Multicharged ion processing for targeted nanostructure formation

Metal silicide films are an integral part of modern microelectronics and find use in multiple contexts, such as Schottky diodes and local interconnects in metal-oxide–semiconductor devices.^1,2 At the nanometer scale, they have been synthesized using a variety of techniques, including silicidation, chemical vapor transport, and chemical vapor deposition. Metal silicides of nanometer dimensions naturally form geometric structures that have distinct shapes and sizes governed by both their nucleation conditions and by strain evolution during growth. One example of such structures is Co on Si which initially forms square shapes that transition to clusters with large aspect ratios. While this and similar systems grown in the low deposition limit exhibit transitional growth behaviors consistent with the model of Tersoff and Tromp, their nucleation (both location and density) and growth rate are dependent on intrinsic defect sites in the material system.^3,4 Attempts to control the growth have included the preferential placement of silicides using methods such as laser-induced periodic surface structures.⁵ 

The $Cu3$Si system, in particular, has motivated recent interest based on its possible use in energy storage applications, such as batteries and solar cells.^6,7 For example, Xu et al. investigated the use of $Cu3$Si in battery anodes and showed that an excellent electrochemical performance was obtained for a Si–$Cu3$Si–Cu composite electrode structure.⁸ Also, it has been shown recently that $Cu3$Si nanowires exhibit superior field-emission properties and can serve as highly efficient anti-reflective layers and as an effective catalyst in reactions.⁹ 

The motivation for the work we present here is an investigation of the role that void formation from multicharged ion impacts has on the formation and growth of $Cu3$Si nanostructures. Specifically, it is known that annealing is required to successfully form $Cu3$Si nanostructures on $SiO2$/Si.^10–12 As we showed in our recent work, SEM images taken under various annealing conditions reveal a linear, temperature-dependent variation in the density of voids or bare Si regions at the $SiO2$ surface.⁴ The observed dependence is linked to the importance of voids in $Cu3$Si nanostructure formation, particulary within an optimal temperature range of 450–600 $°$C. Multicharged ions are known to create distinct, fluence-dependent features at insulating surfaces, such as $Al2O3$ and $SiO2$⁠.^13–18 Therefore, it is reasonable to consider whether a controlled fluence of multicharged ions could supplement or even replace the annealing steps required for void formation in $Cu3$Si nanostructure growth.

In Sec. II, we present the methods and materials relevant to this work. The data obtained for multicharged ions inserted into the processing steps for $Cu3$Si nanostructure formation are shown in Sec. III. These results are discussed in the context of the energy and charge dissipation necessary to neutralize an incident multicharged ion in Sec. IV followed by our conclusions in Sec. V.

Single crystal, 3-inch diameter, boron doped p-type silicon wafers of (111) and (100) orientations with resistivities of 1–10 $Ω$ cm were employed as substrates for this work. Prior to the oxide growth, the wafers were cleaned following standard RCA clean, which involves cleaning of the substrates in a chemical bath (1:1:5 solution of $NH4$OH + $H2O2+H2$O) for five minutes under ultrasonic agitation. This is followed by etching of cleaned surface with 1% dilute HF for two minutes and triple rinse in de-ionized water for six minutes to remove any native oxide. Following this cleaning, an ultra-thin thermal oxide layer of approximately $1.5±0.05$ nm was grown via dry oxidation of Si at 200 $°$C for 2 h in a hot furnace. Oxide thickness measurements were performed at five different locations on each wafer (center and four corners) using a Sopra model GES-5E spectroscopic ellipsometer. The oxide grown Si wafers were diced into 10 $×$ 10 mm$2$ squares and mounted on Omicron style sample platen and loaded into the target chamber.

The samples were exposed to the focused multicharged ion beams produced in the CUEBIT laboratory which is discussed in detail in Ref. 19. The pressure inside the target chamber during the irradiations was on the order of $10−8$ mbar and the ion fluence was in the range of $1011$–$1012$ per $cm2$⁠. The ion species used for the irradiations were $Ar1+$⁠, $Ar4+$⁠, and $Ar8+$ which have potential or neutralization energies of 15.8, 143.9, and 577.7 eV, respectively. The kinetic energy of all the ion beams was 1 keV. Beam profiles were measured using a (HRBIS-4000 from Beam Imaging Solutions) as well as a translatable Faraday cup. An example beam profile, along with relevant information regarding profiles and fluence calculations, is shown in Ref. 20.

After irradiation, the samples were unloaded from the target chamber and transferred to a deposition chamber. Thin Cu films were deposited on these irradiated samples at temperature of 600 $°$C using a McAllister eBeam Evaporator at a pressure of $5×10−6$ Torr. The deposition was kept to a 0.1 mA flux current and a deposition time of 5 min. The deposition rate was estimated to be 1 monolayer/min (ML/min). The sequence of steps involved in the growth process of the resulting $Cu3$Si nanostructures is shown in Fig. 1. Following deposition, surface morphologies of the sample were investigated using a Hitachi S4800 scanning electron microscope. Orientation and chemical morphology of the nano-islands were established using energy dispersive X-ray (EDX) analysis. Figures 2(a) and 2(b) and Table I show the size distribution (length, width, area, density) of the nanostructures formed for $SiO2$/p-Si(100).

Here, we demonstrate the use of MCIs as a processing tool in the growth of epitaxial copper silicide on p-doped Si substrates. Multiple samples were exposed to beams of 1 keV $Arq+$ ions $(q=1,4,8)$ at the CUEBIT facility¹⁹ as shown schematically in Fig. 1. Subsequent copper deposition at $600°$C for 5 min at a deposition rate of $∼$1 ML/min led to the formation of a high density of nano-islands of $Cu3$Si. As is evident from the SEM image in Fig. 1 the high density growth corresponds to the region irradiated by the MCIs. Traditionally, efficient epitaxial nanostructure growth on Si/$SiO2$ substrates requires the formation of voids through annealing of the samples at high temperatures (600–1000 $°$C) for 10–12 hours.^4,21,22 In our case, this step was replaced by irradiation with MCIs and nanostructures were still formed.

Representative data obtained for a Si(100) substrate are shown in Figs. 2 and 3. From Fig. 2, one can see a variation in the shapes of the $Cu3$Si nanostructures that depends on the $Arq+$ $(1≤q≤8)$ charge state used and on the ion fluence. For low fluence values (⁠$1−2.5×1011cm−2$⁠) and $q≤4$⁠, the observed structures are squares reflecting the fourfold symmetry of the substrate. At a higher fluence (⁠$≥2.5×1012cm−2$⁠) and $q≤4$⁠, a change in the observed shape from square to nanowire is seen. No square-shaped nanostructures were observed for $Ar8+$ ions even at the lowest fluence, and the observed shapes were either nanowire or a combination of nanowire and rectangular islands. In the latter case, the rectangular islands are observed to be present on the top of nanowires and positioned approximately at the center. The orientations of the nanowires were found to be along [$1¯$10] and [110] directions without any preference.

Figure 3 shows the density of the $Cu3$Si nano-islands as a function of fluence with a representative SEM image as the inset. The mean density values were obtained by processing multiple SEM images that spanned the irradiated region of each sample. The density scales linearly indicating that each additional ion-surface interaction (increased fluence) is leading to the formation of additional $Cu3$Si nanostructures (increased density) on the $SiO2$/p-Si(100) substrate. However, this is also $q$-dependent, with the higher charge state (+8) leading to significantly more nanostructure formation. Similar results were seen on the Si(111) substrate. The fluence-dependent area of the $Cu3$Si nanostructures is shown in Fig. 4 for $Ar4+$ and $Ar8+$ irradiations and deposition times of 5 min. These data are compared with the time-dependent area of the $Cu3$Si nanostructures formed through the deposition and thermal annealing process (no irradiation) over a 1–5 min deposition time range. It is evident that the areas of the ion-assisted $Cu3$Si nanostructures are much larger compared to those formed via the thermally annealed process.

The results we have obtained for $Arq+$ irradiation of p-type Si indicate that the ions and their charge state can play a significant role in the formation of $Cu3$Si nanostructures. As we noted in Sec. I, voids in the $SiO2$ layer are required to promote the formation of the silicide islands. In the absence of a thermal annealing step, the ion beam irradiation appears to be a generator of these voids as evidenced by the clear delineation between irradiated and non-irradiated growth regions in the SEM image of Fig. 1. Moreover, the island density can be tied quantitatively to the specific ion irradiation fluence as shown in Fig. 3. For both $Ar4+$ and $Ar8+$ ions the linear relationship between fluence and feature density indicates that each impact generates a possible island nucleation site with some $q$-dependent probability. This is consistent with fluence-dependent features observed in multicharged ion measurements on other substrates such as $Al2O3$⁠.¹³ 

The slope of the lines from Fig. 3 can be used to calculate a feature density per ion which is plotted in Fig. 5 as a function of the ion charge state. Fitting this dependence to a power law for our p-Si(100) substrate, i.e., density $∝qγ$⁠, we find $γ=0.95±0.04$ or a nearly linear dependence on charge state. For the p-Si(111) substrate we find a slightly lower value of $γ=0.90±0.04$⁠. According to the Coulomb explosion model proposed by Tona et al.,²³ the relation between the sputter yield of positive ions $(Yp)$ and potential energy of a multicharged ion $(Ep)$ was determined to be $Yp∝EpΓ$⁠, where $Γ$ was found to be 0.6 for the sputter yields of $O+$ ions due to $Iq+$ ion impacts on native $SiO2$/Si(111) targets. For our $Arq+$ ion charge states, $q=1$⁠, 4 and 8 correspond to reneutralization or potential energies of 15.8, 143.9, and 577.7 eV, respectively. Assuming that our observed $Cu3$Si nanostructures densities are formed due to voids created by the multicharged ion irradiation at the surface, which in turn depend on the number of bonds broken (or equivalently, the sputter yield), we can assume that the densities of the nanostructures are directly related to the sputter yield for the $SiO2$ substrate. Hence treating the densities to be proportional to the underlying sputter yield, and following the same relation of $Yp∝EpΓ$ as in Tona et al., we find $Γ=0.58±0.18$ and $0.53±0.06$ for $Arq+$ on $SiO2$/Si(100) and $SiO2$/Si(111), respectively. This is remarkably similar to the Tona et al. result,²³ and the observed differences could be attributed to the differing oxides used in the two experiments (native vs thermally grown). Although we can see that our data are consistent with the Coulomb explosion model, it is worth noting that charge state enhanced preferential sputtering²⁴ and defect mediated sputtering²⁵ have been observed to play a role in other studies.

Regarding the morphology of the nanostructures formed, we observe that there is a transition from squares to wires for $q=4$ and $8$⁠. Such a transition is consistent with the Tersoff and Trump model³—an energetics based strain driven shape transition model. According to this model, the nanostructure formation minimizes the strain energy of an island per unit volume. The model predicts a critical size below which a compact shape (⁠$L=W$⁠) is observed, but when exceeded, rectangular nanoshapes (⁠$L>W$⁠) are observed. Figure 2(b) shows the critical size at which the shape transitions from a square to a wire for $SiO2$/p-Si(100).

The formation of $Cu3$Si nanostructures on p-type Si depends on void formation in the silicon substrate. These voids are traditionally created by thermal annealing at high temperatures (⁠$450$–$600°$C) for a period of $∼$12 h. In this work, we have shown that the annealing step can be replaced by controlled multicharged ion irradiation, as demonstrated by the comparison of the density of $Cu3$Si nanoislands for irradiated and non-irradiated samples. Specifically, silicon oxide samples were irradiated with multicharged $ArQ+$ ion beams with charge state Q (Q = 1,4,8) and corresponding neutralization (potential) energies $Ep$ (⁠$Ep=15.8$⁠, 143.9, and 577.7 eV). The irradiated samples were subsequently exposed to a Cu thin film deposition process. The surface and chemical morphology of the resulting $Cu3$Si nanostructures were measured using SEM and energy dispersive x-ray analysis, respectively, and a significantly higher density of nanoislands was seen in the samples that were irradiated with multicharged ions. Further, the calculated feature density per ion showed a near linear dependence on charge state (slope = 0.90$±$0.04). Assuming that nanostructure formation depends on void formation, and void formation in turn directly depends on sputtering, sputter yields may be considered to be directly dependent on the densities of nanostructure formation. Based on a Coulomb explosion model, a power law relating the sputter yields to the neutralization energy of multicharged ions was proposed by Tona et al.²³ By applying the proposed power law to the density of nanostructure formation, we obtain a sputter yield of $0.58±0.18$ and $0.53±0.06$ due to $ArQ+$ irradiation for oxides on Si(100) and Si(111) substrates, respectively, which is in close agreement to the sputter yield of 0.6 reported for $O+$ ions due to $IQ+$ irradiation on oxides grown on Si(111) substrates.

This research was funded by the Defense Advanced Research Projects Agency (Army Research Office Agreement No. W911NF-13-1-0042).

The authors have no conflicts to disclose.

E. S. Srinadhu: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). D. D. Kulkarni: Investigation (equal); Writing – review & editing (equal). D. A. Field: Investigation (equal). J. E. Harriss: Data curation (equal); Investigation (equal). C. E. Sosolik: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Project administration (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.