost layer of the sample. They are the molecules that were protruding from the surface, and therefore those that have the weakest binding energy to the solid. Their ordered distribution mirrors the size of the (amorphous) unit cell reproduced in the three spatial dimensions to create the sample. It is obvious from this picture that if the molecules were significantly larger, extending toward the depth of the sample, and somewhat more entangled with their neighbors, they could not be desorbed from the sample with these bombardment conditions. This is the case of sample PE14002. Figure 7c,f might give the wrong impression that clusters with a low kinetic energy per mass unit simply desorb the topmost molecules without disturbing the surface of the solid. In Figure 8, side views taken from the movie of the same interaction (7.5 keV Ar5000 bombardment of PE1388 with 45° incidence) illustrate the importance of the molecular relocation in the sample even in the case of soft desorption. Each color in the solid indicates a different horizontal slab of 1 nm thickness. Note that what appears as fragments in the figure is actually pieces of intact molecules only partly represented because the sample is shown in cross-section. The final crater reaches a depth of ∼6 nm, and the crater edges are asymmetric as a result of the oblique incidence. The surface layers tend to be pushed into the sample at the back end of the crater and upward at the front end, forming a protruding rim that pushes molecules toward the vacuum. Figure 8c shows the extent of the molecular mixing and relocation induced by the impact. Although the emission is soft in the sense that there is minimal or no molecular fragmentation, the surface undergoes a pronounced rearrangement during the expansion of the crater, especially in the crater rim. The features seen in Figure 8c are

of a few nanometers of thickness. The average total energy (potential + kinetic) per CG unit in each slab was calculated and multiplied by the total number of CG units in a molecule, providing us with the average total energy per molecule remaining in the slab at each time. The division of the total energy in each slab by the number of CG “atoms” is justified by the fact that the number of particles in a certain depth range varies strongly with the development of the impact crater. The results are shown in Figure 5 for the cases of 10 keV Ar250 (1 eV/amu) and Ar1700 (0.15 eV/amu) at 0° and 45° incidence. In this example, the slab thickness is 4 nm. These plots demonstrate the effect of the incidence angle on the energy deposited in the sample. When going from 0° to 45°, the maximum energy transferred in the top 4 nm layer increases, while the energy deposited in the subsurface decreases. The blue curve, corresponding to the energy carried by the projectile atoms, shows that they retain a significant fraction of their initial energy when the incidence is 45° and that this fraction increases with the cluster size (1 keV for Ar1700). Upon massive cluster bombardment at 45°, a large number of Ar atoms are indeed backscattered in the vacuum. Then, we tested the simple hypothesis that the sputtered mass might be correlated to the energy deposited in the topmost layer. Different slab thicknesses were used, going from 3 to 5 nm, and the best correlation was found for a value of 4 nm (Figure 6). The correlation is not perfect (R2 = 0.87 for the

Figure 6. Correlation between the sputtered mass (absolute values) and the average energy per molecule in the top 4 nm of the surface of PE1388.

linear regression), but the important point here is that the two sets of data (0° and 45°) merge and can be reasonably fit with the same linear regression. Figure 6 shows indeed that, for the range of Ar angles, sizes, and energies considered, the sputtered mass is proportional to the energy transferred in the top 4 nm of the sample. The fit suggests the existence of a threshold, but the energy value is somewhat arbitrary since the energy is averaged over the entire slab, involving at the same time fastmoving atoms and molecules at rest, and an energy distribution shape that strongly varies with time. Although the two sets of data points are quite scattered and overlap, separate fits point to a higher threshold and a steeper slope for the 0° bombardment. The low surface energy range, 25874

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

Figure 7. Side and top views of the initial positions of the sputtered material (violet) and the projectile (red) for PE1388 bombarded by 10 keV Ar250 (1.0 eV/amu) at 0° incidence (a,d), 10 keV Ar250 at 45° incidence (b,e), and 7.5 keV Ar5000 (0.038 eV/amu) at 45° incidence (c,f).

been recently proposed as an alternative to noble gas clusters for improved ionization of the analyte in SIMS experiments.47−49 In order to gain a better understanding of the difference between such molecular gas clusters and argon clusters in terms of sputtering dynamics, induced damage, and emission, a series of simulations was computed using methane clusters (45° incidence) with comparable total masses as the argon employed so far. The scaled sputtered masses obtained with CH4 cluster bombardment are reported and compared to the Ar data in Figure 9. The curve shape is similar for CH4 and

Figure 9. Comparison of the dependences of the sputtered masses of PE1388 on the kinetic energy per mass unit for (CH4)n and Arn clusters (45° incidence). Inset: Energy-dependence of the fraction of the mass sputtered as intact molecules. The arrow indicates the data corresponding to 10 keV (CH4)420. Figure 8. Cross-sectional views of the bombardment of PE1388 by 7.5 keV Ar5000 (0.038 eV/amu) at 45° incidence. Each color defines a different layer of 1 nm thickness. The bottom of the sample and the projectile atoms are omitted for clarity.

for Ar clusters, with a linear segment beyond 0.1 eV/amu and a nonlinear part below that energy. Nevertheless, the scaled sputtered mass is lower for methane, about half of the values obtained for argon on average, as indicated by the linear regressions used to fit the data (Table 1). The explanation of this difference is investigated by comparing the manners in which the different clusters transform and spend their translational energy upon impact.

expected to continue to evolve and relax over much longer times.62 Sputtering with Methane Clusters. As was described in the Introduction, organic and water molecular clusters have 25875

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

polymer. Here it is important to stress that the vibrational energy is probably overestimated in the classical MD model.46 In real methane, the vibrational energy levels are quantized, so that transfer of energy quantities that are smaller than the energy of the first vibrationally excited level (0.16 eV) would be impossible. In contrast, the quantification of the rotational levels should never be limiting (0.001 eV). The black dashed curves correspond to argon. Figure 10a represents the total potential energy of the Ar clusters, i.e., the maximum energy that can be consumed in case of complete atomization of the cluster. Figure 10b shows the cluster translational energy remaining at the end of the bombardment event. These values indicate that, for Ar clusters, a maximum of ∼1 keV can be lost for sputtering. Between CH4 and Ar, the ratio of the remaining energy (useful for sputtering) is close to 3/4, which contributes to explain the lower sputtering yields observed with methane, but not quite quantitatively. Another effect, more difficult to quantify, could be related to the different energy densities deposited in the surface by the two clusters. Indeed, for a given mass, the CH4 clusters are significantly larger than the Ar clusters. The surface energy density is therefore smaller, which could influence the emission yields. The impacts of 10 keV Ar250 and 10 keV (CH4)420 (similar energy/amu) with PE1388 are compared in Figure 11. The sputtered masses of these two events differ by a factor of 1.7. At the same time, the horizontal cross-sectional area of the methane cluster is ∼70% larger than that of the argon cluster. Time-selected snapshots of the interaction (Figure 11a−f) essentially show that the crater induced by the methane cluster is smaller and more asymmetric than that resulting from the argon cluster impact. Accordingly, the images of the sputtered molecules drawn at their original positions before the impact (Figure 11a,d) indicate that the volume sputtered by methane is less deep and less extended laterally, also with some asymmetry, in contrast with the case of argon. In Figure 11g−j is shown the evolution of the projectile atom positions in the first 2.5 ps. The first observation is that the reflection and dispersion of the projectile in the lateral direction are faster and more complete for methane, while argon tends to stick together for a longer time, despite the smaller binding forces, which can be attributed to its higher elemental mass, resulting in less deviation upon collision in the sample. Argon penetrates deeper in the polymer than methane, as is also obvious from the figure. The more pronounced asymmetry noticed for the methane cluster-induced crater mirrors the motion of the projectile. Finally, the translational energy at the end of the trajectory (Figure 10b) quantitatively describes the faster reflection and higher velocities of the methane molecules and fragments after the impact. In comparison, argon implants more and, therefore, transfers more of its energy to the solid. To our knowledge, a study of the influence of the projected cluster energy density at the surface does not exist for this kind of clusters and targets. Nevertheless, analogies with other cases described in the literature can be made. First, it was shown that the penetration depth of heavy clusters is proportional to the cluster projected momentum, i.e., the momentum divided by the cross-sectional area of the cluster, which can be seen as a “momentum density”.63 The different projectile penetration depths and crater shapes for Ar and CH4 clusters are consistent with these observations. Second, Aoki et al. studied the effect of cluster density for 20 keV Ar2000 bombardment of silicon by MD and showed that “diluting” the cluster artificially (by

In Figure 10a are represented the curves of the different components of the internal energy as a function of the cluster

Figure 10. Projectile mass dependence of the internal energy increase (a), and the translational energy left in the projectile atoms (b), at the end of the trajectories, for 5 keV (CH4)n impacts on PE1388 (45° incidence). In (a), the total potential energy (PE) gained upon impact, green, is separated from the ro-vibrational kinetic energy (Ro-vib KE) of the CH4 molecules, red. For comparison, the total (initial) potential energy of the Arn clusters, is signified by the dashed black line (atomization energy). In (b), the dashed black line represents the translational energy of 5 keV Arn clusters at the end of the trajectories, for the same mass range.

mass at the end of the trajectory, for impacts with a total energy of 5 keV (45° incidence). The variation of the total potential energy (green) shows a maximum at small cluster mass, and then a decrease, followed by a sustained increase up to the largest cluster sizes. This variation includes the effect of covalent bond-breaking in the CH4 molecules, prevalent for the small cluster sizes, and the effects of the intermolecular bondscissions and the accumulation of potential energy in the vibrational modes. The kinetic energy of vibration and rotation (red) tends to decrease with increasing cluster size. The sum of the potential and intramolecular kinetic energies are in the range 550−1100 eV at the end of the event, i.e., up to >20% of the initial cluster energy. Finally the remaining center-of-mass (translational) energy, shown in Figure 10b, is between 800 and 1200 eV for the whole series of CH4 clusters. Roughly, the translational energy exhibits a maximum where the internal energy undergoes a minimum, around 6 × 104−7 × 104 amu ((CH4)2809). Overall, ∼2 keV of energy are either transferred in internal energy or kept as translational energy of the methane clusters at the end of the interaction, which means that ∼40% of the total initial energy is wasted for the sputtering of the 25876

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C

Figure 11. Cross-sectional views of the bombardment of PE1388 by (a−c) 10 keV Ar250 (1.0 eV/amu) and (d−f) (CH4)420 (0.99 eV/amu) under 45° incidence. The initial position of the sputtered material (violet) is also represented in (a) and (d). Separate side views restricted to the projectile atoms are shown in (g)−(j). Red is for argon and green for methane.

data, since the sample itself is coarse-grained in our simulations, and detailed conclusions about sample fragmentation would require a full atomistic description.

stretching it in the lateral dimensions) led to a strong reduction of the crater depth.64 While the “dense” Ar cluster acts a single spherical object, the clusters with a larger surface area do not penetrate the Si surface. They observed that the energy transfer to the surface varied relatively slightly over the series of projectiles, but the sputtering yield went from zero to a maximum for intermediate projected areas, then decreased again to zero. Finally, the molecular fragmentation induced in the polymer by CH4 clusters is compared to that of Ar clusters (inset of Figure 10). Despite the significant scatter in the data, the highenergy fragmentation (>1 eV/amu) is generally larger with CH4 clusters but, at the same time, the transition toward negligible fragmentation appears to be shifted toward higher scaled projectile energies. The latter observation might just be related to the fact that a comparatively larger part of the energy of the methane clusters is spent in other modes, as shown in Figure 10, and therefore the effective energy per mass unit is reduced with methane. In any case, the higher amount of fragmentation calculated at high energy invalidates the hypothesis of a softer emission induced by methane clusters (see Introduction). Taking into account that the H atoms of the CH4 molecules have the mass of tritium in the simulations (see the Computational Details), the methane clusters contain >8 times more atoms than the argon clusters for an equivalent total mass (1.66 times more C atoms than Ar atoms). In spite of a smaller mass/volume ratio, the number of atoms per unit of volume is also much larger for the methane clusters. The structure of the cluster is therefore one of finer and lighter “grains”, with a smaller spacing between them, which is more commensurate with the target structure than the Ar clusters, even with the CG representation of the sample. We speculate that the observed difference of fragmentation at high energy might be related to this difference of “granularity” of the projectiles. Instead of interacting with several units of the PE chain at the time and pushing them collectively like Ar atoms tend to do, C and 3H atoms interact more locally and might disrupt bonds more easily, with 1 eV/amu and beyond. The fragmentation of the sample and the methane molecules at high energy could produce protons useful for molecular ionization.40 Nevertheless, caution must be taken to not overinterpret the



CONCLUSION The new simulations of sputtering of model (macro)molecular samples by argon clusters with an incidence of 45° show all the features and even a quantitative agreement with the reported experimental data. The effect of the sample molecular weight on the sputter yield for molecules of more than 1000 amu is explained by the extra covalent binding and entanglement of the sample, while the angle-dependent variation of the yield is due the different depths at which the projectile energy is deposited and to the blocking effect of the projectile at normal incidence, as was proposed by other authors. In addition, our analysis suggests that the break in the scaled dependences of the sputter yields could be due to the observed (mis)match between the velocity of the projectile and the properties of the solid. But as this new body of results shows, one parameter alone cannot be singled out, and it is the combination of parameters (deposited energy, projectile velocity, geometry of the interaction for large clusters, and even projectile energy density) that explains the sputter yields of macromolecular samples under large cluster bombardment. This conclusion is in agreement with other recent computer simulation studies of sputtering involving different samples and/or different cluster projectiles. The bombardment simulations conducted with methane clusters bring additional information about the difference in dynamical behavior between atomic and molecular clusters. The latter spend a significant part of their energy in breaking their covalent bonds (at high energy per amu) but also in exciting the internal modes of vibration and rotation, an energy that is wasted for sputtering. These simulations suggest that sputtering by methane clusters is not softer than that initiated by argon clusters.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +3210473596. 25877

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C Notes

reference material using large argon cluster ions. Anal. Chem. 2010, 82, 98−105. (14) Shard, A. G.; Havelund, R.; Seah, M. P.; Spencer, S. J.; Gilmore, I. S.; Winograd, N.; Mao, D.; Miyayama, T.; Niehuis, E.; Rading, D.; Moellers, R. Argon cluster ion beams for organic depth profiling: results from a VAMAS interlaboratory study. Anal. Chem. 2012, 84, 7865−7873. (15) Niehuis, E.; Möllers, R.; Rading, D.; Cramer, H. G.; Kersting, R. Analysis of organic multilayers and 3D structures using Ar cluster ions. Surf. Interface Anal. 2013, 45, 158−162. (16) Wehbe, N.; Mouhib, T.; Delcorte, A.; Bertrand, P.; Moellers, R.; Niehuis, E.; Houssiau, L. Comparison of fullerene and large argon clusters for the molecular depth profiling of amino acid multilayers. Anal. Bioanal. Chem. 2014, 406, 201−211. (17) Bich, C.; Havelund, R.; Moellers, R.; Touboul, D.; Kollmer, F.; Niehuis, E.; Gilmore, I. S.; Brunelle, A. Argon cluster ion source evaluation on lipid standards and rat brain tissue samples. Anal. Chem. 2013, 85, 7745. (18) Fletcher, J. S. Latest applications of 3D ToF-SIMS bio-imaging. Biointerphases 2015, 10, 018902. (19) Ninomiya, S.; Ichiki, K.; Yamada, H.; Nakata, Y.; Seki, T.; Aoki, T.; Matsuo, J. Molecular depth profiling of multilayer structures of organic semiconductor materials by secondary ion mass spectrometry with large argon cluster ion beams. Rapid Commun. Mass Spectrom. 2009, 23, 3264−3268. (20) Mouhib, T.; Poleunis, C.; Wehbe, N.; Michels, J. J.; Galagan, Y.; Houssiau, L.; Bertrand, P.; Delcorte, A. Molecular depth profiling of organic photovoltaic heterojunction layers by ToF-SIMS: comparative evaluation of three sputtering beams. Analyst 2013, 138, 6801−6810. (21) Fleischmann, C.; Conard, T.; Havelund, R.; Franquet, A.; Poleunis, C.; Voroshazi, E.; Delcorte, A.; Vandervorst, W. Fundamental aspects of Arn+ SIMS profiling of common organic semiconductors. Surf. Interface Anal. 2014, 46, 54−57. (22) Gillen, G.; Roberson, S. Preliminary evaluation of an SF5+ polyatomic primary ion beam for analysis of organic thin films by secondary ion mass spectrometry. Rapid Commun. Mass Spectrom. 1998, 12, 1303. (23) Weibel, D.; Wong, S.; Lockyer, N.; Blenkinsopp, P.; Hill, R.; Vickerman, J. C. A C60 primary ion beam system for time of flight secondary ion mass spectrometry: its development and secondary ion yield characteristics. Anal. Chem. 2003, 75, 1754−1764. (24) Winograd, N. The magic of cluster SIMS. Anal. Chem. 2005, 77, 142A−149A. (25) Nieuwjaer, N.; Poleunis, C.; Delcorte, A.; Bertrand, P. Depth profiling of polymer samples using Ga+ and C60+ ion beams. Surf. Interface Anal. 2009, 41, 6. (26) Mouhib, T.; Delcorte, A.; Poleunis, C.; Bertrand, P. C60 molecular depth profiling of bilayered polymer films using ToF-SIMS. Surf. Interface Anal. 2011, 43, 175−178. (27) Mahoney, Ch. M., Ed. Cluster secondary ion mass spectrometry: principles and applications, 1st ed.; John Wiley & Sons, Inc.: New York, 2013. (28) Garrison, B. J.; Postawa, Z. Computational view of surface based organic mass spectrometry. Mass Spectrom. Rev. 2008, 27, 289−315. (29) Delcorte, A.; Restrepo, O. A.; Czerwinski, B. In Cluster secondary ion mass spectrometry, 1st ed.; Mahoney, Ch. M., Ed.; John Wiley & Sons, Inc.: New York, 2013. (30) Aoki, T. Molecular dynamics simulations of cluster impacts on solid targets: implantation, surface modification, and sputtering. J. Comput. Electron. 2014, 13, 108−121. (31) Anders, Ch.; Urbassek, H. M.; Johnson, R. E. Linearity and additivity in cluster-induced sputtering: A molecular-dynamics study of van der Waals bonded systems. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 155404. (32) Aoki, T.; Matsuo, J. Molecular dynamics simulations of surface modification and damage formation by gas cluster ion impacts. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 242, 517−519. (33) Rzeznik, L.; Czerwinski, B.; Garrison, B. J.; Winograd, N.; Postawa, Z. Microscopic insight into the sputtering of thin polystyrene

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Profs. B. J. Garrison and Z. Postawa for fruitful discussion about the scaling of the sputtering data. B. J. Garrison is gratefully acknowledged for the full access granted to her MD simulation code. A.D. is a Senior Research Associate of the Belgian Fonds National pour la Recherche Scientifique (FNRS). Computational resources were provided by the Institut de Calcul Intensif et de Stockage de Masse (CISM) at Université Catholique de Louvain. The theoretical and computational biophysics group of the University of Illinois at Urbana−Champaign is acknowledged for the development and free access to the VMD software (Humphrey, W.; Dalke, A.; Schulten, K. VMDVisual Molecular Dynamics. J. Mol. Graphics 1996, 14, 33−38).



REFERENCES

(1) Mahoney, J. F.; Perel, J.; Ruatta, S. A.; Martino, P. A.; Husain, S.; Lee, T. D. Massive cluster impact mass spectrometry: A new desorption method for the analysis of large biomolecules. Rapid Commun. Mass Spectrom. 1991, 5, 441−445. (2) Takats, Z.; Wiseman, J. M.; Cooks, R. G. Ambient mass spectrometry using desorption electrospray ionization (DESI): Instrumentation, mechanisms and applications in forensics, chemistry and biology. J. Mass Spectrom. 2005, 40, 1261−1275. (3) Chan, C. C.; Bolgar, M. S.; Miller, S. A.; Attygalle, A. B. Desorption ionization by charge exchange (DICE) for sample analysis under ambient conditions by mass spectrometry. J. Am. Soc. Mass Spectrom. 2010, 21, 1554−1560. (4) Hiraoka, K.; Mori, K.; Asakawa, D. Fundamental aspects of electrospray droplet impact/SIMS. J. Mass Spectrom. 2006, 41, 894− 902. (5) Tempez, A.; Schultz, J. A.; Della-Negra, S.; Depauw, J.; Jacquet, D.; Novikov, A.; Lebeyec, Y.; Pautrat, M.; Caroff, M.; Ugarov, M.; Bensaoula, H.; Gonin, M.; Fuhrer, K.; Woods, A. Orthogonal time-offlight secondary ion mass spectrometric analysis of peptides using large gold clusters as primary ions. Rapid Commun. Mass Spectrom. 2004, 18, 371−376. (6) Rickman, R. D.; Verkhoturov, S. V.; Hager, G. J.; Schweikert, E. A. Multi-ion emission from large and massive keV cluster impacts. Int. J. Mass Spectrom. 2005, 245, 48−52. (7) Yamada, I. Investigation of ionized cluster beam bombardment and its applications for materials modification. Radiat. Eff. Defects Solids 1992, 124, 69−80. (8) Yamada, I.; Matsuo, J.; Toyoda, N.; Kirkpatrick, A. Materials processing by gas cluster ion beams. Mater. Sci. Eng., R 2001, 34, 231− 295. (9) Ninomiya, S.; Nakata, Y.; Ichiki, K.; Seki, T.; Aoki, T.; Matsuo, J. Measurements of secondary ions emitted from organic compounds bombarded with large gas cluster ions. Nucl. Instrum. Methods Phys. Res., Sect. B 2007, 256, 493. (10) Oshima, S.; Kashihara, I.; Moritani, K.; Inui, N.; Mochiji, K. Soft-sputtering of insulin films in argon-cluster secondary ion mass spectrometry. Rapid Commun. Mass Spectrom. 2011, 25, 1070−1074. (11) Gnaser, H.; Ichiki, K.; Matsuo, J. Strongly reduced fragmentation and soft emission processes in sputtered ion formation from amino acid films under large Arn+ (n ≤ 2200) cluster ion bombardment. Rapid Commun. Mass Spectrom. 2012, 26, 1−8. (12) Ninomiya, S.; Ichiki, K.; Yamada, H.; Nakata, Y.; Seki, T.; Aoki, T.; Matsuo, J. Precise and fast secondary ion mass spectrometry depth profiling of polymer materials with large Ar cluster ion beams. Rapid Commun. Mass Spectrom. 2009, 23, 1601−1606. (13) Lee, J. L. S.; Ninomiya, S.; Matsuo, J.; Gilmore, I. S.; Seah, M. P.; Shard, A. G. Organic depth profiling of a nanostructured delta layer 25878

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879

Article

The Journal of Physical Chemistry C films on Ag{111} induced by large and slow Ar clusters. J. Phys. Chem. C 2008, 112, 521−531. (34) Rzeznik, L.; Czerwinski, B.; Garrison, B. J.; Winograd, N.; Postawa, Z. Molecular dynamics simulations of sputtering of organic overlayers by slow, large clusters. Appl. Surf. Sci. 2008, 255, 841−843. (35) Restrepo, O. A.; Delcorte, A. Argon cluster sputtering of a hybrid metal−organic surface: a microscopic view. J. Phys. Chem. C 2013, 117, 1189−1196. (36) Czerwinski, B.; Rzeznik, L.; Paruch, R.; Garrison, B. J.; Postawa, Z. Effect of impact angle and projectile size on sputtering efficiency of solid benzene investigated by molecular dynamics simulations. Nucl. Instrum. Methods Phys. Res., Sect. B 2011, 269, 1578−1581. (37) Rzeznik, L.; Postawa, Z. Molecular dynamics computer simulations of sputtering of benzene sample by large mixed Lennard-Jones clusters. Nucl. Instrum. Methods Phys. Res., Sect. B 2014, 326, 185−189. (38) Postawa, Z.; Kanski, M.; Maciazek, D.; Paruch, R. J.; Garrison, B. J. Computer simulations of sputtering and fragment formation during keV C60 bombardment of octane and β-carotene. Surf. Interface Anal. 2014, 46, 3−6. (39) Cristaudo, V.; Poleunis, C.; Czerwinski, B.; Delcorte, A. Ar cluster sputtering of polymers: effects of cluster size and molecular weights. Surf. Interface Anal. 2014, 46, 79−82. (40) Delcorte, A.; Cristaudo, V.; Lebec, V.; Czerwinski, B. Sputtering of polymers by keV clusters: microscopic views of the molecular dynamics. Int. J. Mass Spectrom. 2014, 370, 29−38. (41) Russo, M. F., Jr.; Garrison, B. J. Mesoscale energy deposition footprint model for kiloelectronvolt cluster bombardment of solids. Anal. Chem. 2006, 78, 7206−7210. (42) Anders, Ch.; Ziegenhain, G.; Zimmermann, Z.; Urbassek, H. M. Cluster-induced crater formation. Nucl. Instrum. Methods Phys. Res., Sect. B 2009, 267, 3122−3125. (43) Restrepo, O. A.; Prabhakaran, A.; Delcorte, A. Interaction of energetic clusters (Au3, Au400 and C60) with organic material and adsorbed gold nanoparticles. Nucl. Instrum. Methods Phys. Res., Sect. B 2011, 269, 1595−1599. (44) Delcorte, A.; Restrepo, O. A.; Czerwinski, B.; Garrison, B. J. Surface sputtering with nanoclusters: the relevant parameters. Surf. Interface Anal. 2013, 45, 9−13. (45) Delcorte, A.; Restrepo, O. A.; Hamraoui, K.; Czerwinski, B. Cluster impacts in organics: microscopic models and universal sputtering curves. Surf. Interface Anal. 2014, 46, 46−50. (46) Anders, Ch.; Urbassek, H. M. Effect of molecular dissociation energy on the sputtering of molecular targets. J. Phys. Chem. C 2010, 114, 5499−5505. (47) Moritani, K.; Kanai, M.; Goto, K.; Ihara, I.; Inui, M.; Mochiji, K. Secondary ion emission from insulin film bombarded with methane and noble gas cluster ion beams. Nucl. Instrum. Methods Phys. Res., Sect. B 2013, 315, 300−303. (48) Wucher, A.; Tian, H.; Winograd, N. A mixed cluster ion beam to enhance the ionization efficiency in molecular secondary ion mass spectrometry. Rapid Commun. Mass Spectrom. 2014, 28, 396−400. (49) Sheraz, S.; Barber, A.; Fletcher, J. S.; Lockyer, N. P.; Vickerman, J. C. Enhancing secondary ion yields in time-of-flight secondary ion mass spectrometry using water cluster primary beams. Anal. Chem. 2013, 85, 5654−5658. (50) Delcorte, A.; Garrison, B. J.; Hamraoui, K. Dynamics of molecular impacts on soft materials: from fullerenes to organic nanodrops. Anal. Chem. 2009, 81, 6676−6686. (51) Seah, M. P. Universal equation for argon gas cluster sputtering yields. J. Phys. Chem. C 2013, 117, 12622−12632. (52) Cumpson, P. J.; Portoles, J. F.; Barlow, A. J.; Sano, N. Accurate argon cluster-ion sputter yields: Measured yields and effect of the sputter threshold in practical depth-profiling by x-ray photoelectron spectroscopy and secondary ion mass spectrometry. J. Appl. Phys. 2013, 114, 124313. (53) Paruch, R. J.; Garrison, B. J.; Mlynek, M.; Postawa, Z. On universality in sputtering yields due to cluster bombardment. J. Phys. Chem. Lett. 2014, 5, 3227−3230.

(54) Paruch, R. J.; Postawa, Z.; Garrison, B. J. Seduction of finding universality in sputtering yields due to cluster bombardment of solids. Acc. Chem. Res. 2015, DOI: 10.1021/acs.accounts.5b00303. (55) Seah, M. P.; Spencer, S. J.; Shard, A. G. Angle dependence of argon gas cluster sputtering yields for organic materials. J. Phys. Chem. B 2015, 119, 3297−3303. (56) Smiley, E. J.; Postawa, Z.; Wojciechowski, I. A.; Winograd, N.; Garrison, B. J. Coarse-grained molecular dynamics studies of clusterbombarded benzene crystals. Appl. Surf. Sci. 2006, 252, 6436−6439. (57) Delcorte, A.; Garrison, B. J. Sputtering polymers with buckminsterfullerene projectiles: A coarse-grained molecular dynamics study. J. Phys. Chem. C 2007, 111, 15312−15324. (58) Hamraoui, K.; Delcorte, A. Effects of molecular orientation and size in sputtering of model organic crystals. J. Phys. Chem. C 2010, 114, 5458−5467. (59) Stuart, S. J.; Tutein, A. B.; Harrison, J. A. A reactive potential for hydrocarbons with intermolecular interactions. J. Chem. Phys. 2000, 112, 6472−6486. (60) Garrison, B. J.; Postawa, Z.; Ryan, K. E.; Vickerman, J. C.; Webb, R. P.; Winograd, N. Internal energy of molecules ejected due to energetic C60 bombardment. Anal. Chem. 2009, 81, 2260−2267. (61) Mücksch, Ch.; Anders, Ch.; Gnaser, H.; Urbassek, H. M. Dynamics of L-phenylalanine sputtering by argon cluster bombardment. J. Phys. Chem. C 2014, 118, 7962−7970. (62) Anders, C.; Bringa, E. M.; Fioretti, F. D.; Ziegenhain, G.; Urbassek, H. M. Crater formation caused by nanoparticle impact: a molecular dynamics study of crater volume and shape. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 235440. (63) Popok, V. N.; Vučković, S.; Samela, J.; Järvi, T. T.; Nordlund, K.; Campbell, E. E. B. Stopping of energetic cobalt clusters and formation of radiation damage in graphite. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 205419. (64) Aoki, T.; Seki, T.; Matsuo, J. Study of density effect of large gas cluster impact by molecular dynamics simulations. Nucl. Instrum. Methods Phys. Res., Sect. B 2009, 267, 2999−3001.

25879

DOI: 10.1021/acs.jpcc.5b07007 J. Phys. Chem. C 2015, 119, 25868−25879